Type I IFN (IFN-I) signaling is detrimental to cells and mice infected with Listeria monocytogenes. In this study, we investigate the impact of IFN-I on the activity of listeriolysin O (LLO), a pore-forming toxin and virulence protein released by L. monocytogenes. Treatment of macrophages with IFN-β increased the ability of sublytic LLO concentrations to cause transient permeability of the plasma membrane. At higher LLO concentrations, IFN-β enhanced the complete breakdown of membrane integrity and cell death. This activity of IFN-β required Stat1. Perturbation of the plasma membrane by LLO resulted in activation of the p38MAPK pathway. IFN-β pretreatment enhanced LLO-mediated signaling through this pathway, consistent with its ability to increase membrane damage. p38MAPK activation in response to LLO was independent of TLR4, a putative LLO receptor, and inhibition of p38MAPK neither enhanced nor prevented LLO-induced death. IFN-β caused cells to express increased amounts of caspase 1 and to produce a detectable caspase 1 cleavage product after LLO treatment. Contrasting recent reports with another pore-forming toxin, this pathway did not aid cell survival as caspase1-deficient cells were equally sensitive to lysis by LLO. Key lipogenesis enzymes were suppressed in IFN-β-treated cells, which may exacerbate the membrane damage caused by LLO.
Listeriolysin O (LLO)3 is a cholesterol-dependent cytolysin, a family of toxins secreted by Gram-positive bacteria (1, 2). Structural studies have provided much insight into the molecular mechanism of pore formation by this family. Initially, the secreted monomer binds to cholesterol-containing membranes via the conserved undecapeptide motif in domain 4 (3, 4). Lateral diffusion and oligomerization of the monomers is followed by structural changes in the molecule (5, 6), resulting in pores consisting of 35–50 monomers, with a diameter of 25–35 nm (7). LLO, together with the other listerial cholesterol-dependent cytolysins (seeligerolysin and ivanolysin), are distinct within the wider family for having an optimum activity at acidic rather than neutral pH (8), attributed to a triad of acidic residues unique to this subgroup of toxins (9). These trigger rapid and irreversible denaturation of the structure at neutral pH at temperatures above 30°C. However, it has been recently shown that binding and permeabilization of cells is possible at neutral pH, provided the cholesterol content of the membrane is sufficiently high (10).
LLO is critically required during Listeria infection, and Listeria deficient for LLO are avirulent in mouse infection studies (11). LLO expression allows escape of bacteria from phagosomes into the cytoplasm and secondary cell-to-cell spreading, although in some cell types its loss may be compensated for by bacterial phospholipases (12). The acidic pH optimum, together with rapid degradation within the cell (13), restricts its intracellular activity and prevents premature death of the host cell (13, 14). The action of LLO during infection, however, may not be restricted to the phagosome, given that the toxin will be released from extracellular bacteria and from dead cells. Indeed, lesions in the spleen resulting from extensive lymphocyte death during Listeria infection are shown to be due to extracellular LLO release (15). These exogenous actions can contribute to the pathogenesis of infection, for example, the apoptotic lymphocyte death previously mentioned is reported to trigger a down-regulation of the inflammatory response, to the detriment of the host (16).
Exogenous LLO, and also other pore-forming toxins, were shown to trigger signaling events in a variety of cell types (2). In some instances, including MAPK activation (17) and calcium signaling (18, 19, 20), this is attributed to its pore-forming ability at the plasma membrane and also potentially at organelle membranes. It is also suggested that signaling may be triggered by LLO binding to a cell surface receptor, specifically TLR4 (21), although no direct binding was shown. An alternative model for LLO-induced signaling is also proposed, based on the observation that LLO oligomerization induces lipid raft aggregation in the cell membrane (22). Subsequent pore formation is not required for this process, nor for the tyrosine kinase activation, induced presumably through signaling complex congregation in the raft. We wanted to examine which signals could be induced in macrophages by exogenous LLO, and also compare this with the signals induced by Listeria infection, to understand to what extent LLO contributes to signaling induced by the whole bacteria.
