The FASEB Journal • FJ Express Summary Why Escherichia coli ␣-hemolysin induces calcium oscillations in mammalian cells–the pore is on its own ¨ nver,* Florian Dreyer,* Andreas Koschinski,*,1 Holger Repp,*,1,2 Baris U † Dierk Brockmeier,* Angela Valeva, Sucharit Bhakdi,† and Iwan Walev†,2 *Rudolf-Buchheim-Institute of Pharmacology, Justus-Liebig-University, Giessen, Germany; and † Institute of Medical Microbiology and Hygiene, Johannes-Gutenberg-University, Mainz, Germany To read the full text of this article, go to http://www.fasebj.org/cgi/doi/10.1096/fj.05-4561fje SPECIFIC AIMS The hemolysin of Escherichia coli (HlyA) is prototypic of a large family of pore-forming toxins that are produced by many medically important gram-negative pathogens. It was reported that HlyA deregulates membrane Ca2⫹ channels, thus inducing periodic low-frequency Ca2⫹ oscillations that trigger transcriptional processes in mammalian cells. Our aim in this study was to delineate the mechanism underlying the Ca2⫹ oscillations that are induced by HlyA. PRINCIPAL FINDINGS 1. HlyA induces nonperiodic Ca2ⴙoscillations in renal cells Within ⬃30 s after application of HlyA (5 ng/ml) to human kidney tubule epithelial (IHKE) cells, intracellular Ca2⫹ concentration ([Ca2⫹]i) increased and exhibited oscillations during further monitoring. It was previously reported that HlyA provoked periodic Ca2⫹ oscillations in primary rat proximal tubule (RPT) cells, with a constant periodicity of Ca2⫹ peaks of 12 min. However, we found no constant periodicities but instead found large variations of oscillation kinetics and peak-to-peak spans. Examples are shown in Fig. 1A–D for Ca2⫹ tracings and in Fig. 1E for false-color images of [Ca2⫹]i. No Ca2⫹ oscillations were induced by the nonhemolytic HlyA mutant S177C/K564R/K690R, which is fully bindable to the plasma membrane but unable to form membrane pores (Fig. 1G). Statistical analysis of time spans between directly neighboring Ca2⫹ peaks in Fig. 1A-D and 48 other Ca2⫹ tracings yielded a most frequent peak-to-peak span (not to be confused with peak periodicity) of 70 s, whereas spans above 10 min were very rare. Peak-to-peak span distribution was dependent on HlyA concentration, whereupon higher concentrations led to shorter time spans. 0892-6638/06/0020-0973 © FASEB 2. A periodicity of Ca2ⴙ peaks is pretended by a power spectrum analysis We subjected the Ca2⫹ data sets to a power spectrum analysis, which was previously used to identify HlyAinduced periodic Ca2⫹ oscillations of low frequency. Despite marked heterogeneity of the original Ca2⫹ tracings and rare occurrence of peak-to-peak spans above 10 min, this analysis yielded a periodicity of Ca2⫹ peaks with a dominant wavelength of 12.8 min, which was in remarkable agreement with the previously reported value of 12.0 min. Surprisingly, in Ca2⫹ tracings monitored in the absence of HlyA and devoid of Ca2⫹ elevations, a periodicity of ⬃12 min was again computed. Furthermore, analysis of a signal consisting only of white noise superimposed to a discretely drifting baseline yielded the same periodicity. In conclusion, power spectrum analysis appeared as an inadequate mathematical model for a proper interpretation of the Ca2⫹ data sets. 3. Nifedipine does not inhibit Ca2ⴙ oscillations When voltage-gated L-type Ca2⫹ channels in IHKE cells were blocked by nifedipine, HlyA still led to Ca2⫹ oscillations (Fig. 1F). This contrasted with previous data of RPT cells where nifedipine abolished the Ca2⫹ response to HlyA. A striking phenomenon that we observed during the Ca2⫹ measurements can account for this difference. Nifedipine-treated cells that were in the microscopic field of view for several minutes during Ca2⫹ imaging, i.e., irradiated by ultraviolet (UV) light, 1 These authors contributed equally to this work. Correspondence: Frankfurter Str. 107, Rudolf-BuchheimInstitute of Pharmacology, Justus-Liebig-University, 35392 Giessen, Germany. Email: [email protected]; and Obere Zahlbacher Str. 67, Institute of Medical Microbiology and Hygiene, Johannes-Gutenberg-University, 55101 Mainz, Germany. Email: [email protected] doi: 10.1096/fj.05-4561fje 2 973 Figure 1. Ca2⫹ oscillations induced by HlyA in IHKE cells loaded with the fluorescent Ca2⫹ indicator Fura-2/acetoxymethyl ester. A–D) Cells were exposed to HlyA (5 ng/ml) 3 min after start of recording. Traces represent examples of time courses of [Ca2⫹]i of 4 single cells. E) Single frames from an image series of [Ca2⫹]i. Frames 1– 4 correspond to numbers 1– 4 above trace in A. Traces in A and B were recorded from cells that are marked with A and B. F) Superposition of Ca2⫹ responses of 49 IHKE cells in the presence of nifedipine (100 M). Application of HlyA (5 ng/ml) is indicated by arrow. Gap represents time when field of view was changed to an area that had not been UV irradiated before. G) Ca2⫹ tracing of a single IHKE cell on consecutive exposure to a nonhemolytic HlyA mutant (⫽nh HlyA, 25 ng/ ml) and wild-type HlyA (2.5 ng/ml). did indeed not show Ca2⫹ oscillations after application of HlyA, whereas Ca2⫹ oscillations were observed in cells that had not “seen” UV light (Fig. 1F). Thus, the previously reported suppressive effect of nifedipine on HlyA-induced Ca2⫹ oscillations was uncovered as experimental artifact. Additionally, HlyA still led to Ca2⫹ oscillations in HEK293 cells when a combination of blockers of voltage-gated and receptor-operated Ca2⫹ channels was present, including nifedipine (100 M), Cd2⫹ (2 mM), and SK&F 96365 (25 M). 4. Ca2ⴙ oscillations and pore formation cease rapidly on removal of HlyA The effect of HlyA on [Ca2⫹]i is essentially dependent on the presence of toxin in extracellular medium. Instantaneously after the start of a washout in HlyAtreated IHKE cells, [Ca2⫹]i decreased and reached initial concentration after ⬃3 min. Reapplication of HlyA again led to Ca2⫹ oscillations, which disappeared after the restart of washout. After the start of washout, disappearance of Ca2⫹ oscillations and HlyA pores followed the same time course. This synchronism was discovered by subjecting single IHKE cells exposed to HlyA to simultaneous measurement of [Ca2⫹]i and pore formation, using at the same time Ca2⫹ imaging and electrophysiological patch-clamp recording. A few 974 Vol. 20 May 2006 seconds after the start of washout the number of open pores decreased and went to zero within ⬃4 min. A second HlyA application again led to pore formation. The short life span of HlyA pores was confirmed by propidium iodide influx measurements. Cells became permeable to dye during incubation with HlyA, but membranes returned to the impermeable state shortly after toxin removal. 