Amyloid Formation Modulates the Biological Activity of a Bacterial Protein

The aggregation of proteins into amyloid fibrils is the hallmark feature of a group of late-onset degenerative diseases including Alzheimer, Parkinson, and prion diseases. We report here that microcin E492, a peptide naturally produced by Klebsiella pneumoniae that kills bacteria by forming pores in the cytoplasmic membrane, assembles in vitro into amyloid-like fibrils. The fibrils have the same structural, morphological, tinctorial, and biochemical properties as the aggregates observed in the disease conditions. In addition, we found that amyloid formation also occurs in vivo where it is associated with a loss of toxicity of the protein. The finding that microcin E492 naturally exists both as functional toxic pores and as harmless fibrils suggests that protein aggregation into amyloid fibrils is an evolutionarily conserved property of proteins that can be successfully employed by bacteria to fulfill specific physiological needs. The aggregation of proteins into amyloid fibrils is the hallmark feature of a group of late-onset degenerative diseases including Alzheimer, Parkinson, and prion diseases. We report here that microcin E492, a peptide naturally produced by Klebsiella pneumoniae that kills bacteria by forming pores in the cytoplasmic membrane, assembles in vitro into amyloid-like fibrils. The fibrils have the same structural, morphological, tinctorial, and biochemical properties as the aggregates observed in the disease conditions. In addition, we found that amyloid formation also occurs in vivo where it is associated with a loss of toxicity of the protein. The finding that microcin E492 naturally exists both as functional toxic pores and as harmless fibrils suggests that protein aggregation into amyloid fibrils is an evolutionarily conserved property of proteins that can be successfully employed by bacteria to fulfill specific physiological needs. Bacteriocins form a vast and diverse group of antibacterial proteins that have been found in most lineages of Eubacteria and Archaebacteria. They kill or inhibit the growth of closely related species by binding to specific cell surface receptors and by forming ion channels in the cytoplasmic membrane, degrading DNA, blocking protein translation, or inhibiting peptidoglycan synthesis (1Riley M.A. Annu. Rev. Genet. 1998; 32: 255-278Crossref PubMed Scopus (199) Google Scholar). They have been proposed to be involved in maintaining bacterial diversity by mediating interactions at the population and community levels (2Riley M.A. Goldstone C.M. Wertz J.E. Gordon D. J. Evol. Biol. 2003; 16: 690-697Crossref PubMed Scopus (63) Google Scholar). Microcin E492 (Mcc) 1The abbreviations used are: Mcc, microcin E492; LB, Luria-Bertani; PIPES, 1,4-piperazinediethanesulfonic acid; CR, Congo red; ThT, thioflavin T; PK, proteinase K. is a low molecular weight bacteriocin naturally produced by Klebsiella pneumoniae RYC492 that is active against several strains of Enterobacteriaceae (3de Lorenzo V. Martinez J.L. Asensio C. J. Gen. Microbiol. 1984; 130: 391-400PubMed Google Scholar, 4de Lorenzo V. Arch. Microbiol. 1984; 139: 72-75Crossref PubMed Scopus (76) Google Scholar, 5Destoumieux-Garzon D. Thomas X. Santamaria M. Goulard C. Barthelemy M. Boscher B. Bessin Y. Molle G. Pons A.M. Letellier L. Peduzzi J. Rebuffat S. Mol. Microbiol. 2003; 49: 1031-1041Crossref PubMed Scopus (66) Google Scholar). Its mechanism of action has been shown to involve the formation of pores in the cytoplasmic membrane of target cells and subsequent lethal loss of membrane potential (5Destoumieux-Garzon D. Thomas X. Santamaria M. Goulard C. Barthelemy M. Boscher B. Bessin Y. Molle G. Pons A.M. Letellier L. Peduzzi J. Rebuffat S. Mol. Microbiol. 2003; 49: 1031-1041Crossref PubMed Scopus (66) Google Scholar, 6de Lorenzo V. Pugsley A.P. Antimicrob. Agents Chemother. 1985; 27: 666-669Crossref PubMed Scopus (66) Google Scholar, 7Lagos R. Wilkens M. Vergara C. Cecchi X. Monasterio O. FEBS Lett. 1993; 321: 145-148Crossref PubMed Scopus (61) Google Scholar). Whereas most known microcins are only produced when exponential growth ceases and the cells enter the stationary phase (8Baquero F. Moreno F. FEMS Microbiol. Lett. 1984; 23: 117-124Crossref Scopus (160) Google Scholar), Mcc activity has been reported to be highest during the exponential phase and to suddenly drop to virtually undetectable levels during the stationary phase (3de Lorenzo V. Martinez J.L. Asensio C. J. Gen. Microbiol. 1984; 130: 391-400PubMed Google Scholar, 9de Lorenzo V. J. Antibiot. 