Pore-formation and signaling events are linked to cell death induced by LLO and other pore-forming toxins, at concentrations less than those required for instant lysis of nucleated cells. Reports for LLO-induced death include apoptotic death in dendritic cells (23), and in T lymphocytes (15), where caspase-dependent and -independent death has been shown. Ion fluxes initiated by pore opening are suggested to be the trigger for death in the case of various other pore-forming toxins, including Staphylococcus aureus α-toxin (24, 25) and aerolysin from Aeromonas hydrophila (26).
The case of T lymphocyte death upon LLO treatment is particularly intriguing, as it is shown to be up-regulated upon pretreatment of the lymphocytes with type I IFN (IFN-I) (27). IFN-I production, induced during early stages of Listeria infection in vivo, is detrimental to the innate immune response of the host (27, 28, 29). Macrophages are host cells for invading Listeria within the spleen and liver. When activated, they are able to destroy phagocytosed bacteria. In vitro experiments, however, show infection of resting macrophages results in a slow, predominantly necrotic death (30), and this death is increased by IFN-β production and signaling triggered by the cytoplasmic invasion of the macrophage (31). In this case, LLO expression by Listeria is necessary for cytoplasmic invasion and IFN-β production, but the necrotic death is in part attributed to IFN-dependent up-regulation of inducible NO synthase (63). The lytic activity of LLO may both contribute to the death of infected macrophages and affect noninfected cells as in the case of T lymphocytes. Therefore, we sought to determine whether IFN-I could also affect the response of macrophages to LLO alone. We report that pretreatment of macrophages with IFN-β critically increased the extent of pore formation by LLO as well as the rate of death. This may prove important in explaining how IFN-I influences the pathogenesis of Listeria infection.
Materials and Methods
Plasmids encoding six-histidine tagged rLLO, either full-length (rLLO529) or N-terminally truncated (rLLO493; a gift of M. Mitsuyama, Kyoto University, Kyoto, Japan) (32) within the pQE31 vector, were transformed into XL-10 Gold Escherichia coli (Stratagene). rLLO expression in cultures for purification was induced by addition of 0.3 mM isopropyl β-d-thiogalactoside, and the culture was grown for a further 3 h at 28°C. The pellet from a 600-ml culture was resuspended in 10 ml of lysis buffer (0.5 M sodium chloride, 50 mM sodium dihydrogen phosphate (pH 7.0)), lysozyme (end concentration 1 mg/ml) was added, and incubated on ice for 30 min. This was followed by sonication for 6 × 10 s and centrifugation at 12,000 × g for 30 min at 4°C. The supernatant was passed through a 27 G needle, and 250 mM imidazole elution buffer was added such that the end imidazole concentration was 20 mM. A total of 500 μl of prewashed His-bind beads (Qiagen) were added and the mixture rotated for 1 h at 4°C. Beads were captured by pouring through a column, washed with 5 × 10 ml of lysis buffer (pH 7.0) containing 20 mM imadazole, then trapped protein was eluted with 5 ml of lysis buffer (pH 7.0) containing 250 mM imadazole. The eluate was dialyzed twice into storage buffer (0.5 M sodium chloride, 10 mM sodium phosphate (pH 7.0), 0.5 mM EDTA (33)). Endotoxin was removed by multiple passes through Detoxi-Gel endotoxin removal columns (Pierce), and end concentration was measured as <20 pg/ml for the 90 μg/ml stock solution (Rapid Endotest Service; Cambrex). Protein was stored frozen in aliquots at −80°C. Protein concentration was measured by Bradford assay (Bio-Rad), using BSA standards.