5. Ca2ⴙ elevations result from Ca2ⴙ influx through HlyA pores By simultaneous monitoring of pore formation and [Ca2⫹]i in IHKE cells exposed to HlyA, we detected an increase in [Ca2⫹]i directly after pore opening and a good temporal correlation between formation and closure of HlyA pores and Ca2⫹ elevations. We hypothesize that the decrease after an elevation reflects the restoration of cellular Ca2⫹ homeostasis by physiological Ca2⫹ redistribution after HlyA pore closure (Fig. 2). CONCLUSIONS AND SIGNIFICANCE The results of this study lead us to a new explanation for Ca2⫹ oscillations that occur in mammalian cells treated with HlyA. Contrary to a previous report, we showed that the ability of HlyA to induce Ca2⫹ oscilla- The FASEB Journal KOSCHINSKI ET AL. Figure 2. Schematic overview. Ca2⫹ oscillations induced by HlyA occur only in the presence of the toxin and are not blocked by nifedipine. After washout of HlyA, Ca2⫹ oscillations cease rapidly. Readdition of the toxin again leads to Ca2⫹ oscillations. We assume that the decrease after a Ca2⫹ elevation is managed by plasma membrane and sarco/endoplasmic reticulum Ca2⫹ pumps (illustrated by circles with blades). tions does not rely on endogenous L-type Ca2⫹ channels. Furthermore, we found that the effect of HlyA on [Ca2⫹]i is not dependent on the activity of any voltagegated or receptor-operated membrane Ca2⫹ channels. Instead, Ca2⫹ oscillations result from pulsed influxes of Ca2⫹ through short-lived HlyA pores, which are rapidly closed or removed from the membrane. The causal connection was directly seen in simultaneous patchclamp measurements of pore formation and Ca2⫹ measurements, where Ca2⫹ elevations occurred as a consequence of opening of HlyA pores. Pore formation by HlyA happens at random, which explains our observation that the Ca2⫹ peaks in individual cells occur in a stochastic sequence. This contrasts a previous claim on the existence of periodic low-frequency Ca2⫹ oscillations induced by HlyA. However, the mathematical model that this claim was based on turned out as inadequate for a proper interpretation of our measured Ca2⫹ data. Moreover, we found that the mean time span between two Ca2⫹ elevations was dependent on the concentration of HlyA, whereupon WHY E. COLI ␣-HEMOLYSIN INDUCES CALCIUM OSCILLATIONS higher concentrations led to shorter time spans. This concentration dependence definitely rules out the idea that HlyA molecules may switch on in the target cell a periodic “Ca2⫹ oscillation run” with one defined frequency. Washout experiments showed that pore formation, propidium iodide influx, and Ca2⫹ oscillations only occurred as long as HlyA molecules were present at the extracellular side. The observation that the Ca2⫹ oscillations ceased with the start of washout demonstrates further that the cell has no lasting memory on the encounter with HlyA regarding [Ca2⫹]i. Whensoever Ca2⫹ oscillations are initiated by HlyA, the pore is on its own. Nonperiodic (“chaotic”) Ca2⫹ oscillations resulting from the interplay of formation and disappearance of the pores along with cellular Ca2⫹ redistribution will obviously vary depending on toxin concentration and susceptibility and may provide the starting point of myriad reactions, but on a very individual basis, in cells attacked by pore-forming toxins such as HlyA. 975 The FASEB Journal • FJ Express Full-Length Article Why Escherichia coli ␣-hemolysin induces calcium oscillations in mammalian cells—the pore is on its own ¨ nver,* Florian Dreyer,* Andreas Koschinski,*,1 Holger Repp,*,1,2 Baris U † Dierk Brockmeier,* Angela Valeva, Sucharit Bhakdi,† and Iwan Walev†,2 *Rudolf-Buchheim-Institute of Pharmacology, Justus-Liebig-University, Giessen, Germany; and † Institute of Medical Microbiology and Hygiene, Johannes-Gutenberg-University, Mainz, Germany Escherichia coli ␣-hemolysin (HlyA), archetype of a bacterial pore-forming toxin, has been reported to deregulate physiological Ca2ⴙ channels, thus inducing periodic low-frequency Ca2ⴙ oscillations that trigger transcriptional processes in mammalian cells. The present study was undertaken to delineate the mechanisms underlying the Ca2ⴙ oscillations. Patch-clamp experiments were combined with single cell measurements of intracellular Ca2ⴙ and with flowcytometric analyses. Application of HlyA at subcytocidal concentrations provoked Ca2ⴙ oscillations in human renal and endothelial cells. However, contrary to the previous report, the phenomenon could not be inhibited by the Ca2ⴙ channel blocker nifedipine and Ca2ⴙ oscillations showed no constant periodicity at all. Ca2ⴙ oscillations were dependent on the pore-forming activity of HlyA: application of a nonhemolytic but bindable toxin had no effect. Washout experiments revealed that Ca2ⴙ oscillations could not be maintained in the absence of toxin in the medium. Analogously, propidium iodide flux into cells occurred in the presence of HlyA, but cells rapidly became impermeable toward the dye after toxin washout, indicating resealing or removal of the membrane lesions. Finally, patchclamp experiments revealed temporal congruence between pore formation and Ca2ⴙ influx. We conclude that the nonperiodic Ca2ⴙ oscillations induced by HlyA are not due to deregulation of physiological Ca2ⴙ channels but derive from pulsed influxes of Ca2ⴙ as a consequence of formation and rapid closure of HlyA pores in mammalian cell membranes.