INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Pathogenic mycobacterial species have long been associated with serious human disease and mortality. For most of these pathogens, such as Mycobacterium tuberculosis, M. leprae, and M. marinum, the ability to survive and replicate within human macrophages is considered a major virulence determinant and one which contributes significantly to the ability of these organisms to persist for years in the host (4, 15, 17, 35). Infection with mycobacterial pathogens elicits a strong inflammatory response. Much of this response is felt to be due to the presence of the large amount of indigestible lipid associated with the lipid-rich mycobacterial cell wall. In contrast, M. ulcerans, the causative agent of Buruli ulcer, is unique among human mycobacterial pathogens in that the bacteria are primarily extracellular and there is a limited inflammatory response to the infection. Despite the large necrotic ulcer which results from infection with M. ulcerans, there is little evidence of an acute inflammatory response (12, 20).

In 1965, Connor and Lunn (13) suggested that the extensive necrosis of M. ulcerans infections was due to a diffusible substance, such as an exotoxin, causing the ulcers. In 1974, Read et al. reported the presence of a cytopathic activity in the culture supernatants of M. ulcerans (36). Several reports have characterized the effects of M. ulcerans culture supernatant on tissue culture cells (primarily mouse fibroblast L929 cells), in guinea pig skin, and mouse footpad models (24, 36). Sterile filtered M. ulcerans culture supernatant caused adherent L929 cells to round up and lift off the tissue culture plate (termed a cytopathic effect). Intradermal injection of the sterile filtrate (SF) caused small ulcer-like lesions in guinea pig skin (24, 36). This finding suggested that a toxin could be, in part, responsible for the pathogenesis of Buruli ulcer. The most recent paper on the M. ulcerans toxin ascribes immunosuppressive properties to the SF (33).

Several papers published in the 1970s described attempts at defining the biochemical properties of this toxin (21, 24, 36). Investigations by Krieg et al. describe the toxin as a heat-stable substance present in the SF, cytoplasmic fraction, and particulate fraction but absent from the cell wall (24). Further work from this same laboratory concluded that the toxin was composed of "a high-molecular-weight phospholipoprotein-polysaccharide complex" (21).

In this report, we describe partial purification of a toxin from M. ulcerans and characterize the M. ulcerans toxin as a relatively apolar lipid. We have also further defined the biological activities of the SF and partially purified toxin by quantitating cytopathic activity in mouse fibroblast L929 cells. The lack of an acute inflammatory response to infection with M. ulcerans, the cytopathology of the toxin on tissue culture cells, and the lipid nature of the toxin suggest that the M. ulcerans toxin may function in a unique, as yet uncharacterized mechanism.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Eukaryotic cell culture. L929 mouse fibroblast cells (ATCC CCL1) were purchased from the American Type Culture Collection and passaged in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% heat-inactivated fetal calf serum (FCS; Gibco BRL, Grand Island, N.Y.).

Bacterial culture. M. ulcerans 1615 (ATCC 35840) was obtained from the American Type Culture Collection and is part of the Trudeau Collection. It was originally isolated in Malaysia from a human patient (32). M. ulcerans and M. marinum 1218 (ATCC 927) were passaged (1:20 dilutions) in Middlebrook 7H9 medium (Difco, Detroit, Mich.) supplemented with 10% Middlebrook oleic acid, bovine serum albumin (BSA), dextrose, and catalase enrichment (OADC supplement) at 32°C without shaking in Falcon T-185 flasks. Culture filtrates from M. ulcerans and M. marinum were harvested from cultures in late exponential growth phase since prior work established that toxin production is optimal at this time (24). Virulence was confirmed by using a guinea pig model of infection as described previously (24). This medium is referred to as M7H9.