1985; 38: 340-345Crossref PubMed Scopus (16) Google Scholar). Surprisingly, Mcc was found to be similarly produced and accumulated both in exponential and stationary phases (10Corsini G. Baeza M. Monasterio O. Lagos R. Biochimie (Paris). 2002; 84: 539-544Crossref PubMed Scopus (24) Google Scholar), thereby ruling out an arrest of Mcc production and/or degradation as causes of this loss of activity. In addition, mass spectrometry studies revealed no difference in the molecular weight of Mcc purified from either growth phases, indicating that this effect was not related to any post-translational modifications of the protein under our experimental conditions. 2S. Bieler, R. Lagos, and C. Soto, unpublished data. This led us to hypothesize that the sudden disappearance of activity in the stationary phase could be because of a change in Mcc conformation and/or oligomerization state. We report here that purified active Mcc is able to be converted upon in vitro incubation from a soluble to an insoluble state consisting of amyloid-like fibrils. Electron microscopy, structural, biochemical, and kinetics studies indicate that these fibrils have properties identical to those previously described for the amyloid aggregates that are formed in the course of amyloid diseases. We also found that the assembly of Mcc into amyloid fibrils is accompanied by a loss of the antibacterial activity of the protein. Furthermore, amyloid fibrils are also formed in vivo, and this correlates with a sudden drop in Mcc activity in the culture. We therefore propose that amyloid formation modulates the activity of Mcc in vivo and is responsible for the previously reported loss of activity in the stationary phase. These data represent the first demonstration that amyloid formation in prokaryotes controls in vivo the biological activity of a protein and changes the phenotype of the bacteria. Thus far, formation of amyloid has been associated with several human diseases involving the misfolding, aggregation, and tissue deposition of proteins (11Soto C. FEBS Lett. 2001; 498: 204-207Crossref PubMed Scopus (290) Google Scholar, 12Dobson C.M. Trends Biochem. Sci. 1999; 24: 329-332Abstract Full Text Full Text PDF PubMed Scopus (1708) Google Scholar, 13Carrell R.W. Lomas D.A. Lancet. 1997; 350: 134-138Abstract Full Text Full Text PDF PubMed Scopus (820) Google Scholar). Our findings suggest that the process of protein aggregation into amyloid fibrils is a conserved mechanism that may have an important role not only in triggering tissue damage and disease but also in modulating the biological activity of diverse proteins. Bacterial Strains and Plasmids—The Escherichia coli strains and plasmids used in this study are described in Table I.Table IBacterial strains and plasmidsStrain/plasmidGenotypeSource or referenceE. coli strainsVCS257DP50 supF [supE44 supF58 hsdS3(rB-mB-) dapD8 lacY1 glnV44 Δ(gal-uvrB)47 tyrT58 gyrA29 tonA53 Δ(thyA57)]StratageneBL21F- ompT hsdS(rB-, mB-) gal dcmAmersham BiosciencesBL21(DE3)BL21 λ(DE3)NovagenPlasmidspJEM15Mcc determinants28Wilkens M. Villanueva J.E. Cofre J. Chnaiderman J. Lagos R. J. Bacteriol. 1997; 179: 4789-4794Crossref PubMed Google Scholarnp108Mcc determinants mceG::Tn527Lagos R. Baeza M. Corsini G. Hetz C. Strahsburger E. Castillo J.A. Vergara C. Monasterio O. Mol. Microbiol. 2001; 42: 229-243Crossref PubMed Scopus (66) Google Scholarp11α2Ampr, fragment of tubulin cloned in pT7-728Wilkens M. Villanueva J.E. Cofre J. Chnaiderman J. Lagos R. J. Bacteriol. 1997; 179: 4789-4794Crossref PubMed Google Scholar Open table in a new tab Growth Conditions—Bacteria were grown at 37 °C with shaking in LB (Luria-Bertani) Miller medium (14Miller J.H. Experiments in Molecular Genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York1972: 431-435Google Scholar) that was filtered through a 0.22-μm cellulose acetate filter system (Corning). When specified, the medium was buffered with 100 mm PIPES-NaOH at pH 6.5. Antibiotics were used at the following concentrations: ampicillin, 100 μg/ml; kanamycin, 50 μg/ml. Cultures were started using 1:1000 dilutions of fresh overnight cultures to minimize the amount of amyloid-like Mcc from the inocula. Purification of Mcc—E. coli VCS257pJEM15 was grown at 37 °C with shaking to an absorbance at 600 nm of 1.5 in M9 minimal medium containing 0.2% glucose, 0.2% sodium citrate, 1 g/liter casamino acids, 1 mg/liter thiamine, and 100 mg/liter ampicillin. 