Cells, reagents, and infection
Bone marrow-derived macrophages were obtained by culture of bone marrow from 7- to 11-wk-old mice of C57BL/6 genetic background in L cell-derived CSF-1 as described (34). MyD88-deficient (35) and TLR4-deficient mice (36) were as described and provided by S. Akira (Osaka University, Osaka, Japan). Stat1-deficient (37) mice were as described and provided by D. Levy (New York University School of Medicine, New York). Caspase 1-deficient mice (38) were a gift of A. Zychlinsky (Max Planck Institute for Infection Biology, Berlin, Germany). Recombinant murine IFN-β (Calbiochem) was added to a final concentration of 500 U/ml. SB203580 (Calbiochem) was added 30 min before LLO addition to a final concentration of 10 μM. LLO treatment of cells was done in DMEM medium with 3% FBS. Dextran 500 (m.w. 500,000; GE Healthcare) was dissolved in this medium where indicated before use. Infection of macrophages with Listeria monocytogenes (strain LO28 (39)) was as described in (31), using wild-type or LLO-deficient (Δhly) bacteria. Infection of macrophages with Salmonella typhimurium strain SR11 was as described in Ref. 40 .
Assays of cell death
Crystal violet staining and measurement of lactate dehydrogenase (LDH) release was as described in (31).
Western blot analysis
Western blot analysis was done as described in Ref. 41 . Rabbit phosphospecific Ab against p38 MAPK and rabbit antiserum against p38 MAPK were obtained from New England Biolabs and used at dilution 1/1000, analyzing separate blots for the same samples to prevent overlapping signals. Ab recognizing Thr202 and Tyr204 phosphorylated ERKs (phospho-ERK) was obtained from Cell Signaling Technologies, Ab recognizing ERKs (pan-ERK) was obtained from BD Transduction Laboratories, both used 1/1000. Ab recognizing IκBα was obtained from Santa Cruz Biotechnology and used 1/1000. Ab recognizing Tyr701 phosphorylated Stat1 was obtained from New England Biolabs and used 1/1000, while Ab recognizing the C terminus of Stat1α was as described in Ref. 41 . Ab recognizing precursor caspase 1 and the p10 subunit from cleavage was from Santa Cruz Biotechnology and used 1/100. All primary Abs except caspase 1 for short exposure were probed with fluorescence-labeled secondary Ab (Molecular Probes). This was detected by the infrared imaging system Odyssey (LI-COR) and quantified using associated software. Caspase 1 for short exposure was probed with HRP-labeled secondary Ab and visualized using ECL.
Patch clamp measurements
Macrophages were plated onto glass coverslips and treated where indicated with IFN-β overnight. For measurement, macrophages on coverslips were superfused with an extracellular solution containing (in millimoles) 150 NaCl, 5.4 KCl, 0.5 MgCl2, 1.8 CaCl2, 10 HEPES, 5 glucose (pH 7.4) adjusted with NaOH. The recording pipette contained (in millimoles) 140 potassium aspartate, 2 MgCl2, 2 CaCl2, 10 HEPES, adjusted to pH 7.4 with KOH. Currents were recorded in the perforated patch clamp configuration using amphothericin B (600 μg/ml) in the pipette solution. All measurements were performed at 20–22°C. Electrodes pulled of borosilicate capillaries had a tip resistance of 5–6 M Ω. After patch formation, an equilibrium period of 5 min followed, which was succeeded by a control period of 5 min at a holding potential of −30 mV during which no electrical activity could be observed. Then, LLO was added to the bathing solution at concentrations of 2.5 or 5.0 μg/ml. First channel openings were detected ∼1 min after addition of LLO. Electrophysiological measurements were conducted with an Axopatch-1D patch clamp amplifier (Axon Instruments) at a cutoff frequency (−3 dB) of 2 kHz. Currents were filtered at 5 kHz and sampled at 10 kHz. Data acquisition was executed directly to a PC with pCLAMP 6 software (Axon Instruments). Current analysis was performed with ASCD software (G. Droogmans, Katholieke Universiteit Leuven, Leuven, Belgium).
Reverse transcription and real-time PCR
Total RNA extraction, cDNA synthesis, and real-time PCR was conducted as described in Ref. 42 , using the GAPDH housekeeping gene as an endogenous control with primers as described. Primers for HMGCoA reductase and fatty acid synthase were as described in Ref. 43 . Primers for caspase 1 were as described in Ref. 44 , primers for caspase 11 were as described in Ref. 45 , primers for apoptosis-associated speck-like protein containing a CARD were as described in Ref. 46 .