Cytopathic assay. L929 cells were plated at 7 × 10⁴/well in 24-well plates or 5 × 10³/well in 96-well tissue culture plates and allowed to adhere overnight. Mycobacterial cultures were passed through 0.22-µm-pore-size sterile filters, and the SF was added to the L929 cells as follows: SF was diluted 1:2 with sterile phosphate-buffered saline (PBS), and serial dilutions were added to L929 cells at 1/20 total volume. Mycobacterial medium was also added as controls. At 24 and 48 h posttreatment, the L929 cells were inspected microscopically for rounded up and detached cells. In each experiment, the last dilution which had greater than 90% rounded up and detached cells was arbitrarily called 1 cytopathic activity unit (CPU). CPU were calculated as follows: 1 U/µl of SF added at the last positive dilution × total number of microliters of the assay.

Protease digestion of SF. M. ulcerans SF or M7H9 medium was incubated with chymotrypsin (10-fold excess by weight) for 1 h at 37°C. If indicated, Bowman-Birk soybean inhibitor was added at twice the concentration of chymotrypsin. For proteinase K treatment, M. ulcerans SF or M7H9 medium was treated with proteinase K (200 µg/ml) for 1 h at 37°C, and, if indicated, 4 mM PefaBloc SC (Boehringer Mannheim, Indianapolis, Ind.) was added. Cytopathic assays were performed as described above. Chymotrypsin and Bowman-Birk inhibitor were purchased from Sigma (St. Louis, Mo.); proteinase K and PefaBloc were purchased from Boehringer Mannheim.

Partial purification of toxin. Aliquots of 1,000 ml of SF from log-phase cultures of M. ulcerans or M. marinum were concentrated and extracted with chloroform-methanol (2:1, vol/vol) for 2 h at room temperature (19). After separation of the aqueous phase (which contains proteins, salts, and other highly polar molecules) and organic phase (contains most lipids), the organic phase was dried and the remaining material was precipitated with ice-cold acetone to separate the nonsoluble phospholipids from less polar lipids (6). For cytopathic assays, aliquots of all fractions were resuspended in sterile PBS or M7H9 and added to adherent L929 cells as described above. Radiolabeled, [1-¹⁴C]acetic acid sodium salt (specific activity, 59.0 mCi/mmol; Amersham) was added to mycobacterial cultures (0.5 µCi/ml of culture) for 6 days. For thin-layer chromatography (TLC) analysis, lipids were resuspended in chloroform-methanol (2:1, vol/vol) and loaded onto glass-backed Whatman K6 silica gel TLC plates (Alltech Associates, Inc., Deerfield, Ill.). The plates were developed twice with chloroform-methanol (96:4, vol/vol). Radiolabeled lipids were detected on a Molecular Dynamics PhosphorImager. Individual lipid spots were scraped off the glass backing and eluted with methanol. The eluted lipids were dried down and resuspended in PBS for cytopathic assays and DNA synthesis assays.

Actin staining. L929 cells were plated on glass coverslips in six-well tissue culture plates at a density of 1.5 × 10⁶ cells/coverslip. After overnight growth, the semiconfluent monolayer was exposed to toxin or M7H9 medium (as a control) and stained after 3, 4, or 10 h. Toxin-treated cells and control cells were fixed for 20 min with 4% paraformaldehyde and solubilized for 4 min in 0.1% saponin; F-actin was stained with rhodamine-phalloidin (10 U/ml; Molecular Probes, Eugene, Oreg.). Stained cells were visualized on a Bio-Rad MRC 1000 laser confocal microscope.

Flow cytometry. L929 cells were plated at 4 × 10⁴ cells/well in 3 ml of medium in six-well tissue culture plates. The next day, 150 µl of M7H9 medium, SF from M. marinum or M. ulcerans, or the acetone-soluble lipid fraction was added to the L929 cells. After 48 h, the supernatant and adherent cells were harvested, centrifuged, and resuspended in 1 ml of PBS-0.1% FCS. Cells were counted, centrifuged, and resuspended in fluorescence-activated cell sorting buffer (PBS, 0.1% Nonidet P-40, 20 µg of RNase A [Boehringer Mannheim] per ml, 50 µg of propidium iodide [Molecular Probes] per ml) at a concentration of 10⁶ cells/ml. Cells were immediately sorted and analyzed on a FACStar instrument modified for five-parameter operation (Becton Dickinson Immunocytometry Systems, San Jose, Calif.).

