Macrophage Responses to Bacterial Toxins: a balance between activation and suppression

Abstract

Toxins secreted by bacteria can impact the host in a number of different ways. In some infections, toxins play a crucial and central role in pathogenesis (i.e. anthrax), while in other bacterial infections, the role of toxins is less understood. The cholesterol-dependent cytolysins (CDCs), of which streptolysin O is a prototype, are a class of pore-forming toxins produced by many Gram positive bacteria and have only been studied in a few experimental infection models. Our laboratory has demonstrated that CDCs have effects on macrophages that are both pro-and anti-inflammatory. Here we review evidence that CDCs promote inflammation by driving the secretion of IL-1β and HMGB-1 from macrophages in a NLRP3 dependent manner, while also causing shedding of membrane microvesicles from cells that can interact with macrophages and inhibit TNF-α release. CDCs thus impact macrophage function in ways that may be both beneficial and detrimental to the host.

Introduction to macrophage biology

Macrophages are one of the first-line responders of the immune system to pathologic conditions. In order to appropriately react to and deal with these problems, macrophages possess a large repertoire of germ-line encoded receptors whose activation leads to cytotoxicity, cytokine production, and recruitment of other immune cells to propagate the immune response. Inappropriate and chronic activation of macrophages can also lead to a variety of diseases, including atherosclerosis, autoimmune diseases, and contact hypersensitivity. The best-studied family of these receptors is the Toll-like Receptors (TLRs), which recognize Pathogen Associate Molecular Patterns (PAMPs). Different TLRs possess unique specificities for certain PAMPs. For example, TLR4 recognizes bacterial lipopolysaccharide (LPS). Engagement of TLRs “primes” macrophages, initiating a gene program that includes the up-regulation of protective genes, induction of cytotoxicity, and cytokine synthesis (1).

A distinct set of protective genes upregulated by LPS-primed macrophages is the NOD-like receptor (NLR) family. Some members of this family of receptors respond to specific insults through the activation of caspase-1, which allows for the activation and release of many proteins, including IL-1β (2). Although the function of many NLRs remains unknown, one member, NLRP3, responds to a variety of stimuli, including ATP, crystals, potassium ionophores and bacterial pore-forming toxins (PFTs) (3). Following activation, NLRP3 associates with apoptosis-associated speck-like protein containing a CARD domain (ASC, also called Pycard) to activate caspase-1, forming a complex known as the inflammasome (4). How NLRP3 is activated, however, and by such a wide range of stimuli, remains unknown. Although potassium efflux and the generation of reactive oxygen species have been proposed to be potential inflammasome activating signals, it is not clear how they stimulate NLRP3-driven responses or if they directly participate in the process.

Cholesterol-Dependent Cytolysins

Pore forming toxins (PFTs) are the largest family of toxins, produced by over 20 members of gram positive bacteria, including Clostridium, Streptococcus, Listeria, and Bacillus sp. (5, 6). One class of PFTs, cholesterol dependent cytolysins (CDCs), require cholesterol for binding and are produced as soluble monomers that oligomerize on eukaryotic membranes to form pores up to 50 nm in diameter (7). Up to 50 monomers can form a pore and oligomerization, which is dependent on toxin concentration, is the rate determining step (7, 8). CDCs bind to cholesterol on cell membranes through three hydrophobic loops present in domain four of the molecule (9). Initially, a pre-pore complex is formed on the membrane, which transforms into a fully formed pore when a conserved, tryptophan rich undecapeptide in domain 4 inserts into the membrane followed by the two alpha helices of domain 3 becoming two extended β hairpins (9). Most CDCs are produced by extracellular bacteria. However, intracellular bacteria such as Listeria release CDCs inside the phagosome.

The effect of pore formation on the cell varies depending on the toxin dose, the duration of the exposure, and the cell type. The dose of toxin that cells are exposed to during infection in vivo cannot be readily determined, but is thought to vary based on the distance between the target cell and the site of infection. It is argued that the function of the pore complex may be to allow entry of bacterial virulence factors into the host cell. This process has been termed cytolysin-mediated translocation (10, 11). Cholesterol-dependent cytolysin pores have also been used as a research tool to transfer proteins into cells (12).

Pore-forming toxins stimulate pro-inflammatory responses from myeloid cells

Previous studies from our group demonstrated that the cholesterol-dependent cytolysin (CDC) listeriolysin O (LLO) expressed in E. coli induces secretion of pro-inflammatory cytokines IL-1β, IL-6, and IL-12 from myeloid cells (13). We also observed that purified toxin proteins from this family, including anthrolysin O, tetanolysin O, and streptolysin O, can each directly stimulate release of IL-1β from murine bone-marrow derived macrophages (BMDM). This led us to explore the mechanism of toxin-induced IL-1β release. While we found no evidence for NFκB activation, which is needed for IL-1β gene transcription and accumulation of pro-IL-1β protein, these toxins did cause release of mature IL-1β from LPS primed macrophages in a NLRP3 dependent manner (14). This result will be discussed in more detail in the following sections.

