Prion-like characteristics of the bacterial protein Microcin E492

Microcin E492 (Mcc) is a pore-forming bacteriotoxin. Mcc activity is inhibited at the stationary phase by formation of amyloid-like aggregates in the culture. Here we report that, in a similar manner as prions, Mcc naturally exists as two conformers: a β-sheet-rich, protease-resistant, aggregated, inactive form (Mccia), and a soluble, protease-sensitive, active form (Mcca). The exogenous addition of culture medium containing Mccia or purified in vitro-generated Mccia into the culture induces the rapid and efficient conversion of Mcca into Mccia, which is maintained indefinitely after passaging, changing the bacterial phenotype. Mccia prion-like activity is conformation-dependent and could be reduced by immunodepleting Mccia. Interestingly, an internal region of Mcc shares sequence similarity with the central domain of the prion protein, which is key to the formation of mammalian prions. A synthetic peptide spanning this sequence forms amyloid-like fibrils in vitro and is capable of inducing the conversion of Mcca into Mccia in vivo, suggesting that this region corresponds to the prion domain of Mcc. Our findings suggest that Mcc is the first prokaryotic protein with prion properties which harnesses prion-like transmission to regulate protein function, suggesting that propagation of biological information using a prion-based conformational switch is an evolutionary conserved mechanism.


Results
Two forms of Mcc are functionally, biochemically, and structurally distinct. Two distinct forms of Mcc have been observed during the growth phase of Klebsiella pneumoniae RYC492 strain 21,25 . The active form of Mcc (hereafter designated as Mcc a ) is produced during the exponential phase of growth, while the inactive form of Mcc (hereafter termed Mcc ia ) appears during the stationary phase (Fig. 1A). Such a phenomenon could easily be replicated by growing Mcc producing E. coli VCS257 cells harboring the pJEM15 plasmid in M9 minimal medium at 37 °C. Mcc activity was determined by the critical dilution method (CDM), as described in "Materials and Methods" (Fig. 1B). The activity of Mcc increased exponentially, peaking at 9 h in the exponential phase. The activity started to decline around 12 h, becoming undetectable at 24 h in the stationary phase (Fig. 1C). The loss of Mcc activity in the stationary phase was paralleled by the formation of an altered conformation of Mcc that resulted in an increase in protease resistance. For this experiment, aliquots of culture medium from stationary and exponential phases were subjected to limited proteolysis with proteinase K (PK; 1-1000 μ g/ml) for 30 min at 37 °C. The aliquots coming from the stationary phase, containing mostly Mcc ia , showed a remarkable resistance to proteolysis compared to Mcc a , coming from the exponential phase, as shown by immunoblot analysis (Fig. 1D). Strikingly, a large proportion of Mcc ia remained undigested even after incubation with a very high concentration of PK (1000 μ g/ml), resembling the behavior of PrP Sc .
The conversion of Mcc a into Mcc ia that occurs in vivo during the bacteria culture growth was reproduced in vitro using purified Mcc. The highly purified Mcc a (400 μ g/ml) from the exponential phase (9 h) of Mcc producing bacteria was allowed to aggregate in 50 mM PIPES, pH 6.5 containing 500 mM NaCl at 37 °C and the activity of Mcc before (0 h) and after 24 h of incubation was determined by CDM. Purified Mcc a before aggregation was highly active, whereas after 24 h of incubation, Mcc activity diminished substantially (P = 0.0052), resembling the in vivo conversion of Mcc a into Mcc ia (Fig. 1E). The change in bacteriotoxin activity by in vitro aggregation was associated with the acquisition of protease resistance. An immunoblot analysis before and after PK digestion clearly showed that Mcc a was sensitive to PK, whereas in vitro-generated Mcc ia (by 24 h of incubation) exhibited remarkable resistance to PK (Fig. 1F). These findings indicate that Mcc structural changes, in the absence of any other factor, are responsible for the functional changes in biological activity.