—Koschinski, A., ¨ nver, B., Dreyer, F., Brockmeier, D., ValRepp, H., U eva, A., Bhakdi, S., and Walev, I. Why Escherichia coli ␣-hemolysin induces calcium oscillations in mammalian cells—the pore is on its own. FASEB J. 20, E80 –E87 (2006) ABSTRACT Key Words: pore-forming toxins 䡠 RTX toxins 䡠 gram-negative bacteria 䡠 membrane lesions ESCHERICHIA COLI ␣-hemolysin (HlyA) is prototypic of a large family of pore-forming toxins that are produced by many medically important gram-negative pathogens. E80 Common to all are the presence of a Ca2⫹ binding nonapeptide repeat sequence and the requirement for posttranslational fatty acylation at one or two lysine residues for the acquisition of pore-forming activity (1, 2). In the absence of Ca2⫹ or fatty acylation, binding capacity is retained but the toxins are unable to adopt the pore-forming configuration (3, 4). The question of whether HlyA binds to specific cellular receptors is currently under investigation (5, 6); the toxin binds to protein-free lipid bilayers to form pores of ⬃1 nm diameter (7), similar to the pore size described in animal cell membranes (8). Pores formed in planar lipid bilayers flicker between an open and closed state dependent on the membrane potential (7, 9). In addition to their direct cytotoxic capacity, poreforming toxins can trigger a multitude of cellular responses that may in turn produce important longrange effects in the mammalian host organism. Many reactions are triggered by the uncontrolled flux of monovalent and divalent ions across the plasma membrane (10). Studies with S. aureus alpha-toxin and streptolysin O have disclosed that nucleated cells are able to repair a limited number of lesions (11, 12) and that this is accompanied by transcriptional responses such as the activation of NF-B and subsequent cytokine production (13, 14). Thus, similar to endotoxin, pore-forming toxins may provoke a large spectrum of long-term effects, and delineating the mechanisms underlying transcriptional activation represents a novel field of research with broad potential relevance. Oscillations in the cytoplasmic concentration of free Ca2⫹ have recently been recognized to represent a central event in transcriptional regulation (15, 16). Therefore, the finding that subcytocidal attack by HlyA is accompanied by periodic low-frequency Ca2⫹ oscilla1 These authors contributed equally to this work. Correspondence: Frankfurter Str. 107, Rudolf-BuchheimInstitute of Pharmacology, Justus-Liebig-University, 35392 Giessen, Germany. Email: [email protected]; and Obere Zahlbacher Str. 67, Institute of Medical Microbiology and Hygiene, Johannes-Gutenberg-University, 55101 Mainz, Germany. Email: [email protected] doi: 10.1096/fj.05-4561fje 2 0892-6638/06/0020-0080 © FASEB tions, transcriptional activation, and interleukin-6 production in renal epithelial cells represented an important discovery that might be extrapolatable to other pore-forming agents (17). The question relating to the cause of the HlyA-induced periodic Ca2⫹ oscillations also appeared to be resolved in that publication: Ca2⫹ oscillations were reportedly suppressed in the presence of the Ca2⫹ channel blocker nifedipine. Thus, in addition to being a pore-forming toxin, HlyA appeared to be endowed with the capacity of influencing physiological Ca2⫹ channels in mammalian cells (17). The present investigation was undertaken to make a closer investigation of this putative bifunctionality of HlyA. The capacity of the toxin to induce Ca2⫹ oscillations in mammalian cells was confirmed. However, Ca2⫹ oscillations showed no constant periodicity at all. Furthermore, a different conclusion was reached regarding the cause of Ca2⫹ oscillations. Evidence is presented that they are not due to toxin-dependent alterations of intrinsic Ca2⫹ channel activity but to pulses of Ca2⫹ flux through short-lived toxin pores, which are rapidly closed or removed from the plasma membrane of target cells. MATERIALS AND METHODS Cell culture starting with the mutant K564R/K690R to form S177C/ K564R/K690R. Mutant toxins containing Cys177 (S177C and S177C/K564R/K690R) were labeled with [3H]N-ethyl-maleimide. The binding capacity of these labeled toxins to erythrocytes was determined. The bindability of labeled mutant S177C and the hemolytically inactive mutant S177C/K564R/ K690R was identical. Protein purification of toxins was carried out as described previously (1,4). Electrophysiological recording Cells were plated in 35 mm dishes 48 h before the experiments. Cells (⬃4⫻105 per dish) were washed with extracellular (bath) solution (in mM: 140 NaCl, 3 KCl, 2 MgCl2, 2 CaCl2, 10 glucose, and 10 HEPES, pH 7.35 adjusted with ⬃4 mM NaOH), and whole-cell recordings were started within 5–10 min after the washing procedure, using a bath volume of 2 ml. Recording pipettes had a resistance of 5–10 M⍀ when filled with pipette solution (in mM: 140 K⫹ glutamate, 10 NaCl, 2 MgCl2, and 10 HEPES, pH 7.3 adjusted with ⬃4 mM KOH). A free intracellular Ca2⫹ concentration ([Ca2⫹]i) of 100 nm was obtained using 100 M of the Ca2⫹ chelator BAPTA and a total Ca2⫹ concentration of 30 M, assuming an apparent dissociation constant Kd of 0.24 M (pH 7.3) for the Ca2⫹ BAPTA complex. Solutions were filtered through 0.2 m pore size filters. Whole-cell current recording and data analysis were performed as described previously (18,19). HlyA was applied with a pipette directly into the bath solution within 2–5 min after a stable recording configuration had been obtained. All measurements were performed at 20 – 22°C. Data are mean ⫾ sem unless stated otherwise. Measurement of intracellular ATP Human embryonic kidney (HEK293) cells were maintained in a humidified, 6% CO2-94% air atmosphere in a mixture of Dulbecco’s modified Eagle’s medium (DMEM) and Ham’s F-12 medium (1:1, v/v) containing 10% (v/v) fetal calf serum (PAN Systems, Aidenbach, Germany) and 2 mM l-glutamine without antibiotics. The human endothelial cell line EAhy926 (a kind gift of Cora-Jean S. Edgell, University of North Carolina, Chapel Hill, NC), a hybridoma derived from the fusion of human umbilical vein endothelial cells with the A549 cell line, was grown in DMEM/Nutrient Mix F-12 (1:1) with Glutamax I and pyrodoxin (Life Technologies, Paisley, Scotland) supplemented with 10% fetal calf