The CDC tetanolysin O induces pore formation and membrane damage in macrophages

Tetanolysin O (TLO) treated BMDM show evidence of pore formation via the uptake of membrane impermeable dyes. This uptake was cholesterol dependent, as the addition of free cholesterol to the media was able to inhibit dye uptake. Cells treated with lower doses of TLO were able to recover membrane integrity over time and stain positively for calcein, a live cell indicator, and exclude ethidium homodimer, a membrane impermeable dye. In addition to dye uptake assays indicating toxin dependent pore formation in BMDM, TLO caused release of cellular contents. Lactate dehydrogenase (LDH) is a ~140 kDa enzyme present in most cells and is normally retained within the cytosol. When LPS primed BMDM are treated with TLO, LDH is released from the cells. Furthermore, HMGB1, a nuclear non-histone protein, was seen in the supernatants of cells exposed to TLO. Thus, the release of both LDH and HMGB1 from TLO treated cells further confirmed membrane damage in BMDM as a result of TLO exposure.

NLRP3 dependent release of IL-1β from macrophages

Through the use of BMDM from NLRP3 and caspase-1 knock-out mice, we have shown that TLO-induced secretion of mature IL-1β is NLRP3 inflammasome dependent, and is independent of another inflammasome formed by NLRC4 (14). This processing and secretion of IL-1β requires caspase-1 activity, as processing was impaired in caspase-1 knock-out BMDM as well as with pharmacological inhibitors of caspase-1. The addition of potassium chloride to the extracellular media also inhibited the processing and secretion of mature IL-1β, as did an inhibitor of iPLA2 (14). The requirement for potassium efflux was not surprising, considering it was known both that it was necessary for IL-1 processing and secretion (15) and that it occurs following other stimuli of the NLRP3 (16). Furthermore, only the lower doses of TLO were able to induce the processing and secretion of IL-1β. Lytic doses resulted in the release of pro-IL-1β from the cells, indicating the loss of cell integrity. This release of unprocessed cytokine was seen in both the wild type and NLRP3 knock-out BMDM, indicating that this release was NLRP3 inflammasome independent. LDH and HMGB1 release were also found to be NLRP3 inflammasome dependent only at lower doses of TLO.

Our laboratory has also developed a transfection system to stably express ASC in D2-SC1 cells, which lack endogenous ASC, but express NLRP3, as well as caspase-1. Through the use of this system, we were able to reconstitute the NLRP3 inflammasome. As previously described in cells transfected with ASC (17), we were able to visualize punctate ASC aggregates in cells treated with TLO, or the potassium ionophore nigericin, indicating inflammasome formation (Fig. 1a). D2-ASC cells were able to process and secrete IL-1β in response to nigericin, while native D2 cells not expressing ASC were unable to process and secrete the cytokine (Fig. 1b). This result confirmed that ASC was required for nigericin induced processing and secretion of IL-1β in addition to NLRP3 and caspase-1, and is in agreement with previous work demonstrating ASC dependent processing of IL-1β (18).

Fig 1.

ASC-dependent NLRP3 inflammasome activation in D2-SC1. (A) ASC aggregated in response to 4 h of LPS priming followed by treatment of D2-ASC with a sublytic dose TLO for 30 min at 37°C. Nuclei were stained with DAPI and ASC was stained with anti-ASC antibody clone 2EI-7 (Millipore). (B) IL-1β secretion was detected via ELISA in response to 20 μM nigericin treatment for 30 min following 4 h of LPS priming in D2-ASC, but not D2 neomycin control cells. 50 mM potassium chloride (KCl) was used to inhibit NLRP3 inflammasome dependent IL-1β secretion.

Toxin Induces Release of Microvesicles from cells

In addition to the immunological responses described above from BMDM, all cells must physically/structurally react to the threat PFTs pose. The mechanism through which they do so remains controversial, with competing proposals suggesting inactivation of the PFT through blockade and/or disassembly, internalization of the PFT or shedding of the toxin from the cell (reviewed in (19)). Research on one endogenous PFT, the membrane attack complex of Complement, has suggested both endocytosis and shedding (20–22). Other PFTs induce shedding of entire intestinal microvilli (23, 24). Research on bacterial PFTs, including α-toxin and Streptolysin O, has been interpreted primarily to support a model of endocytosis for protection from toxin (25–27).

However, this model presents a number of theoretical problems. Internalization of toxin-laden membranes would be expected to permeabilize the endo/lysosomal system, spilling acid hydrolases into the cytoplasm. The danger of intracellular pore formation is illustrated by the lethal activity of LLO, which does not even form pores until in an acidic environment (28). Thus, endocytosis of bacterial products is generally associated with increased damage, not protection, for cells.

Recent work in our laboratory has shown using Streptolysin O that cells do not clear bacterial PFTs through endocytosis (29). Instead, cells sequester the PFT on blebs, block their cytosolic contact, and shed these blebs into the media through the process termed ectocytosis (29). Since membrane is donated to the plasma membrane from the endo/lysosomal system during repair, unshed blebs may carry endosomal markers and inadvertently be mistaken for endosomes. Similar shedding has been shown with the membrane attack complex (20), suggesting this may be a general mechanism of toxin resistance.