200 ml of culture supernatant were passed through a Sep-Pak C18 cartridge (Waters). The cartridge was washed with 3 ml of 65% methanol and 3 ml of 25% acetonitrile. Mcc was eluted with 3 ml of 50% acetonitrile and concentrated by a CentriVap Concentrator (Labconco). Samples were brought to pH 8.0 by adding NaOH and incubated at 37 °C for 15 min to ensure full solubilization of aggregates formed during the concentration step. Finally, samples were centrifuged at 75,000 × g for 1 h at 4 °C to remove any remaining aggregates, and supernatants were stored at –20 °C. Purity was calculated to be >90% by amino acid composition analysis (420/H Amino Acid Analysis System; Applied Biosystems). Matrix-assisted laser desorption ionization time-of-flight mass spectrometry performed in a Voyager DE STR biospectrometry work station (Applied Biosystems) revealed one single peak at 7887.5 Da corresponding to the previously reported molecular mass of mature Mcc without post-translational modifications (5Destoumieux-Garzon D. Thomas X. Santamaria M. Goulard C. Barthelemy M. Boscher B. Bessin Y. Molle G. Pons A.M. Letellier L. Peduzzi J. Rebuffat S. Mol. Microbiol. 2003; 49: 1031-1041Crossref PubMed Scopus (66) Google Scholar, 15Pons A.M. Zorn N. Vignon D. Delalande F. Van Dorsselaer A. Cottenceau G. Antimicrob. Agents Chemother. 2002; 46: 229-230Crossref PubMed Scopus (29) Google Scholar). Electron Microscopy—Samples of purified Mcc or bacterial culture were placed onto 300-square mesh nickel grids (Maxtaform), negatively stained with 2% uranyl acetate for 90 s, and examined in a Philips EM410 electron microscope. Circular Dichroism Analysis—CD spectra were obtained using an AVIV CD spectrometer model 215 with a 0.1-cm path length cell. Five consecutive readings at 1-nm bandwidth, 0.25-s response time, and 1-nm resolution were taken over the wavelength range 190–260 nm at 25 °C, averaged, buffer subtracted, and noise reduced. Mean residue molar ellipticities of Mcc were calculated using 7886.5 as molecular weight and 84 as the number of residues. Samples were sonicated prior to analysis to ensure homogeneity. Congo Red Binding—For spectral analysis, aggregated Mcc was obtained by incubating the protein at 750 μg/ml in aggregation buffer (10 mm PIPES-NaOH, pH 6.5, 0.5 m NaCl) at 37 °C with agitation for 15 h. Samples were briefly sonicated and incubated with 12.5 μm Congo red (CR) (Aldrich) for 15 min at 37 °C. Absorbance was measured from 400 to 700 nm with a resolution of 1 nm using a DU 530 spectrophotometer (Beckman Coulter). The spectrum of bound CR was obtained by subtracting the spectrum of Mcc alone from the spectrum of Mcc in the presence of CR. The spectrum of free CR was obtained by subtracting the spectrum of the buffer from the spectrum of CR in buffer in the absence of Mcc. To quantitate amyloid formation by CR assay, Mcc samples (200 μg/ml) were incubated with 44 μm CR for 15 min at 37 °C and centrifuged at 16,000 × g for 5 min at room temperature. The absorbance of supernatants was recorded at 490 nm using an EL800 Universal Microplate Reader (Bio-Tek Instruments). The percentage of free CR was obtained by comparison with the absorbance of a sample devoid of protein. Thioflavin T Fluorescence—For spectral analysis, aggregated Mcc was briefly sonicated, diluted in the aggregation buffer to a final concentration of 3 μm and added 5 μm thioflavin T (ThT) (Aldrich). After 10 min of incubation at room temperature, fluorescence measurements were carried out on a Fluorolog 2 spectrofluorimeter (Spex Industries). The excitation wavelength was set at 430 nm, emission wavelengths ranged from 440 to 700 nm, and slit widths were 6 mm for excitation and emission. To quantitate amyloid formation by ThT assay, fluorescence emission was measured in the presence of ThT at 480 nm. The excitation wavelength was set at 430 nm, and slit widths were 6 mm for excitation and emission. Fluorescence emission is expressed as arbitrary units. Proteinase K Digestion—Samples were incubated for 20 min at 37 °C under agitation with the indicated concentrations of proteinase K (PK) (Roche Applied Science). The reaction was stopped by boiling the samples at 102 °C for 5 min in NuPAGE LDS sample buffer (Invitrogen). To assess the resistance of soluble and aggregated Mcc to increasing concentrations of PK, purified Mcc was diluted at 15 μg/ml in culture medium (LB medium, 100 mm PIPES-NaOH, pH 6.5) and either directly digested (soluble) or incubated for 15 h at 37 °C with agitation prior to digestion (aggregated). Turbidity Assay—Mcc was diluted in aggregation buffer at a final concentration of 850 μg/ml and incubated at 37 °C