serum (Life Technologies), 100 U/ml penicillin, and 100 g/ml streptomycin (Life Technologies) at 37°C in humidified air containing 6% CO2. Human renal proximal tubule-derived cells (IHKE) were kindly provided by Dr. Michael Gekle (Institute of Physiology, University of Wu¨rzburg, Wu¨rzburg, Germany). The cells were originally from Dr. Steen Mollerup (Department of Toxicology, National Institute of Occupational Health, Oslo, Norway) who kindly permitted us to use them. Cells were grown in Ham’s F-12 modified DMEM medium supplemented with 1.18 g/l NaHCO3, 5.37 g/l HEPES, 5 mg/l human apo-transferrin, 5 mg/l bovine insulin, 36 g/l hydrocortisone, 10 g/l mouse epidermal growth factor, 5 g/l Na-selenite, and 10% fetal calf serum. Cells were grown in a 37°C, 94% air-6% CO2 incubator. Expression of HlyA, mutagenesis, and toxin purification Hemolytically inactive mutant in which Lys at position 564 and 690 was replaced by Arg (K564R/K690R) (1) was kindly provided by C. Hughes and V. Koronakis (Department of Pathology, Cambridge University, Cambridge, UK). Mutation of Ser-177 into Cys in cysteine-less wild-type (WT) HlyA was described previously (4). The same procedure was followed WHY HlyA INDUCES CALCIUM OSCILLATIONS HEK293 cells in 96-well plates were treated with various concentrations of HlyA for 60 min, after which the supernatants were removed. Cellular ATP concentrations were determined after the cells were lysed with 0.2 ml of 0.5% (v/v) Triton X-100, using a commercial ATP-bioluminescence kit (Boehringer Mannheim). Intracellular ATP was expressed as the percentage of luminescence relative to that of negative controls (without toxin). Ratiometric Ca2ⴙ imaging Cells were plated on 35 mm plastic petri dishes that had been modified for fluorescence measurements (18). Cells were loaded for 30 min at 37°C with the fluorescent Ca2⫹ indicator Fura 2-acetoxymethyl ester (AM) (Calbiochem-Novabiochem, Bad-Soden, Germany) dissolved in extracellular solution at a concentration of 5 M. The solution contained 16 M Pluronic F-127 (Molecular Probes, Eugene, OR) for a better dispersion of Fura 2-AM. Loaded cells were washed extensively with extracellular solution and further incubated for 10 min at 37°C. Measurements of [Ca2⫹]i were performed at room temperature with a Leica DM IRB inverse microscope and a VisiChrome imaging system (Visitron, Puchheim, Germany). Excitation was at 350 and 380 nm, and emission was determined at 510 nm. [Ca2⫹]i was determined ratiometrically from the captured data using the Metafluor imaging software. Biometric analysis A time period of 25.55 min (512 measurements, sampling interval 3 s) of each individual original Ca2⫹ tracing was subjected to a discrete Fourier transformation, followed by E81 computation of the respective power spectrum. According to the sampling period (T⫽25.55 min), multifolds of the basic angular velocity (0⫽2/T) constitute the Fourier transformation, corresponding to wavelengths of 25.55, 12.775, 8.517, 6.388, etc., min. In addition, the signals of these data sets were first corrected by possible baseline drifts using a trend function that consisted in a second order polynomial function, i.e., a constant, a linear, and a quadratic term. The trend function was estimated by a Gaussian least square method. After subtraction of the trend, the signals were treated as described above. No further data manipulations or standardization was carried out. trace. As a negative membrane holding potential of –50 mV was used, the membrane current trace steps down during a pore opening and up during a pore closing. Thus, Fig. 1A also demonstrates that the HlyA pores did not remain continuously open but oscillated between open and closed states (see also Figs. 4B and 5B and C). Despite ongoing pore formation, the cells remained viable when toxin concentrations of ⱕ10 ng/ml were employed, and no decreases in cellular ATP levels were noted (Fig. 1B). Online supplemental material HlyA induces nonperiodic Ca2ⴙ oscillations in renal cells Figure S1 is a time-lapse video of monitoring of [Ca2⫹]i in IHKE cells after exposure to HlyA (2.5 ng/ml). The original recording period was 90 min. Figure S2 shows Ca2⫹ tracings of 52 different IHKE cells after exposure to HlyA (2.5 ng/ml). Figure S3 shows a time-lapse video of [Ca2⫹]i in IHKE cells after exposure to HlyA (twice 2.5 ng/ml) in the presence of nifedipine. Ca2⫹ oscillations become visible after the change of the field of view. Figure S4 is a time-lapse video of the effect of HlyA (2.5 ng/ml) on [Ca2⫹]i in serum-starved (24 h) IHKE cells. Figure S5 shows Ca2⫹ tracings of 30 different serumstarved (24 h) IHKE cells exposed to HlyA (2.5 ng/ml). RESULTS Formation of HlyA pores in HEK293 cells HEK293 cells were initially used to characterize the pore-forming effects of HlyA in the whole-cell configuration of the patch-clamp technique. HEK293 cells possess only low endogenous ion channel activity at certain membrane potentials (20) and are well suited for electrophysiological experiments to measure the effect of pore-forming toxins. This cell type already has been successfully used in our group to study pore formation by listeriolysin of Listeria monocytogenes (18). Pore formation was observed within seconds after application of HlyA and followed a steep concentrationresponse curve (Fig. 1A). At a concentration of 5 ng/ml HlyA, ⬃230 s elapsed from the opening of the first pore to the opening of the tenth pore. A 10-fold higher concentration reduced this time span to ⬃40 s. With the patch-clamp-technique, a determination of single pores is possible, since each pore opening becomes visible by a stepwise change in the membrane current IHKE cells were loaded with Fura 2-AM and used to study the effect of HlyA on [Ca2⫹]i. IHKE cells were chosen as human equivalent to primary rat proximal tubule (RPT) cells that were used in a previous study to demonstrate the effect of HlyA on [Ca2⫹]i (17). IHKE cells are easily accessible for electrophysiological and Ca2⫹ measurements. Within ⬃30 s after application of 5 ng/ml HlyA to the cells, [Ca2⫹]i increased and exhibited oscillations during