Mcc ia is a β-sheet rich, amyloid-like aggregate. One of the hallmark properties of mammalian and yeasts prions is the formation of β -sheet rich amyloid-like structures by a seeding/nucleation mechanism 8,28 . Interestingly, the conversion of Mcc a into Mcc ia is also associated with the formation of amyloid-like structures, which bind to the amyloid-specific dyes: Thioflavin T (ThT) and Congo red (CR). Incubation of purified Mcc a in vitro led to the formation of amyloid-like structures through a nucleation-dependent mechanism as illustrated by the ThT binding assay ( Fig. 2A). The amyloid nature of the aggregates was further confirmed by a CR binding assay, in which CR bound to Mcc ia displayed a characteristic increase in absorption and a spectral shift to reach a peak at ~540 nm compared to Mcc a (Fig. 2B). Analysis by fourier-transformed infrared spectroscopy (FTIR) revealed a maximum absorption at ~1654 cm −1 for Mcc a , consistent with a predominance of α -helix/random coil structure. However, Mcc ia exhibited the maximum absorption at ~1639 cm −1 indicating that β -sheets are the predominant structure (Fig. 2C) 29 . Further evidence for the structural differences between Mcc a and Mcc ia was the recognition with the conformational antibody A11. This antibody has been shown to specifically recognize β -sheet-rich oligomers produced during the process of amyloid formation of many different proteins 30 . Using a dot blot assay, we found that while Mcc ia was readily recognized by A11, Mcc a was not (Fig. 2D). Finally, while no detectable polymeric structures were seen for Mcc a under electron microscopy, an aliquot of Mcc ia was greatly enriched in unbranched amyloid-like fibrils of variable length (Fig. 2E).
To further investigate whether conversion of Mcc a into Mcc ia amyloid aggregates is the key event in changing the activity of the protein, we studied the effect of known and general inhibitors of amyloid formation on Mcc activity. For this purpose, we used curcumin and CR, which have been shown in several reports to prevent protein misfolding and aggregation into amyloid-like structures formed by diverse proteins [31][32][33] . Mcc producing bacteria were grown in the absence (control) or presence of either 25 μ M curcumin or 50 μ M CR. Aliquots of culture medium were subjected to Mcc activity assay and limited proteolysis. The addition of curcumin or CR significantly increased the time in which Mcc was active, suggesting an inhibitory effect of these molecules on the Mcc producing bacteria were grown in minimal medium at 37 °C. Aliquots were removed periodically, and activity was measured by CDM (see "Materials and Methods"). Activity was measured in duplicate samples and error bars indicate standard deviation (S.D). (D) Structural changes of Mcc reflected by differences on resistance to proteolytic degradation. Immunoblots of PK-treated aliquots of culture at the exponential and stationary phase, containing Mcc a (9 h in culture) and Mcc ia (24 h in culture), respectively. Samples in Lanes 2-5 and 7-10, were treated with increasing concentrations of PK (1, 10, 100 and 1000 μ g/ml). Samples in lanes 1 and 6 were left untreated. E: The activity of purified Mcc a (400 μ g/ml) before and after aggregation into Mcc ia at 37 °C for 24 h was measured by CDM. Error bars indicate S.D. The differences were statistically analyzed by the student t-test (** P = 0.0052). F: Acquisition of protease-resistance upon in vitro conversion of Mcc a into Mcc ia . Aliquots of purified Mcc a (t = 0 h) and after conversion into Mcc ia (t = 24 h) were subjected to the same treatment with PK as described in panel D. Mcc a (lanes 1-5) and Mcc ia (lanes 6-10). Samples in Lanes 2-5 and 7-10, were treated with increasing concentrations of PK (1, 10, 100 and 1000 μ g/ml). Samples in lanes 1 and 6 were left untreated.