DNA and protein synthesis measurement. DNA synthesis measurements were performed as described previously (25). Briefly, 1.5 µCi of [methyl-³H]thymidine (specific activity, 25 Ci/mmol; Amersham) was added to L929 cells in 96-well plates for 1 h at 37°C. After 1 h, supernatants were removed, and cells were washed with PBS and trypsinized. All supernatants, washes, and trypsinized cells were pooled and harvested on an Inotech cell harvester. The counts/well was determined by scintillation counting of harvested cells. All samples were done in quadruplicate, and statistics were done with StatView program. For protein synthesis measurements, L929 supernatant was removed and transferred to a duplicate 96-well plate. The L929 cells were washed twice with methionine-free medium, and these washes were added to the respective supernatants. The supernatants and washes were then centrifuged to collect nonadherent cells, the medium was removed, and the cells that were present in the supernatant and washes were resuspended in methionine-free medium and added to the original plates. Fourteen microcuries of [³⁵S]methionine (specific activity 1,000 Ci/mmol; Amersham) in methionine-free medium plus the sample treatment was added to L929 cells for 1 h at 37°C. Cells were harvested, and counts/well was determined as described above.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Cytopathic activity of M. ulcerans toxin is protease resistant and heat stable. Previous literature reported that the toxin was a phospholipoprotein-polysaccharide complex (21). To further ascertain the importance of a protein component, we examined whether exposure of the SF to chymotrypsin and proteinase K resulted in a decrease in cytopathic activity. The presence of chymotrypsin or proteinase K in the M7H9 medium control alone caused L929 cells to round up (Table 1, samples 1 and 4). However, this effect occurred within 1 h, whereas the cytopathic effect produced by exposure to M. ulcerans SF required 12 to 24 h of exposure. When specific protease inhibitors were added to the reaction, the nonspecific early cytotoxic effect was completely inhibited (Table 1, samples 3 and 6). When M. ulcerans SF was treated with these proteases, and the proteases were subsequently inactivated by specific inhibitors, there was no decrease of CPU of the SF (Table 1, samples 7 to 12). The protease resistance of the SF suggested that the toxin may not have an active protein component. Since most proteins are heat sensitive, we tested the heat stability of the SF. As reported previously (21, 36), exposure of twofold dilutions of SF to 100°C for 30 min resulted in the retention of greater than 85% cytopathic activity (data not shown). These data, taken together, suggest that the active component might not be protein.

M. ulcerans toxic activity copurified with lipid extracts from SF. To begin toxin purification, M. ulcerans SF and M. marinum SF were subjected to a standard Folch extraction to separate lipids from other molecules (see Materials and Methods). After separation of the aqueous and organic phases, an aliquot of the organic phase was precipitated with ice-cold acetone to separate the phospholipids from other less polar lipids. All fractions were dried down, resuspended in equal volumes of PBS, and tested for CPU on L929 cells. As shown in Table 2, the cytopathic activity was retained in the acetone-soluble fraction of the organic phase. Separation of the non-acetone-soluble lipids from acetone-soluble lipids did not diminish the total CPU, suggesting that the active component is not a phospholipid. The observation that this toxic component is protease resistant and partitions into the organic phase is consistent with its being a lipid. As a control, all of these fractionation steps were performed with culture supernatant from a pathogenic strain of M. marinum (1218R). No fractions were found to be cytopathic to L929 cells, supporting the conclusion that M. ulcerans releases a species-specific lipid toxin molecule(s).