further monitoring (Fig. 2A-E). It was previously reported that HlyA provoked periodic Ca2⫹ oscillations in RPT cells, with a constant periodicity of Ca2⫹ peaks of 12 min (17). However, we found large variations of oscillation kinetics and peakto-peak spans in human tubule cells. Figure 2A-D shows Ca2⫹ tracings of four individual cells after application of HlyA, already giving an impression of the broad variability of the Ca2⫹ oscillations. False-color images of [Ca2⫹]i in individual cells are demonstrated in Fig. 2E. A time-lapse video of Ca2⫹ monitoring of IHKE cells is available online (Fig. S1), as well as Ca2⫹ tracings of ⬃50 different IHKE cells (Fig. S2), demonstrating impressively that the Ca2⫹ oscillations showed no constant periodicities of the peaks and large variations of the peak-to-peak spans. We then applied the identical mathematical method that was previously used to identify HlyA-induced periodic Ca2⫹ oscillations of low frequency (17), i.e., a power spectrum analysis that is based on a discrete Fourier transformation. Despite the marked heterogeneity of the original Ca2⫹ tracings, the analysis of the data sets yielded a periodicity of Ca2⫹ peaks with a dominant wavelength of 12.8 min, which was in remarkable agreement with the previously reported value of 12.0 min in RPT cells (17). As a control, Figure 1. A) Pore-forming activity of HlyA in HEK293 cells. Relationship of toxin concentration and time between opening of the first and the tenth pore (⌬tp1–10). Insets show courses of pore formation in single cells after exposure to (1) 50 ng/ml and (2) 6.25 ng/ml HlyA. Trace goes downward after a pore opening and upward during a pore closing. Membrane holding potential was –50 mV. This negative potential avoids activation of endogenous Ca2⫹ and K⫹ channels in HEK293 cells. B) Effect of HlyA on intracellular ATP concentration of HEK293 cells. Cells were exposed to HlyA at given concentrations for 60 min, 37°C, and cellular ATP was determined. Error bars indicate sd. E82 Vol. 20 May 2006 The FASEB Journal KOSCHINSKI ET AL. Figure 2. Ca2⫹ oscillations induced by HlyA in IHKE cells loaded with the fluorescent Ca2⫹ indicator Fura 2-AM. A–D) Cells were exposed to HlyA (5 ng/ml) 3 min after start of recording. Traces represent examples of time courses of [Ca2⫹]i of 4 single cells. E) Single frames from an image series of [Ca2⫹]i. Frames 1– 4 correspond to numbers 1– 4 above trace in A. Traces in A and B were recorded from cells that are marked with A and B. F) Superposition of Ca2⫹ responses of 49 IHKE cells in the presence of nifedipine (100 M). Application of HlyA (5 ng/ml) is indicated by arrow. Gap represents time when field of view was changed to an area that had not been UV irradiated before. G) Ca2⫹ tracing of a single IHKE cell on consecutive exposure to a nonhemolytic HlyA mutant (⫽nh HlyA, 25 ng/ml) and WT HlyA (2.5 ng/ml). tracings monitored in the absence of HlyA were also analyzed. Surprisingly, a periodicity of peaks with a wavelength of ⬃12 min was again computed, although no Ca2⫹ elevations existed in the control tracings. Furthermore, analysis of a signal consisting only of white noise superimposed to a discretely drifting baseline yielded the same peak value. Thus, it became clear that this mathematical model was not adequate for a proper interpretation of the measured Ca2⫹ data. Indeed, due to the properties of discrete Fourier transformation, the power spectrum will provide information for discrete frequencies only, depending on the sampling period and sampling frequency. In the experimental constellation we used, which is comparable to the constellation described in the previous report (17), the power spectrum is computed for wavelengths of 25.5, 12.8, 8.5, etc., min. The discrete Fourier transformation power spectrum analysis is not suitable to trace oscillations for these low frequencies, since variations in baseline do produce prominent peaks at the same discrete frequencies, which can lead to a fundamental misinterpretation of measured data. We next determined the statistical distribution of the time spans between directly neighboring Ca2⫹ peaks in Fig. 2A-D and 48 other Ca2⫹ tracings. Figure 3 shows that the most frequent peak-to-peak span was 70 s and that spans in the range of 1–3 min occurred more frequently compared to longer time spans, whereas WHY HlyA INDUCES CALCIUM OSCILLATIONS peak-to-peak spans above 10 min were very rare events. Importantly, these peak-to-peak spans represent the time span between a Ca2 ⫹peak to its direct neighbor. They should not be confused with peak periodicity or frequency of consecutive peaks. When the toxin concentration was increased, the peak-to-peak span distribution shifted to shorter time spans, as illustrated in Fig. 3, inset, by a typical Ca2⫹ tracing where two different HlyA concentrations were applied consecu- Figure 3. Frequency distribution of time spans between consecutive Ca2⫹ peaks in Ca2⫹ tracings monitored in 52 IHKE cells. Cells were loaded with Fura 2-AM and exposed to HlyA (5 ng/ml). Analysis period was 55 min. Curve was fitted to data using weighted average of 9 nearest neighbors. Bin width was 10 s. Inset shows Ca2⫹ monitoring of a single IHKE cell in the presence of 2.5 and 7.5 ng/ml HlyA. E83 tively. The finding that the peak-to-peak span distribution is dependent on HlyA concentration clearly refutes the idea that the HlyA molecule might have the inherent feature to induce Ca2⫹ oscillations of a discrete frequency. To address the question of a possible cell-type limitation of our observations, we further investigated the Ca2⫹ response to HlyA in HEK293 and EAhy926 cells. The EAhy926 line is representative of human vascular endothelial cells and was used to have a complement to IHKE cells, which have an epithelial nature. In HEK293 and EAhy926 cells, Ca2⫹ oscillations were also observed but with no evidence of a constant periodicity (data not shown). The results presented yet were obtained with nonsynchronized cells. Thus, one might speculate that the observed nonperiodic Ca2⫹ oscillations could be cellcycle dependent. To address this