with agitation. The aggregation of the protein was monitored by measuring the absorbance at 562 nm using a BioPhotometer 6131 (Eppendorf). Sedimentation Assay and Seeding of Mcc Aggregation—Samples of Mcc at 150 μg/ml in aggregation buffer were incubated at 37 °C with agitation for the indicated times and centrifuged at 16,000 × g for 5 min at room temperature. Protein concentration in supernatants was measured using a Micro BCA protein assay kit (Pierce). The amount of aggregated protein was obtained by subtracting this value from the total concentration of Mcc. Aggregated Mcc used to seed aggregation was obtained by incubating the protein at 150 μg/ml in aggregation buffer at 37 °C with agitation for 15 h. Sonicated aggregated Mcc was obtained by submitting the aggregated protein to sonication three times for 3 s at 60% power using a Vibra-Cell ultrasonic processor (Sonics & Materials). Mcc Activity Assay by the Critical Dilution Method—Samples to be tested were serially diluted in 50% methanol. In the case of cultures, cells were previously removed by centrifugation at 16,000 × g for 5 min at room temperature. 4-μl drops of diluted samples were laid onto LB agar plates previously covered with 3 ml of LB top agar containing 2 × 107 Mcc-sensitive E. coli BL21(DE3)p11α2 cells. After 15 h of incubation at 37 °C the activity was determined by the critical dilution method and expressed in arbitrary units/ml (16Mayr-Harting A. Hedges A.J. Berkeley R.C.W. Norris J.R. Ribbons D.N. Methods in Microbiology. 7A. Academic Press, New York1972: 315-422Google Scholar). Mcc Activity Assay by Counting Surviving Cells—90 μl of samples to be tested were added to 10 μl of E. coli BL21(DE3)p11α2 cells grown overnight in LB medium. After 30 min of incubation at 37 °C with shaking, surviving cells were counted by plating serial dilutions onto LB agar. The activity was expressed as a percentage, taking as a reference (100%) the killing activity obtained using a Mcc sample that was not incubated. Immunoblotting—Samples were diluted in sample buffer (Invitrogen), boiled at 102 °C for 5 min, and loaded on NuPAGE 4–12% bis-Tris gels (Invitrogen). Proteins were transferred to Hybond-ECL nitrocellulose membranes (Amersham Biosciences), incubated with a rabbit antiserum raised against a C-terminal fragment of Mcc (SGSGYNSATSSSGSGS) (1:3000), washed, incubated with a horseradish peroxidase-linked anti-rabbit donkey antibody (1:5,000; Amersham Biosciences), and developed with ECL plus (Amersham Biosciences). Soluble Mcc Assembles into Amyloid-like Fibrils in Vitro— Negative stain electron microscopy of samples of soluble purified Mcc incubated in LB medium, 100 mm PIPES, pH 6.5 (culture medium) for 15 h at 37 °C revealed ∼10-nm-wide unbranched amyloid-like fibrils of varying lengths (Fig. 1, A and B), whereas no fibrils could be detected in samples of the purified protein that had not been incubated (Fig. 1C). Fibrils undistinguishable from those obtained in culture medium were observed upon incubation in 100 mm PIPES-NaOH, pH 6.5, 0.5 m NaCl (Fig. 1D). Circular dichroism studies indicated that whereas soluble Mcc adopts a random coil conformation in aqueous buffer (Fig. 2A, solid line) or a mainly α-helical one in methanol (Fig. 2B, solid line), aggregated Mcc is rich in β-sheet structures in both solvents (Fig. 2, A and B, dashed lines). These results are consistent with the stable cross-β-sheet structure that has been proposed for amyloid fibrils (17Sunde M. Blake C. Adv. Protein Chem. 1997; 50: 123-159Crossref PubMed Google Scholar). A similar structural conversion from a soluble α-helical/random coil conformation into β-sheet-rich aggregates is the hallmark and likely the triggering event in protein misfolding diseases (11Soto C. FEBS Lett. 2001; 498: 204-207Crossref PubMed Scopus (290) Google Scholar, 18Soto C. Nat. Rev. Neurosci. 2003; 4: 49-60Crossref PubMed Scopus (1121) Google Scholar). A typical feature of amyloid fibrils that is used to differentiate them from other types of aggregates is the specific binding and staining with dyes such as CR and ThT (19Glenner G.G. N. Engl. J. Med. 1980; 302: 1283-1292Crossref PubMed Scopus (1310) Google Scholar, 20Sipe J.D. Cohen A.S. J. Struct. Biol. 2000; 130: 88-98Crossref PubMed Scopus (856) Google Scholar). Like previously described amyloids (21Klunk W.E. Jacob R.F. Mason R.P. Methods Enzymol. 