View larger version (34K): [in this window] [in a new window]   FIG. 1.   TLC of acetone-soluble lipids from M. ulcerans. M. ulcerans culture was radiolabeled with [1-14C]acetate, and acetone-soluble lipids were prepared. The lipids were resolved on a glass-backed TLC plate in chloroform-methanol (2:1, vol/vol) and detected on a PhosphorImager. Individual lipids were scraped off the glass backing, eluted with methanol, and tested for cytopathic activity. Arrows indicate lipids which were scraped off the TLC plate, and the results of the cytopathic assay are noted with plus and minus signs.

M. ulcerans effect on cytoskeleton of L929 cells. To examine the cellular events involved in the cytopathic activity on L929 cells, we investigated the kinetics of toxin activity on the cytoskeleton by staining treated cells with the actin-specific stain rhodamine-labeled phalloidin. At 4 h, over 50% of the toxin-treated cells were rounded compared to the M7H9-treated cells (Fig. 2A and B). By 10 h, stress fibers were absent from the majority of cells and localized foci of F-actin could be found on the cell periphery (Fig. 2D). Although obvious changes in the cytoskeleton occur within 4 h, cells do not become detached from the monolayer until 24 to 36 h. Thus, the cytopathic effect is accompanied by cytoskeletal rearrangement which can be seen as early as 4 h following exposure to the toxin.

View larger version (87K): [in this window] [in a new window]   FIG. 2.   Effect of toxin on cytoskeleton of L929 cells. L929 cells were exposed to either M7H9 medium or M. ulcerans SF and stained with the actin-specific stain phalloidin-red. (A) L929 cells exposed to M7H9 medium for 4 h show normal actin structure with stress fibers. (B) L929 cells exposed to M. ulcerans SF for 4 h show some rearrangement of actin. (C) M7H9-treated L929 cells at 10 h. (D) SF-treated L929 cells at 10 h show dramatic cytoskeleton rearrangements.

M. ulcerans toxin causes cell cycle arrest in L929 cells. The effect of the toxin on L929 cells is not cytocidal but reversible, even after 48 h of exposure to the SF (41). These observations suggested to us that the toxin from M. ulcerans may place the L929 cells in cell cycle arrest. We used flow cytometric analysis to determine which stage of the cell cycle the toxin-treated cells were in after 48 h of exposure. As shown in Fig. 3, after 48 h of exposure, M7H9 medium- or M. marinum SF-treated L929 cells exhibited a typical profile of cycling cells, with 61 or 57%, respectively, in G₀/G₁ and 37 or 41%, respectively, in S, G₂, or M (Fig. 3A and B). After 48 h of exposure to M. ulcerans SF or partially purified toxin, there was a significant shift in the cell population, with 88 or 83%, respectively, in G₀/G₁ and 8 or 7%, respectively, in S, G₂, or M. These data indicate that the toxin causes the L929 cells to arrest in the G₀/G₁ phase of the cell cycle. The fact that the partially purified lipid fraction causes both the cytopathic effect and the cell cycle arrest strongly suggests that both phenotypes may be caused by one or more lipids present in the acetone-soluble fraction.

View larger version (23K): [in this window] [in a new window]   FIG. 3.   Flow cytometry analysis of toxin-treated L929 cells. L929 cells were treated with M7H9 medium (A), M. marinum SF (B), M. ulcerans SF (C), and M. ulcerans acetone-soluble lipid fraction (D) for 48 h. Floating and attached cells were harvested, lysed, and stained with propidium iodide. After 48 h of treatment, there was a significant shift in the percent of M. ulcerans-treated cells (C and D) in the G0/G1-phase of the cell cycle compared to the medium- and M. marinum-treated (A and B) controls.

View larger version (20K): [in this window] [in a new window]   FIG. 4.   Kinetics of DNA and protein synthesis. (A) DNA synthesis was measured at various times after treatment with M7H9 medium (squares), M. marinum SF (diamonds), M. ulcerans SF (circles), or DMEM alone (triangles). (B) Protein synthesis was measured at various hours after treatment with M7H9 medium (squares), M. marinum SF (diamonds), M. ulcerans s SF (circles), or DMEM alone (triangles). All time points and samples were done in quadruplicate, and standard deviations are indicated with bars.