For FACS analysis, macrophages were harvested by incubation for 3–5 min in citric saline (0.135 M potassium chloride, 0.015 M sodium citrate), centrifugated together with the medium from the plate to collect cells in suspension, spun for 7 min at 300 × g at 4°C, and resuspended in PBS containing 0.2% BSA before analysis by FACSCalibur (BD Biosystems). Annexin V/7-aminoactinomycin D (7AAD) staining was conducted according to the manufacturer’s instructions using annexin V-allophycocyanin and 7-AAD (BD Biosciences).
Statistical significance was assessed where indicated using the Student t test, two-tailed, assuming equal variance, and was calculated using Microsoft Excel software. Significance is indicated in figures using the convention: ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001.
IFN-β enhances the lytic activity of LLO
Addition of LLO to primary macrophages triggered cell death in a concentration-dependent manner (Fig. 1, A and B). The time course of LDH release indicated that onset of death was rapid, within the first few hours after toxin addition. The long-term survival of cells was assayed by staining adherent cells with crystal violet 24 h after toxin addition. Cell death was visible in this assay at LLO concentrations at 0.25 μg/ml and above. IFN-β treatment alone did not cause cell death, but a larger number of pretreated macrophages died after LLO addition. IFN-β triggers Stat1-dependent as well as Stat1-independent pathways (47). IFN-β pretreatment of Stat1-deficient macrophages was without consequences for the lytic activity of LLO (Fig. 1 C), showing that the sensitization effect is mediated by Stat1.
Addition of the pan-caspase inhibitor z-VAD-fmk before LLO exposure led to only a small reduction in death, and oligonucleosomal DNA fragmentation (typical of caspase-mediated cell death) was detected only at the highest LLO concentration tested (1.0 μg/ml; data not shown). Cell death, therefore, may occur by both caspase-dependent and -independent mechanisms. The exposure of the membrane lipid phosphatidylserine (PS) on the outer leaflet of the cell membrane is associated with apoptosis, and can be monitored by assaying binding of its ligand annexin V. Combining this assay with monitoring membrane permeability with the nuclear dye 7-AAD, we saw a rapid increase (within the first hour) in the population of annexin V-positive cells (Fig. 1 D). Upon IFN-β pretreatment, more of this annexin V-positive population was also simultaneously 7-AAD positive. This pattern did not change at later time points. It suggests that IFN-β does not increase an apoptotic signal actuated by LLO.
Signal transduction by LLO differs from that stimulated by viable L. monocytogenes
Given the range of signaling pathways reported to be activated by LLO in various cell types, we examined which of these were triggered by LLO treatment of primary macrophages. We observed robust and rapid activation of p38MAPK at concentrations below those causing notable death (Figs. 2,A and 3,B). A truncated form of LLO, missing domain 4 and hence lacking cytolytic activity (32) did not produce this response. As this was purified in parallel and stored in the same buffer, this also indicates contaminating endotoxin is not responsible for the signaling seen. p38MAPK activation occurred after LLO addition in MyD88-deficient and TLR4-deficient cells. In contrast, early activation of p38MAPK by Listeria or LPS required MyD88 or TLR4, respectively (Fig. 2, A and B). p38MAPK activation by Listeria was not dependent on LLO secretion, as activation was also seen upon infection with LLO-deficient bacteria. This is consistent with the demonstrated TLR2 dependence of early p38 activation upon Listeria infection (48), with TLR2 recognizing Listeria cell wall components (49). Partial MyD88 independence of p38MAPK activation was noted only at later stages of Listeria infection (Fig. 2 B). The MyD88-dependent activation of p38MAPK by Listeria, or TLR4-dependent activation by LPS, was maximal at 30 min, while a maximal signal was present already at 10 min after LLO treatment. TLR4 independence of LLO-induced signaling questions the reported role of TLR4 as an LLO receptor (21).
p38MAPK activation can be stimulated by treatment with detergents (17). Similarly, exposure of epithelial cells to pore-forming toxins results in osmotic stress-induced activation of p38MAPK (50). In line with these results, activation of the enzyme by LLO did not occur when cells were treated in medium containing dextran (Fig. 2 C), which offers protection against osmotic stress and hemolysis caused by toxin pore formation (51).