1999; 309: 285-305Crossref PubMed Scopus (311) Google Scholar), aggregated Mcc induced a spectral change in the absorbance of CR with a maximum difference between free CR and Mcc-bound CR at ∼540 nm (Fig. 2C). The addition of aggregated Mcc to a solution of ThT resulted in a fluorescence emission maximum at ∼480 nm (Fig. 2D), which is identical to the fluorescence induced by other amyloid proteins (22LeVine III, H. Methods Enzymol. 1999; 309: 274-284Crossref PubMed Scopus (1213) Google Scholar). Another typical feature of amyloid aggregates is their high resistance to proteolytic degradation, which is partly the reason for the difficulty in clearing them from the body in the disease condition (23Bolton D.C. McKinley M.P. Prusiner S.B. Science. 1982; 218: 1309-1311Crossref PubMed Scopus (1015) Google Scholar, 24Nordstedt C. Naslund J. Tjernberg L.O. Karlstrom A.R. Thyberg J. Terenius L. J. Biol. Chem. 1994; 269: 30773-30776Abstract Full Text PDF PubMed Google Scholar). Although soluble Mcc was totally digested by low concentrations of PK, aggregated Mcc exhibited strong resistance to this enzyme at various concentrations (Fig. 2E). CR binding, ThT fluorescence, and protein sedimentation were employed to monitor amyloid formation by Mcc over time as described under “Experimental Procedures.” The curves obtained (Fig. 3, A–C, solid line) are reminiscent of previously described amyloid formation kinetics (25Harper J.D. Lansbury Jr., P.T. Annu. Rev. Biochem. 1997; 66: 385-407Crossref PubMed Scopus (1420) Google Scholar). Indeed, a lag phase where virtually no amyloid-like structures could be detected was followed by a sudden and fast increase in CR binding, ThT fluorescence, or protein precipitation. Similar kinetics was observed when Mcc aggregation was followed by measuring the increase in turbidity of the solution (data not shown). The time course aggregation data suggest that Mcc, similarly to other amyloid proteins, assembles into fibrils following a typical nucleation-dependent polymerization process. This involves a slow nucleation phase in which the unassembled polypeptides undergo unfavorable association steps to form an ordered oligomeric nucleus that is required to initiate rapid polymerization (25Harper J.D. Lansbury Jr., P.T. Annu. Rev. Biochem. 1997; 66: 385-407Crossref PubMed Scopus (1420) Google Scholar). Such a model implies that the lag phase could be bypassed by adding preformed nuclei or seeds. To test this hypothesis, soluble Mcc was incubated in the presence of previously aggregated Mcc, and protein aggregation was followed by sedimentation. As expected, the lag phase was significantly shortened (Fig. 3C, dashed line) by comparison with the incubation of the soluble protein alone (Fig. 3C, solid line). Because these seeds consisted of mature fibrils that are presumably not as efficient seeds as small oligomeric nuclei, a similar experiment was performed using sonicated fibrils instead of intact ones. Addition of sonicated fibrils resulted in a complete elimination of the lag phase (Fig. 3C, dotted line), indicating that sonication was efficient in breaking the fibrils and multiplying the number of seeds available for polymerization, as we described previously for the conversion of the prion protein (26Saborio G.P. Permanne B. Soto C. Nature. 2001; 411: 810-813Crossref PubMed Scopus (1022) Google Scholar). In addition, this nucleation-dependent amyloid formation model implies that the concentration of soluble Mcc needs to be over a threshold level referred to as critical concentration to initiate polymerization. This critical concentration also corresponds to the concentration of protein that remains in solution when amyloid formation is complete and the system is at equilibrium (25Harper J.D. Lansbury Jr., P.T. Annu. Rev. Biochem. 1997; 66: 385-407Crossref PubMed Scopus (1420) Google Scholar). Fig. 3C shows that only ∼110 μg/ml Mcc can be aggregated from a total of 150 μg/ml incubated protein, which means that under these conditions the critical concentration is ∼40 μg/ml or 5 μm. After this critical concentration is reached, a certain lag time is needed for protein-protein interaction to occur, enabling the formation of a nucleus that further sustains protein polymerization. In Vitro Polymerization of Mcc into Amyloid Fibrils Is Accompanied by a Loss of Activity—The relation between Mcc activity and amyloid formation was investigated by following over time the ability of soluble or aggregated Mcc to kill sensitive cells. The activity of Mcc before and after 24 h of incubation was measured by the critical dilution method, which revealed a loss of more than 90% of the activity upon aggregation (Fig. 4A). In addition, the activity was monitored by counting the