question, IHKE cells were subjected to 12–24 h serum starvation, and then Ca2⫹ measurements were performed. Despite synchronization of the cell cycle, HlyA (2.5 ng/ml)-induced Ca2⫹ oscillations were still nonperiodic, and no synchronization of Ca2⫹ elevations was observable. A timelapse video of Ca2⫹ monitoring of serum-starved IHKE cells is available online (Fig. S4), as well as 30 individual Ca2⫹ tracings of serum-starved IHKE cells (Fig. S5). These data clearly show that also in synchronized cells, the Ca2⫹ response to HlyA is nonperiodic and not synchronous. Nifedipine does not inhibit Ca2ⴙ oscillations When voltage-gated L-type Ca2⫹ channels in IHKE cells were blocked by nifedipine (100 M), Ca2⫹ oscillations were still observable after application of HlyA (Fig. 2F, Fig. S3). This finding contrasted with the observations in RPT cells where nifedipine (100 M) abolished the Ca2⫹ response to HlyA (17). A striking phenomenon that we observed during Ca2⫹ measurements in the presence of nifedipine can account for this difference. Those cells that were in the microscopic field of view for several minutes, i.e., irradiated by ultraviolet (UV) light, did indeed not show Ca2⫹ oscillations after application of HlyA (Fig. 2F, Fig. S3). Once adjusted, the field of view is usually not changed during the further Ca2⫹-monitoring period. However, when, after application of HlyA, the field of view was changed to cells that had not been in the optical path before, i.e., had not “seen” UV light, Ca2⫹ oscillations were observed (Fig. 2F, Fig. S3). This phenomenon was reproduced in three independent experiments in IHKE cells and also in HEK293 and EAhy926 cells. Thus, the previously observed elimination of HlyA-induced Ca2⫹ oscillations by nifedipine was essentially an experimental artifact. Additionally, HlyA still led to Ca2⫹ oscillations in HEK293 cells when a combination of Ca2⫹ channel blockers was present, including nifedipine (100 M); Cd2⫹ (2 mM), which blocks all types of voltage-gated Ca2⫹ channels (21); and SK&F 96365 (25 M), a E84 Vol. 20 May 2006 blocker of receptor-operated Ca2⫹ channels (22). This shows that the effect of HlyA on [Ca2⫹]i is not dependent on the activity of endogenous membrane Ca2⫹ channels. Pore formation and Ca2ⴙ oscillations do not occur in cells treated with nonhemolytic HlyA IHKE and HEK293 cells were treated with the HlyA mutant S177C/K564R/K690R, which is nonhemolytic but fully bindable. In no case were pore formation and Ca2⫹ oscillations ever noted at concentrations of up to 500 ng/ml. An example of the course of [Ca2⫹]i in IHKE cells with consecutive exposure to nonhemolytic HlyA mutant and to WT HlyA is shown in Fig. 2G. These results demonstrated that the ability of HlyA to provoke Ca2⫹ oscillations was not related simply to binding to the plasma membrane but required formation of pores. Pore formation and Ca2ⴙ oscillations cease rapidly on removal of HlyA We then addressed the question of whether the effect of HlyA on [Ca2⫹]i persists after removal of the toxin from the extracellular medium. Figure 4A shows a superposition of the Ca2⫹ responses of 31 IHKE cells after application of HlyA (10 ng/ml). Shortly after the start of a bath perfusion to wash out HlyA from the extracellular medium, [Ca2⫹]i decreased and reached the initial concentration after ⬃3 min. A second application of HlyA (5 ng/ml) again led to Ca2⫹ oscillations, which disappeared within some minutes after restart of the bath perfusion. The disappearance of the Ca2⫹ oscillations after washout of HlyA followed the same time course as observed for the disappearance of pore formation. Figure 4B shows a typical whole-cell registration of pore formation in a single IHKE cell after application of HlyA. Open pores were no longer observed ⬃4 min after the start of the bath perfusion. Reapplication of HlyA again led to pore formation. The most straightforward explanation for the above findings was that after formation, the HlyA pores rapidly disappeared from the plasma membrane, either through closure or by removal from the afflicted membrane area. To corroborate this, influx of propidium iodide was investigated in parallel. As shown in Fig. 4C-E, cells became permeable to the dye during incubation with HlyA, but membranes returned to the impermeable state shortly after toxin removal. This confirmed that pores formed by HlyA were short lived. Evidence that Ca2ⴙ oscillations result from Ca2ⴙ influx through HlyA pores followed by Ca2ⴙ redistribution We then simultaneously monitored pore formation by HlyA (10 ng/ml) and [Ca2⫹]i in IHKE cells. We found an increase in [Ca2⫹]i directly after pore opening and a good temporal correlation between formation and closure of HlyA pores and Ca2⫹ elevations (Fig. 5A, B). The FASEB Journal KOSCHINSKI ET AL. Figure 4. A) Superposition of the Ca2⫹ responses of 31 IHKE cells after exposure to HlyA (10 ng/ml), followed by a washout, and a repeat of that sequence with 5 ng/ml HlyA. B) Monitoring of pore formation in a single HEK293 cell after application of HlyA (10 ng/ ml), followed by superfusion of the bath chamber with extracellular solution to wash out the toxin, and second application of HlyA (10 ng/ml). Insets 1 and 2 show pore formation at a higher time resolution, revealing that pores did not remain continuously open but oscillated between open and closed states. Membrane holding potential was –50 mV. Membrane current trace goes downward during a pore opening and upward during closing. C–E) Demonstration that IHKE cells are permeable for propidium iodide (PI) dependent on the presence of HlyA in the medium. C) Cells exposed to PI alone. D) PI was added to the cells together with 50 ng/ml HlyA. E) PI was added after HlyA removal. After 15 min incubation with PI, samples were subjected to flow cytometric analysis. Cells simultaneously exposed to HlyA and PI showed red fluorescence because of flux of PI into the cells. In contrast, cells exposed to PI after removal of HlyA were not stained. We hypothesized that the decrease after an elevation might reflect the restoration of Ca2⫹ homeostasis by physiological Ca2⫹ redistribution after HlyA pore closure. To test this possibility, IHKE cells were incubated with thapsigargin (1 M), an inhibitor of endoplasmic reticular Ca2⫹ ATPases. We found that the effect of thapsigargin depended on the concentration of HlyA. In the presence of thapsigargin and 2.5 ng/ml HlyA, most IHKE cells showed typical Ca2⫹ oscillations. In the presence of thapsigargin and 5 ng/ml HlyA, some IHKE cells showed typical Ca2⫹ oscillations even during several hours of monitoring. The major part of the cells exhibited Ca2⫹ oscillations only for a limited period of ⬃30 min, and thereafter a continuous increase in [Ca2⫹]i was observed. Some cells showed no Ca2⫹ oscillations but only a continuous elevation in [Ca2⫹]i. When thapsigargin-treated IHKE cells were exposed to 10 ng/ml HlyA, most cells responded with a continuous increase in [Ca2⫹]i. In IHKE cells that were not treated with thapsigargin, HlyA concentrations of at least 20 ng/ml were necessary to induce a continuous elevation in [Ca2⫹]i. DISCUSSION The results of this study led us to a new explanation for Ca2⫹ oscillations that occur in mammalian cells treated WHY HlyA INDUCES CALCIUM OSCILLATIONS with HlyA. Three human cell-types were used for this investigation, namely renal embryonic, renal tubule epithelial, and vascular endothelial cells. The Ca2⫹ oscillations are not thought to be due to any direct or indirect influence of HlyA on intrinsic Ca2⫹ channels. Instead, they are proposed to result from pulsed influxes of Ca2⫹ through short-lived toxin pores, which we believe are rapidly sealed or removed from the membrane. Given the fundamental relevance of cytoplasmic Ca2⫹ oscillations for many cell functions, this conceptual remodeling will be of importance for future investigations in the field. Arguments in support of the proposal are manifold. First, we were unable to confirm the suppressive effect of the Ca2⫹ channel blocker nifedipine on HlyAinduced Ca2⫹ oscillations in any of the three cell-types studied. It is known that nifedipine is very sensitive to UV and daylight (up to 450 nm). It is readily converted into degradation products, which are no longer active against Ca2⫹ channels but instead have cellular as well as toxic effects (23–25). The degradation products might lead to cell damage, with the consequence of missing Ca2⫹ oscillations in UV-irradiated cells. In conclusion, the previously observed elimination of HlyA-induced Ca2⫹ oscillations by nifedipine was essentially not due to Ca2⫹ channel blockade, and the sole E85 Figure 5. A, B) Parallel monitoring of [Ca2⫹]i and pore formation by HlyA (10 ng/ml) in a single IHKE cell. A) Ca2⫹ concentration trace. B) Electrophysiological recording of pore formation. Solid line in B is an inverse display of the Ca2⫹ concentration trace shown in A for a direct comparison of [Ca2⫹]i and pore formation. C) Part of the trace shown in B but at a higher time resolution for a better discrimination of pore openings and closings. Membrane holding potential was –50 mV. Membrane current trace goes downward during a pore opening and upward during closing. support for the postulated crucial involvement of intrinsic L-type Ca2⫹ channels in HlyA-provoked Ca2⫹ oscillations (17) ceased to exist. Furthermore, we found that the effect of HlyA on [Ca2⫹]i is not dependent on the activity of any voltage-gated or receptor-operated membrane Ca2⫹ channels. Secondly, washout experiments provided an important further clue. Removal of HlyA from the medium resulted in rapid disappearance of the pores. At the same time, the ability of propidium iodide to diffuse into the cells was lost and, strikingly, Ca2⫹ oscillations also ceased. The conclusion was rather inescapable that, after formation in mammalian cell membranes, HlyA pores are short lived and disappear within seconds, either by rapid closure or by removal from the membrane. Final support for the pore theory of Ca2⫹ oscillations came from simultaneous patch-clamp measurements of pore formation and Ca2⫹ measurements. These experiments directly showed that Ca2⫹ elevations occurred as a consequence of opening of the pores formed by HlyA. Pore openings occur at random, and the open probability depends on HlyA concentration. Since a rise in [Ca2⫹]i is an effect that is directly conditioned by a pore opening, it is also expected to happen stochastically, and the probability should be dependent on toxin concentration. In contrast to a previous report (17), we found indeed no evidence for the existence of a E86 Vol. 20 May 2006 constant periodicity of the Ca2⫹ peaks, and we observed that the time spans between consecutive Ca2⫹ peaks depended on HlyA concentration. This concentration dependence definitely rules out the idea that HlyA molecules may switch on in the target cell a periodic “Ca2⫹ oscillation run” with one defined frequency. The concept thus took shape that Ca2⫹ fluxed from the extracellular compartment through the stochastically opened HlyA pores to the cytosol and that this was followed by pore closure and a rapid decline of the elevated cytoplasmic Ca2⫹ concentration, indicating restoration of cellular Ca2⫹ homeostasis. This concept is supported by the observed dependence of the Ca2⫹ response on HlyA concentration in IHKE cells treated with thapsigargin. It appears that cells respond to Ca2⫹ influx through HlyA pores with uptake into intracellular Ca2⫹ stores combined with discharge by plasma membrane Ca2⫹ pumps. If a part of these systems is cut off, e.g., by thapsigargin, the amount of Ca2⫹ influx determines whether Ca2⫹ homeostasis can still be maintained over time or not. As the amount of Ca2⫹ influx depends on the number of HlyA pores, it can be well explained that the effect of thapsigargin is dependent on HlyA concentration. In summary, a novel concept has emerged to account for the intriguing finding that a pore-forming toxin can provoke Ca2⫹ oscillations in living cells. The experiments described herein provide evidence that an RTX toxin forms very short-lived pores in cell membranes. In fact, similar rapid pore closure has been noted in artificial lipid bilayers (7–9). Thus, the