Exposure of fibroblasts to M. ulcerans SF alters the ability of cells to progress through G₁. Since it appears that treatment of L929 cells with M. ulcerans toxin results in a block in G₀/G₁, we next examined whether the toxin could inhibit G₁ progression in quiescent L929 cells stimulated by FCS. To determine the kinetics of entry into DNA synthesis, L929 cells which had been incubated with serum-free DMEM for 5 days were stimulated to reenter the cell cycle by addition of medium with 10% FCS (9). As shown in Fig. 5, the synchronized L929 cells reach the peak of DNA synthesis at 18 to 20 h after addition of serum. If M. ulcerans SF was added at the same time as serum, DNA synthesis was greatly reduced, indicating that the toxin prevents the cell cycle rescue of the L929 cells by fresh serum.

View larger version (25K): [in this window] [in a new window]   FIG. 5.   Ability of M. ulcerans SF to inhibit G1 progression in quiescent L929 cells stimulated by FCS. L929 cells were synchronized in G0 by incubation in serum-free DMEM for 5 days. After 5 days, the cells were stimulated by addition of DMEM-10% FCS (squares), and either M7H9 medium (diamonds) or M. ulcerans SF (circles) was added. DNA synthesis was measured at various times. All time points and samples were done in quadruplicate, and standard deviations are indicated with bars.

View larger version (77K): [in this window] [in a new window]   FIG. 6.   Determination of restriction point of toxin's action in growth-arrested L929 cells. Growth-arrested L929 cells were stimulated with DMEM-10% FCS, and at various times after addition, M7H9 medium (hatched bars) or M. ulcerans SF (solid bars) was added. DNA synthesis was measured at 20 h poststimulation. All time points and samples were done in quadruplicate, and standard deviations are indicated with bars.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Early work on M. ulcerans toxin using crude cell extracts or sterile culture filtrate suggested that the M. ulcerans toxin had a lipid component. In this work, biochemical characterization of the toxin has been extended considerably by showing that toxic activity copurifies exclusively with a specific lipid fractionacetone-soluble lipidsobtained from M. ulcerans SF. When lipid species from this fraction are separated by TLC, toxic activity is found in 2 of the 11 spots. This is the first time toxic activity has been obtained with a well-defined lipid species. We are in the process of determining the chemical structure of these lipids by preparative TLC, followed by mass spectrometry, infrared spectrometry, and ¹H and ¹³C nuclear magnetic resonance spectroscopy.

We have also further defined the biological activities of the toxin on mouse fibroblast L929 cells. Exposure of cells to either M. ulcerans SF or acetone-soluble lipids results in major changes in the distribution of F-actin in L929 cells. Furthermore, both the partially purified acetone-soluble lipid fraction and the M. ulcerans toxin place L929 cells in the G₁ phase of the cell cycle. None of these biological changes were observed with any fractions from M. marinum culture filtrate, despite the fact that M. marinum is taxonomically very similar to M. ulcerans and has a number of identical cell wall components, including phenolic glycolipids (14, 42).

We find, in agreement with Read et al. (36), that the cytopathic activity of the M. ulcerans SF is heat stable. Although we repeated their results concerning high molecular weight (greater than 100,000) by using Amicon filters, we found that toxic activity bound to the membrane made this method unreliable (data not shown). The fact that the toxin may be bound to BSA in the medium further complicates this method of size determination. Using ammonium sulfate precipitation, density gradient ultracentrifugation, and proteolytic and hydrolytic enzymes, Hockmeyer et al. (21) reported that the toxin was trypsin resistant, pronase sensitive, and lipase sensitive. They also reported that the toxin precipitated over a wide range of ammonium sulfate concentrations and was enriched for in a high-density lipoprotein fractionation which separates lipoproteins from unlipidated proteins (21). We have repeated some of these experiments, and our data are generally consistent with their findings. However, we find no evidence of an active protein component. We suggest that the pronase may have been contaminated with lipase. Furthermore, a lipid would be present in the lipoprotein fraction and in the ammonium sulfate precipitations, particularly if it were bound to BSA.