number of surviving cells after exposure to Mcc aliquots incubated for various times. The results show that the activity was high during a period of 10 h, which was followed by a sudden and fast loss of activity (Fig. 4B). The kinetics of the change in activity mirrored closely the kinetics of amyloid formation as shown in Fig. 3, suggesting that the loss of activity was related to protein aggregation. To rule out the possibility that the loss of cytotoxic activity was because of protein degradation, we compared the amount of protein before and after 24 h of incubation. As shown in Fig. 4C, lanes 11 and 12, no difference was observed in the total amount of Mcc before and after incubation. The evolution of the PK resistance of Mcc upon incubation was also assessed. Mcc remained fully sensitive to PK for almost 10 h and then became increasingly resistant to this enzyme (Fig. 4C), which correlates with the loss of activity (Fig. 4B). As shown in Fig. 2, formation of amyloid is paralleled by acquisition of protease resistance; therefore, we conclude from these experiments that Mcc loses its toxic activity upon fibrillar aggregation. The Loss of Mcc Activity in Vivo Coincides with Amyloid Formation—A culture of Mcc-producing bacteria (E. coli BL21pJEM15) was monitored by periodically measuring its optical density at 600 nm and the Mcc activity present in the medium (Fig. 5A). The activity was found to accumulate during growth to reach its highest level at the end of the exponential phase of growth (9 h) and then to dramatically drop (10 h) and remain at virtually undetectable levels during the stationary phase. This phenomenon is not restricted to our experimental conditions involving a recombinant E. coli strain. Indeed, an identical pattern of activity was previously reported in the original K. pneumoniae RYC492 strain, where both active and inactive forms of Mcc were observed in the exponential and stationary phases of growth, respectively (3de Lorenzo V. Martinez J.L. Asensio C. J. Gen. Microbiol. 1984; 130: 391-400PubMed Google Scholar, 9de Lorenzo V. J. Antibiot. 1985; 38: 340-345Crossref PubMed Scopus (16) Google Scholar). However, an explanation for this phenomenon had been missing ever since. To determine whether amyloid was related to this sudden loss of Mcc activity, the in vivo formation of amyloid was evaluated by PK digestion (Fig. 5B), sedimentation assay (Fig. 5C), and electron microscopy (Fig. 6). Fig. 5B shows an immunoblot of culture samples treated or not with PK. Although the total amount of Mcc was found to gradually increase during cell growth, almost no PK-resistant Mcc could be detected up to 9 h of culture. Strikingly, a strong PK resistance was observed after 10 h and was maintained in subsequent time points, which coincides with the reported loss of activity. In agreement with our in vitro data showing a negative correlation between Mcc activity and resistance to PK digestion (Fig. 4), these in vivo results suggest that the formation of amyloid also occurs in a bacterial culture and that it profoundly affects the activity of the protein. To further study the relationship between Mcc aggregation and activity, we performed sedimentation assays on culture samples to analyze the amount of soluble and aggregated protein at each time, as described under “Experimental Procedures.” The results show that although most Mcc appeared to be soluble up to 9 h, virtually no soluble protein could be detected at 10 h and later (Fig. 5C). These results are in agreement with what was observed in vitro, where Mcc was found to become rapidly insoluble after several hours of incubation (Fig. 3C). This finding is again in accordance with the proposition that amyloid formation is responsible for the reported loss of Mcc activity. Finally, to confirm that amyloid-like fibrils were indeed formed in vivo, samples of E. coli BL21pJEM15 culture grown for 15 h were analyzed by electron microscopy. Amyloid-like fibrils resembling those obtained in vitro were observed (Fig. 6, A and B), whereas no such structures could be detected in samples of E. coli BL21np108 (a mutant that does not export Mcc) (27Lagos R. Baeza M. Corsini G. Hetz C. Strahsburger E. Castillo J.A. Vergara C. Monasterio O. Mol. Microbiol. 