very short pore life may be an intrinsic property that is governed by the target bilayer itself. Nonperiodic (“chaotic”) Ca2⫹ oscillations resulting from the interplay of formation and disappearance of the pores along with cellular Ca2⫹ redistribution will obviously vary depending on toxin concentration and susceptibility and may provide the starting point of myriad reactions, but on a very individual basis, in cells attacked by pore-forming toxins such as HlyA. We thank C. Zibuschka and S. Weis for expert technical assistance. This work was supported by grants from the Deutsche Forschungsgemeinschaft (RE 1046/1–2 and SFB 490, project C1/D3). REFERENCES 1. Stanley, P., Koronakis, V., and Hughes, C. (1998) Acylation of Escherichia coli hemolysin: a unique protein lipidation mechanism underlying toxin function. Microbiol. Mol. Biol. Rev. 62, 309 –333 2. Welch, R.A. (2001) RTX toxin structure and function: a story of numerous anomalies and few analogies in toxin biology. In Pore-Forming Toxins (van der Goot, G.F., ed.) pp. 85–112, Springer Verlag, Berlin, Germany 3. Hyland, C., Vuillard, L., Hughes, C., and Koronakis, V. (2001) Membrane interaction of Escherichia coli hemolysin: flotation and insertion-dependent labelling by phospholipids vesicles. J. Bacteriol. 183, 5364 –5370 4. Schindel, C., Zitzer, A., Schulte, B., Gerhards, A., Stanley, P., Hughes, C., et al. (2001) Interaction of Escherichia coli hemolysin The FASEB Journal KOSCHINSKI ET AL. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. with biological membranes. A study using cysteine scanning mutagenesis. Eur. J. Biochem. 268, 800 – 808 Lally, E.T., Kieba, I.R., Sato, A., Green, C.L., Rosenbloom, J. Korostoff, J., Wang, J.F., Shenker, B.J., Ortlepp, Robinson, M. K., and Billings, P. C. (1997) RTX toxins recognize an 2 integrin on the surface of human target cells. J. Biol. Chem. 272, 30463–30469 Cortajarena, A.L., Goni, F. M., and Ostolaza, H. (2001) Glycophorin as a receptor for Escherichia coli alpha-hemolysin in erythrocytes. J. Biol. Chem. 276, 12513–12519 Benz, R., Schmid, A., Wagner, and Goebel, W. (1989) Pore formation by the Escherichia coli hemolysin: evidence for an association-dissociation equilibrium of the pore-forming aggregates. Infect. Immun. 57, 887– 895 Menestrina, G., Pederzolli, C., Dalla Serra, M., Bregante, J., and Gambale, F. (1996) Permeability increase induced by Escherichia coli hemolysin A in human macrophages is due to the formation of ionic pores: a patch clamp characterization. J. Membr. Biol. 149, 113–121 Menestrina, G., Mackman, N., Holland, I. B., and Bhakdi, S. (1987) Escherichia coli haemolysin forms voltage-dependent ion channels in lipid membranes. Biochim. Biophys. Acta 905, 109 – 117 Bhakdi, S., Walev, I., Jonas, D., Palmer, M., Weller, U., Suttorp, et al. (1996) Pathogenesis of sepsis syndrome: possible relevance of pore-forming bacterial toxins. Curr. Top. Microbiol. Immunol. 216, 101–118 Valeva, A., Walev, I., Gerber, A., Klein, J., Palmer, M., and Bhakdi, S. (2000) Staphylococcal alpha-toxin: repair of a calcium permeable pore in the target cell membrane. Mol. Microbiol. 36, 467– 476 Walev, I., Bhakdi, S. C., Hofmann, F., Djonder, N., Valeva, A., Aktories, K., and Bhakdi, S. (2001) Delivery of proteins into living cells by reversible membrane permeabilization with streptolysin-O. Proc. Natl. Acad. Sci. U. S. A. 98, 3185–3190 Dragneva, Y., Anuradha, C. D., Valeva, A., Hoffmann, A., Bhakdi, S., and Husmann, M. (2001) Subcytocidal attack by staphylococcal alpha-toxin activates Nf-kappaB and induces interleukin-8 production. Infect. Immun. 69, 2630 –2635 Walev, I., Hombach, M., Bobkiewicz, W., Fenske, D., Bhakdi, S., and Husmann, M. (2002) Resealing of large transmembrane pores produced by streptolysin O in nucleated cells is accompanied by NF-kappaB activation and downstream events. FASEB J. 216, 237–239 WHY HlyA INDUCES CALCIUM OSCILLATIONS 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. Berridge, J. J., Bootman, M. D., and Roderick, H. L. (2003) Calcium signalling: dynamics, homeostasis and remodelling. Nat. Rev. Mol. Cell. Biol. 4, 517–529 Lewis, R. S. (2003) Calcium oscillation in T-cells: mechanisms and consequence for gene expression. Biochem. Soc. Trans. 31, 925–929 Uhlen, P., Laestadius, A., Jahnukainen, T., Soderblom, T., Backhed, F., Ce, G., et al. (2000) Alpha-haemolysin of uropathogenic E. coli induces Ca2⫹ oscillations in renal epithelial cells. Nature 405, 694 – 697 Repp, H., Pamukc¸i, Z., Koschinski, A., Domann, E., Darji, A., Birringer, J., et al. (2002) Listeriolysin of Listeria monocytogenes forms Ca2⫹ permeable pores leading to intracellular Ca2⫹ oscillations. Cell Microbiol. 4, 483– 491 Koschinski, A., Wengler, G., Wengler, G., and Repp, H. (2003) The membrane proteins of flaviviruses form ion-permeable pores in the target membrane after fusion: Identification of the pores and analysis of their possible role in virus infection. J. Gen. Virol. 84, 1711–1721 Berjukow, S., Doring, F., Froschmayr, M., Grabner, M., Glossmann, H., and Hering, S. (1996) Endogenous calcium channels in human embryonic kidney (HEK293) cells. Br J. Pharmacol. 118, 748 –754 Randall, A. D. (1998) The molecular basis of voltage-gated Ca2⫹ channel diversity: is it time for T? J. Membr. Biol. 161, 207–213 Rink, T. J. (1990) Receptor-mediated calcium entry. FEBS Lett. 268, 381–385 De Vries, H., and Beijersbergen van Henegouwen, G. M. (1998) Photoreactivity of nifedipine in vitro and in vivo. J. Photochem. Photobiol. B 43, 217–221 Frick, M., Siber, G., Haller, T., Mair, N., and Dietl, P. (2001) Inhibition of ATP-induced surfactant exocytosis by dihydropyridine (DHP) derivatives: a non-stereospecific, photoactivated effect and independent of L-type Ca2⫹ channels. Biochem. Pharmacol. 61, 1161–1167 Savigni, D.L., Wege, D., Cliff, G.S., Meesters, M. L., and Morgan, E. H. (2003) Iron and transition metal transport into erythrocytes mediated by nifedipine degradation products and related compounds. Biochem. Pharmacol. 65, 1215–1226 Received for publication August 18, 2005. Accepted for publication January 3, 2006. E87
© Copyright 2024