It has been reported that M. ulcerans can be divided into several subgroups based on DNA sequence variation downstream from the 16S rRNA gene (34). We have obtained representative strains from Africa and Australia used in this study from the Victoria Reference Collection (Australia) and are in the process of analyzing these strains with regard to toxin production and animal virulence. Evidence from human infections suggests that all isolates are from lesions with the same characteristic features of Buruli ulcer.

A growing body of literature has shown that progression through the cell cycle involves a number of pathways in which both signal transduction and the actin cytoskeleton are interconnected. The small GTP-binding proteins Rho, Rac, and Cdc4 are all known to be play essential regulatory roles in the actin cytoskeleton and are required for progression through the cell cycle (3, 26). The C3 exoenzyme from Clostridium botulinum and toxins A and B from C. difficile all inactivate Rho to induce a breakdown of the F-actin cytoskeleton and cause cell cycle arrest (2, 11, 16, 22, 23, 28). Escherichia coli cytotoxic necrotizing factor 1 toxin has been shown to activate Rho and induce cell cycle arrest (18, 38). Data presented in this paper show that the M. ulcerans toxin results in changes in the actin cytoskeleton as well as in a failure of cells to progress past G₁. This finding may suggest that one of the Rho family of small GTPases may be targets for this toxin; however, it is also interesting in that the M. ulcerans toxin is a lipid, and all of these other bacterial toxins are proteins. We are in the process of purifying the toxin so that such questions can be addressed.

A number of mycobacterial lipids have been postulated to play a role in pathogenesis and have multiple effects on host cells, particularly immune cells. The biological interactions of mycobacterial lipids such as lipoarabinomannan (LAM), trehalose dimycolate (TDM), sulfatide (SL), and phenolic glycolipids (PGLs) on host cells has been studied extensively. LAM has been shown to induce cytokine production in macrophages, such as tumor necrosis factor alpha (TNF-), interleukin-1 (IL-1), and IL-6 in mouse macrophage cell lines (1), and also inhibit gamma interferon (IFN-)-mediated activation of macrophages as well as scavenge oxygen radicals (10, 39, 40). TDM induces macrophage procoagulant activity, TNF-, and IL-12 and is hypothesized to be necessary for induction of granulomas in pulmonary tuberculosis (5, 27, 31). SL has been shown to inhibit priming of monocytes by lipopolysaccharide, IFN-, IL-1, and TNF- and induce secretion of TNF- and IL-1 in lipopolysaccharide-primed monocytes (8, 29). PGLs have been postulated to play a role in the virulence of M. leprae (7, 37).

We think it unlikely that the M. ulcerans toxin will be any of these previously described lipids for the following reason. LAM is not partitioned into the acetone-soluble fraction. Our data from TLC of the acetone-soluble lipids suggest that the active compound is much less polar than TDM or sulfatides would be expected to be. TDM must also be administered in an oil suspension to interact with host cells, and the activity of acetone soluble lipids is tested in PBS. In addition, acetone-soluble lipids from M. marinum which we have used as a negative control for all of these studies also contain TDM. Finally, although acetone-soluble lipids may contain phenolic glycolipids, data from other laboratories have shown that the PGLs from M. marinum and M. ulcerans are identical (14). Taken together, these data suggest that the M. ulcerans toxin may be a unique mycobacterial lipid with some very interesting biological properties. Additional purification and structural characterization of the toxin will permit studies to address the molecular mechanism of the toxin-induced cell cycle arrest and the possible role of the toxin in the pathogenesis of M. ulcerans.