2001; 42: 229-243Crossref PubMed Scopus (66) Google Scholar) culture grown for 15 h (Fig. 6, C and D). In addition, no fibrils could be observed in samples of E. coli BL21pJEM15 culture grown for 8 h only (data not shown), which is consistent with the finding that Mcc is soluble and sensitive to PK digestion at this time point (Fig. 5). The misfolding, aggregation, and tissue accumulation of proteins into fibrillar amyloid deposits is a hallmark process in a variety of neurological and systemic disorders, collectively called protein misfolding diseases (11Soto C. FEBS Lett. 2001; 498: 204-207Crossref PubMed Scopus (290) Google Scholar, 12Dobson C.M. Trends Biochem. Sci. 1999; 24: 329-332Abstract Full Text Full Text PDF PubMed Scopus (1708) Google Scholar, 13Carrell R.W. Lomas D.A. Lancet. 1997; 350: 134-138Abstract Full Text Full Text PDF PubMed Scopus (820) Google Scholar, 18Soto C. Nat. Rev. Neurosci. 2003; 4: 49-60Crossref PubMed Scopus (1121) Google Scholar). Compelling evidence indicates that the formation of β-sheet-rich oligomeric and fibrillar structures is the triggering factor in these diseases, which comprise more than 20 different disorders, including Alzheimer, Parkinson, and prion diseases as well as diabetes type 2. These findings have led to the notion that amyloid formation is an inherently negative process implicated in pathological conditions in complex organisms. Here we have shown that amyloid formation also occurs in prokaryotes and is involved in modulating the biological activity of Mcc. Strikingly, the biochemical, structural, morphological, and tinctorial properties of the Mcc aggregates are identical to those observed in the disease protein accumulations. In addition, the mechanism of amyloid formation following a seeding-nucleation polymerization process and the formation of cytotoxic oligomeric structures are also hallmark properties of protein aggregation in the disease conditions. Therefore, amyloidogenesis cannot be considered as a process associated exclusively with diseases in higher organisms but should rather be viewed as a conserved mechanism to control the biological function of diverse proteins. The mode of action of Mcc has been shown to involve the formation of ion-permeable pores in the cytoplasmic membrane of target bacteria. Indeed, purified Mcc was found to depolarize and permeabilize the E. coli cytoplasmic membrane (5Destoumieux-Garzon D. Thomas X. Santamaria M. Goulard C. Barthelemy M. Boscher B. Bessin Y. Molle G. Pons A.M. Letellier L. Peduzzi J. Rebuffat S. Mol. Microbiol. 2003; 49: 1031-1041Crossref PubMed Scopus (66) Google Scholar, 6de Lorenzo V. Pugsley A.P. Antimicrob. Agents Chemother. 1985; 27: 666-669Crossref PubMed Scopus (66) Google Scholar) and to form ion channels in phospholipids bilayers in vitro (7Lagos R. Wilkens M. Vergara C. Cecchi X. Monasterio O. FEBS Lett. 1993; 321: 145-148Crossref PubMed Scopus (61) Google Scholar). Interestingly, Mcc is not only active against bacteria but was also found to cause apoptosis and necrosis in some eukaryotic cell lines (29Hetz C. Bono M.R. Barros L.F. Lagos R. Proc. Natl. Acad. Sci. U. S. A. 2002; 99: 2696-2701Crossref PubMed Scopus (129) Google Scholar). The finding that Mcc is able to induce cell death and aggregate into amyloid fibrils is particularly intriguing and could offer a good model by which to study the mechanism of cytotoxicity of disease-causing misfolded protein aggregates. Strikingly, the emerging notion in protein misfolding diseases is that the toxic species might be oligomeric pore-like structures that may cause cell damage by altering ionic homeostasis (30Caughey B. Lansbury P.T. Annu. Rev. Neurosci. 2003; 26: 267-298Crossref PubMed Scopus (1458) Google Scholar). Indeed, ordered prefibrillar oligomeric pores have been proposed to be responsible for cell death in Alzheimer and Parkinson diseases (31Lashuel H.A. Petre B.M. Wall J. Simon M. Nowak R.J. Walz T. Lansbury P.T. J. Mol. Biol. 2002; 322: 1089-1102Crossref PubMed Scopus (692) Google Scholar, 32Lashuel H.A. Hartley D. Petre B.M. Walz T. Lansbury P.T. Nature. 2002; 418: 291Crossref PubMed Scopus (1136) Google Scholar). A similar phenomenon has been proposed to be involved in diabetes type 2, as protofibrillar islet amyloid polypeptide was found to permeabilize synthetic vesicles by a pore-like mechanism (33Anguiano M. Nowak R.J. Lansbury Jr., P.T. Biochemistry. 2002; 41: 11338-11343Crossref PubMed Scopus (255) Google Scholar). Interestingly, small annular Aβ and α-synuclein protofibrils were reported to resemble the cytolytic β-barrel pore-forming toxins from bacteria such as Clostridium perfringens (32Lashuel H.A. Hartley D. Petre B.M. Walz T. Lansbury P.T. Nature. 