The p38MAPK pathway was shown to support cell survival when innate immune cells are exposed to bacterial products including LPS or anthrax lethal toxin (52, 53). We used the specific inhibitor SB203580 to test whether p38MAPK regulates the viability of cells treated with LLO. In agreement with a previous report for another cholesterol-dependent cytolysin, streptolysin O (54), p38MAPK inhibition did not influence the lytic action of LLO (Fig. 2 D).
Contrasting p38MAPK, other examined signaling pathways were not or only marginally stimulated by LLO. Tyrosine-phosphorylated Stat1, a readout for IFN production and signaling, was undetectable even at the highest concentration of LLO tested (Fig. 2,E). Weak activation of ERK2, and some slight degradation of IκBα, could be detected only when a high concentration of the toxin was used (Fig. 2,F and data not shown). These signals were far weaker than those induced by Listeria infection or LPS treatment, which both lead to IFN production, robust IκB degradation, and activation of both ERKs 1 and 2. As in the case of p38MAPK, IκB degradation or ERK activation by viable bacteria showed partial dependence on MyD88, particularly in the early phase of infection (Fig. 2 G).
IFN-β increases transient pore formation and the loss of plasma membrane integrity
Our previous results show that IFN-β increases the activity of cellular pathways contributing to Listeria-induced macrophage death (63). Results shown in Fig. 1 suggest that IFN-I may additionally increase the damage caused to the infected cell by LLO. To further address this possibility, the amount of pore formation upon toxin treatment was assessed by adding toxin to cells with the membrane-impermeant nucleic acid stain propidium iodide present in the medium. Cells were harvested after 1 h and analyzed by FACS to measure propidium iodide uptake. A very bright population, labeled region B, increased in frequency with LLO concentration and with IFN-β pretreatment (Fig. 3 A). This corresponded to cells which had lost significant membrane integrity to allow propidium iodide to enter freely, and were counted as dead. The dim population, labeled region D, corresponded to living cells. This population displayed a small increase in mean fluorescence when toxin was added, compared with untreated cells, though remained distinct from the dead cell population. This is most likely due to temporary pore formation in the cell membranes allowing limited propidium iodide entry, but with later repair of these pores to prevent unlimited entry. Such limited entry has been previously described in the context of perforin-mediated damage (55). As expected, the mean fluorescence of this population increased with LLO concentration, as more pore formation could occur. Notably, IFN-β pretreatment resulted in a higher mean fluorescence than for nonpretreated cells at the same LLO concentration, suggesting it increases membrane damage suffered by cells upon exposure to LLO. Such a phenomenon would explain the increased death observed.
To further confirm these results, we measured ion flux from pore formation in macrophages exposed to LLO, with or without IFN-β pretreatment, using the perforated patch clamp technique to measure currents generated across the plasma membrane of one cell (Fig. 3 B). After a control period without electrical activity, LLO was added to the external solution bathing the cells and, after ∼1 min, single channel currents could be observed. The pore formation progressed rapidly and multiple openings occurred. At least three elementary pore current amplitudes of 16.5 ± 1.5, 32.0 ± 2.0, and 64.1 ± 6.2 pA were measured at a holding potential of −30 mV, being multiples of the smallest pore current amplitude. Pore openings were rare and occurred in bursts. At a concentration of 0.5 μg/ml, LLO caused channel openings only of small amplitude (n = 3). The current was markedly increased in IFN-β-pretreated macrophages (n = 3). Multiple openings were observed, which were not seen in not-pretreated cells under the same experimental conditions. At 1.0 μg/ml LLO (n = 2), obviously more pores were incorporated into the cell membrane as multiple openings were frequently seen, and again membrane current was larger in IFN-β-pretreated cells (n = 3) (data not shown). The frequency of pore openings slowly decreased after ∼10 min.