Macrophage Reaction to Toxin-induced Microvesicles

If cells release toxin-rich microvesicles following PFT attack, responding macrophages would be among the first immune cells to encounter and deal with them. These ectosomes range in size from ~150 nm up to 2 μm, and the concentration of SLO pores can be observed in them (Fig. 2a). We investigated first whether macrophages would interact with and/or internalize these ectosomes. We first labeled ectosomes derived from toxin-treated cells with Cy3 or Cy5 and then incubated them with BMDM for various amounts of time (Fig. 2b–c). We have used defined cell lines as a source of microparticles to rule out any potential confusion with immune complexes (30). We found that ectosomes were readily bound by macrophages at 4°C, and internalized at 37°C (Fig. 2b–c). These ectosomes did not readily colocalize with early endosomes or fluid phase markers like dextran, but did localize to some compartments staining positive for Niemann-Pick Type C protein 1. This protein is a resident of late endosomes and assists with cholesterol metabolism (31). Thus, these ectosomes are readily bound and internalized by BMDM.

Fig 2.

Microvesicle production and uptake by macrophages. (A) Ectosomes were produced by treating 3T3 cells with Streptolysin O for 15 min at 37°C, collected by ultracentrifugation of clarified supernatants, negatively stained and visualized by electron microscopy. Arrowheads indicate SLO pores visible on the ectosomes. (B) Ectosomes produced from Cy5-labeled 3T3 cells were added to BMDM and incubated at 37°C for 60 min (solid line), 30 min (dotted line) or at 4°C for 60 min (dashed, tinted line) prior to analysis of Cy5 fluorescence by FACS. Untreated BMDM are indicated by the filled black line. (C) Ectosomes produced from Cy3 labeled 3T3 cells were added to BMDM for the indicated times and fixed, stained with DAPI and visualized by epifluorescence. MV indicates fluorescently labeled ectosomes. Scale bar is 100 nm in A, 10 μM in B.

It has recently been appreciated that lipid particles internalized by macrophages influence whether they become activated, or are deliberately suppressed. It is possible that these microparticles serve as danger signals, and activate macrophages in preparation of dealing with a threat. Oxidized LDL (oxLDL) is one such lipid particle that activates macrophages and induces an inflammatory response (32). This chronic activation is thought to lead to foam cell formation and atherosclerosis. However, microparticles shed by tumor cells have also been reported to be immunosuppressive (33), and ectocytosis might block further elaboration of the immune response (34). Specifically, for apoptotic bodies, the simultaneous presence or absence of infection determines whether they activate or tolerize the immune system (35). To discern whether ectosomes activate or tolerize macrophages, we treated BMDM with ectosomes and measured their activation status. Macrophages treated with ectosomes showed slight activation, though not to the same extent as priming with LPS or treatment with oxLDL. Furthermore, treatment of BMDM with microvesicles for 2 days turned the macrophages into foam cells, similar to treatment with oxLDL. This foam cell formation was enhanced if the BMDM was simultaneously stimulated with a TLR ligand. This suggests that treatment with lipid particles in an activating environment will lead to macrophage activation and foam cell formation.

However, ectosomes blocked TNF production by TLR-stimulated BMDM. Also, ectosomes blocked the presentation of SIINFEKL to B3Z T cell hybridomas, suggesting that they serve in an inhibitory role. These results stand in contrast with those microvesicles generated from cells in response to ATP and calcium influx. High concentrations of ATP activate the P2X7 receptor, and together with calcium result in microvesicle shedding (36). The microvesicles generated here are a heterogeneous mixture of internal and external membrane-derived particles, and unlike pure ectosomes, drive activation of macrophages in a phospholipase D dependent manner (36). Thus, macrophages can discriminate between vesicle types, though the precise protein and lipid components required for this discrimination remain to be determined.

Conclusions

Pore-forming toxins rapidly stimulate macrophages to release IL-1β and other cellular components such as HMGB-1 that can promote inflammation both locally and systemically. The NLRP3 inflammasome mediates the release of these effectors, and is consistent with its role in other diseases involving infectious or sterile inflammation. During infections with toxin producing bacteria, it seems likely that CDCs would interact with multiple cell types in the local environment. Microvesicles shed by toxin-coated cells would likely be internalized by macrophages, some of which might have bound toxin directly and be secreting pro-inflammatory cytokines as described above. Thus, in vivo at the site of infection, toxins may exert competing effects on macrophages, with a balance observed between pro-inflammatory (direct toxin binding leading to IL-1β and HMGB1 release) and anti-inflammatory pathways (toxin-induced MV from adjacent cells that suppress macrophage TNF-α release). Experimental models to test which pathway may be dominant and whether the balance could be shifted therapeutically will be difficult to establish, but would be invaluable for understanding the interaction between the host and bacteria producing these toxins.