2002; 418: 291Crossref PubMed Scopus (1136) Google Scholar, 34Hotze E.M. Heuck A.P. Czajkowsky D.M. Shao Z.F. Johnson A.E. Tweten R.K. J. Biol. Chem. 2002; 277: 11597-11605Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar), suggesting that a common structure could be responsible for toxic membrane permeabilization. However, the formation of amyloid-like fibrils made of bacterial toxins had not been reported until now, and this parallel was only based on a shared pore-like structure. The oligomerization states of the soluble secreted Mcc and of the toxic pore structures that are formed in target membranes, as well as their relations with the oligomeric intermediates that are present during the amyloid formation process, are currently under investigation in our laboratory. The finding that Mcc aggregation into amyloid fibrils is associated with a loss of toxicity against target bacteria supports the view that mature amyloid fibrils accumulating in amyloidoses may not be toxic but rather may be considered as inert end products or even as a protective mechanism to sequester the toxic intermediates (30Caughey B. Lansbury P.T. Annu. Rev. Neurosci. 2003; 26: 267-298Crossref PubMed Scopus (1458) Google Scholar). Whether Mcc amyloid formation in vivo is a regulated process possibly involving post-translational modifications or interactions with other proteins is an intriguing issue that still needs to be addressed. Mcc activity has been reported to be increased under certain growth conditions by the post-translational covalent linkage to its C terminus of a glucose moiety linked to a trimer of 2,3-dihydroxybenzoylserine (35Thomas X. Destoumieux-Garzon D. Peduzzi J. Afonso C. Blond A. Birlirakis N. Goulard C. Dubost L. Thai R. Tabet J.C. Rebuffat S. J. Biol. Chem. 2004; 279: 28233-28242Abstract Full Text Full Text PDF PubMed Scopus (125) Google Scholar). Although the mechanism by which post-translationally modified Mcc is more active than its non-modified counterpart remains unclear, one possibility would be that this modification stabilizes toxic oligomeric structures and/or prevents the formation of harmless mature amyloid fibrils. Indeed, it has been extensively reported that a variety of post-translational modifications could alter the ability of disease-associated amyloid proteins to aggregate into amyloid fibrils (36Atwood C.S. Martins R.N. Smith M.A. Perry G. Peptides. 2002; 23: 1343-1350Crossref PubMed Scopus (115) Google Scholar). Specific proteins and other compounds have also been shown to promote or inhibit amyloid formation by various proteins (37Soto C. Mol. Med. Today. 1999; 5: 343-350Abstract Full Text Full Text PDF PubMed Scopus (136) Google Scholar). Interestingly, enterochelin, a trilactone siderophore that is naturally produced by Enterobacteriaceae, has been reported to inhibit the activity of Mcc (38Orellana C. Lagos R. FEMS Microbiol. Lett. 1996; 136: 297-303Crossref PubMed Google Scholar). Whether this effect is related or not with amyloid formation remains to be determined. Our findings with Mcc show that protein aggregation into amyloid fibrils might be a generic property of proteins not necessarily associated to a pathological condition. Cytotoxicity may be determined by common features of specific types of aggregates rather than by the primary structure of the proteins involved (39Bucciantini M. Giannoni E. Chiti F. Baroni F. Formigli L. Zurdo J.S. Taddei N. Ramponi G. Dobson C.M. Stefani M. Nature. 2002; 416: 507-511Crossref PubMed Scopus (2164) Google Scholar). However, in contrast with amyloidoses where the protein aggregates accumulate in an uncontrolled manner, in the case of Mcc, bacteria use these properties to kill target cells as part of a normal physiological process that confers them competitive advantage. Amyloid formation was previously reported in bacteria when fibers produced by E. coli called curli were shown to consist of amyloid fibrils (40Chapman M.R. Robinson L.S. Pinkner J.S. Roth R. Heuser J. Hammar M. Normark S. Hultgren S.J. Science. 2002; 295: 851-855Crossref PubMed Scopus (979) Google Scholar). However, unlike Mcc, curli fibers are not involved in cellular toxicity but seem to have structural functions related to the colonization of inert surfaces and biofilm formation. The involvement of amyloid in regulating the biological activity of a bacterial protein suggests that formation of these fibrillar aggregates is an evolutionarily conserved property of proteins that is not only implicated in pathological conditions but also might be used for specific physiological functions. We thank Claudio Hetz, Céline Adessi, June Yowtak, and Octavio Monasterio for helpful discussions and technical advice.