Quantification of p38MAPK activation by low concentrations of LLO showed greater activation in IFN-β-pretreated cells compared with nonpretreated for the same concentration (Fig. 3,C). As this is dependent on membrane permeabilization (Fig. 2 C), increased activation suggests greater membrane damage.
Caspase 1 activation by LLO is enhanced in IFN-β-stimulated macrophages, but makes no contribution to the lytic activity of LLO
The action of pore-forming agents, such as nigericin and aerotoxin, can act as a trigger for inflammasome activation through decreasing intracellular potassium levels (56, 57, 58), and it is suggested that LLO may be important in activation of the inflammasome during Listeria infection (59). Low levels of cleavage of caspase 1 were apparent upon LLO addition to IFN-β-pretreated cells, but not detectable in those not pretreated (Fig. 4,A). In this experiment, we also observed that protein expression of full-length caspase 1 was greater upon IFN-β pretreatment. Up-regulation of caspase 1 gene expression in a Stat1-dependent manner was shown by real-time PCR (Fig. 4,B). Likewise, IFN-β increased the expression of another potential inflammasome component, caspase 11. By contrast, expression of apoptosis-associated speck-like protein containing a CARD, a key inflammasome adapter protein, remained unchanged (Fig. 4,C). Caspase 1 has been linked to cell death caused by infection with Salmonella or Shigella. However, we saw no difference in LLO-induced cell death, or in the up-regulation of death by IFN, in caspase 1-deficient macrophages when compared with wild type (Fig. 4, D and E).
Up-regulation of lipogenesis pathways (via caspase 1 activation) is reported to promote cell survival in response to aerolysin (58), presumably by promoting membrane repair. Interestingly, we observed a considerable down-regulation by IFN-β of HMG CoA reductase and fatty acid synthase (Fig. 5,A), two genes involved in lipogenesis, whose up-regulation was linked to cell survival after aerolysin treatment. The basal levels of expression in Stat1-deficient cells were reduced, but the decrease seen was Stat1 dependent (Fig. 5 B). Unlike for aerolysin, we did not observe up-regulation of these genes upon LLO treatment, with or without IFN-β pretreatment (data not shown). This corresponds with the comparatively low level of caspase 1 activation observed in our system, and may be attributed to a difference in toxin, cell type, or species.
IFN-β is primarily associated with a protective response to viral infection, but its important role in other pathological situations, such as tumor immunity and bacterial infection, is becoming more apparent. In this study, we show that IFN-β can increase the damage sustained and death caused when macrophages are exposed to LLO, the pore-forming toxin secreted by Listeria. This effect pertains directly to the extent of pore formation at the plasma membrane. To our knowledge, such a direct effect of IFN-I on membrane damage has not been previously described. Our studies comparing signal transduction by cells infected with viable L. monocytogenes with that stimulated by LLO suggest that the cytolysin does not make an important contribution to the signals emitted during entry of the bacteria. It may contribute to signaling during later stages of infection, but this remains unclear. However, LLO released from extracellular bacteria or from dead cells may cause damage to neighboring, uninfected cells in analogy to the situation reported for T lymphocytes. This could be particularly relevant when infiltration of infected organs or the formation of abscesses and granulomas leads to dense, macrophage-containing cell aggregates. Because L. monocytogenes infection causes IFN-I production, its ability to increase susceptibility to LLO-mediated membrane perturbation and damage is likely to make a further contribution to the harmful sequelae of infection.
Our studies did not confirm an important role for TLR4 as an LLO receptor. Signaling by either viable Listeria or by LLO was unchanged in TLR4-deficient cells (Fig. 2). Furthermore, exogenous LLO did not stimulate IFN-I production and Stat1 tyrosine phosphorylation that would result from the TIR domain-containing adapter inducing IFN-β pathway downstream of TLR4 (Fig. 2).
More detailed studies of events surrounding pore formation and repair will be needed to delineate how these processes are affected by IFN-β. The inhibitory effect of IFN-β on lipogenesis pathways may prove important (Fig. 5). There are two possibilities for the action of IFN-I on LLO. Given that binding and pore formation by LLO is dependent on membrane composition, it is conceivable that IFN-β causes changes in this such that the extent of pore formation is increased. Alternatively, closure of pores and cell survival requires a membrane resealing process (60), which may be influenced by lipogenesis pathways. Inhibition of this could extend the damage caused by pore formation, leading to increased death stemming directly or indirectly from the continued ion movement and osmotic stress. IFN-β pretreatment of macrophages did not increase phagosomal escape of Listeria (our unpublished observations). This suggests that the composition/repair of the phagosomal membrane is not affected by IFN-β activity or, more likely, that the increased activity of LLO at acidic pH obscures the enhancement of membrane disruption by IFN-β.
The observed increase in intracellular signaling events may be attributed at least in part to the increase in membrane damage. The activation of p38MAPK is due to osmotic stress (Fig. 2), and is increased when cells are IFN-β pretreated (Fig. 3). Pore formation at the membrane by cholesterol-dependent cytolysins including LLO has previously been shown to cause ion fluxes, such as the efflux of potassium ions and the influx of calcium ions, and caspase 1 activation by pore-forming toxins is attributed to ion fluxes. Caspase 1 activation is only observed in pretreated cells (Fig. 4) and may be due both to the increase in damage and the up-regulation by IFN-β of caspases 1 and 11. A role for IFN-β in up-regulating inflammasome activation and cell death has been recently described in the context of Francisella tularensis and Listeria infections (61). Because caspase 1 protein levels remained unchanged under the experimental conditions of this study, the authors concluded that other IFN-up-regulated factors may aid inflammasome activation. Such additional mechanisms might contribute further to the IFN-β effects on LLO-treated cells. Despite the increased caspase 1 activation by LLO following IFN-I treatment, the protease was not required for death (Fig. 4), or PS translocation (data not shown).
Although PS on the cell surface accompanies LLO-induced cell death, the effect of IFN-I is not to increase the proportion of annexin V-positive macrophages, but rather to enhance their progression to a final death stage characterized by permanent plasma membrane permeability (Figs. 1 and 3). This is in contrast to the effect of IFN-I on LLO-treated T-lymphocytes (15, 27), characterized by an increase in PS exposure. Translocation of PS in nucleated cells can be triggered by sustained elevation of cytosolic calcium, without the need for other apoptotic processes (62). Calcium influx is shown to occur in cells treated with LLO (18, 19), and this might trigger exposure in the outer leaflet of the plasma membrane. The decrease caused by IFN-β of cells which expose PS but still remain impermeable to 7AAD, is then presumably due to obstruction of their ability to repair membrane damage.
In conclusion, our study suggests a new mechanism of action of IFN-I during Listeria infection, and this could contribute to the detrimental effects of these cytokines seen during in vivo infection. Pore-forming toxins are common virulence factors among pathogenic bacteria. That a cytokine can increase a toxin’s effect on immune cells is a novel concept and it will be important to understand its role during bacterial infections.
We thank colleagues who provided materials or mouse lines, as mentioned in Materials and Methods, and all colleagues in the T. Decker laboratory for help, discussion, and advice. We also thank Christoph Romanin (Institute of Biophysics, University of Linz, Linz, Austria) for critical reading of the manuscript.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by the Austrian Science Foundation (Fonds zur Förderung der wissenschaftlichen Forschung) through Grants AP17859 and P20522-B05 (to T.D.), SFB 28 (to T.D.), and a PhD fellowship (to H.Z.). Additional funding was provided by the Viennese Foundation for Research and Technology (HOPI Initiative; to T.D.).
Abbreviations used in this paper: LLO, listeriolysin O; IFN-I, type I IFN; LDH, lactate dehydrogenase; 7AAD, 7-aminoactinomycin D; PS, phosphatidylserine.