Scientific RepoRts | 7:45720 | DOI: 10.1038/srep45720 The black line represents the experimental spectra and the colored lines are the deconvoluted components corresponding to the various structural motifs as obtained by computer analysis. The peak assignments was as following: ~1639 and ~1691 cm −1 for β sheets; ~1654 cm −1 for α -helix; ~1648 cm −1 for random structure; ~1667,~1675 and ~1680 cm The differences in Mcc activity between control and treated samples were highly significant (P < 0.0001) in both variables (time and treatment, as well as the interaction between them) as measured by two-way ANOVA. Posthoc analysis by the Bonferroni test revealed also significant differences between treatment and controls at times 9, 12 and 18 h (*P < 0.05; ***P < 0.001). (G) Immunoblots of aliquots removed at different times from cultures used in panel F before (− ) and after treatment (+ ) with PK (3 μ g/ml).  (Fig. 2F). This effect was not due to a change on bacterial growth, since under these conditions the rate of growth was the same in the presence or the absence of these molecules (data not shown). The influence of these molecules on the Mcc conformational change could be further monitored by studying the rate of proteolytic degradation of Mcc by PK. Both CR and curcumin completely inhibited the conversion of PK-sensitive Mcc a into PK-resistant Mcc ia until 12 h as compared to the control, where complete acquisition of PK-resistance was observed at that time point (Fig. 2G). Altogether, these results demonstrate that formation of Mcc ia amyloid fibrils sequesters functional Mcc a into the aggregates, leading to the loss of bacteriotoxin activity.

The exogenous addition of culture medium containing Mcc ia induces the prion-like conversion of Mcc a into Mcc ia in vivo.
The hallmark feature of prions is that the pathological, infectious form (PrP Sc ) interacts with endogenous PrP C and catalyzes its conformational change to produce more PrP Sc . To test whether Mcc ia is capable of catalyzing the Mcc a to Mcc ia conversion, Mcc producing bacteria were grown in minimal media at 37 °C in the absence (control) or in the presence of 10% (v/v) whole culture or 5-20% (v/v) culture supernatant, both taken from the stationary phase of the culture when Mcc ia is naturally produced (see "Materials and Methods"). Small aliquots were periodically removed and subjected to the Mcc activity assay by CDM and proteolysis by PK (3 μ g/ml). As expected, the aliquots from Mcc producing bacteria grown in the absence of Mcc ia showed maximum activity at 9 h that started to decline at 12 h, and was undetectable by 24 h (Fig. 3A). The loss of activity correlated with the appearance of PK-resistant Mcc at 12 h, whereas Mcc was completely sensitive to PK until 9 h (Fig. 3D). The addition of 5 to 20% (v/v) culture supernatant containing Mcc ia significantly diminished the amount of active Mcc present in the bacteria culture in a dose-dependent manner, reaching a complete effect when 10% of the material was added (Fig. 3A). Consistently, the loss of activity was paralleled with the production of PK-resistant Mcc (Fig. 3D). To exclude the possibility that the loss of Mcc activity after addition of bacterial culture medium was due to another factor present in the medium rather than the prion-like activity of Mcc ia , we conducted a control experiment in which culture medium from the stationary phase of Another important feature of prions is that once PrP Sc is formed, it can self-perpetuate indefinitely by serial passages in infected animals. Thus, we wanted to explore whether this also happens with Mcc ia after its first cycle of conversion. For this purpose, Mcc producing bacteria were grown until 48 h (stationary phase; P0). After 48 h, 10% (v/v) of whole culture was added into the fresh culture medium of Mcc producing bacteria and grown for a subsequent 48 h (P1). This cycle was repeated up to 3 passages (Supplementary Figure 3A). Aliquots from all passages were removed, and Mcc activity was analyzed as described before. In P0, where no Mcc ia was added, Mcc activity showed a typical trend; increase and then decrease in activity at the stationary phase. However, in all passages P1 to P3, Mcc activity was undetectable at any time point (Supplementary Figure 3B  Aliquots were removed at indicated times and activity of Mcc was measured by CDM. The differences in Mcc activity between control and the various treatments were highly significant (P < 0.0001) in both variables (time and treatment, as well as the interaction between them) as measured by two-way ANOVA. Post-hoc analysis by the Bonferroni test revealed also significant differences between treatment and controls at times 9 and 12 h (***P < 0.001). (B) Mcc producing bacteria were grown in minimal medium in the presence of 20% (v/v) culture supernatant (S) and or 10% (v/v) whole (W) culture of E. coli VCS257 deprived of Mcc plasmid (Mcc -). Activity of Mcc was measured by critical dilution. In this case the differences between control and treatments were not significant as analyzed by two-way ANOVA. (C) Mcc producing bacteria were grown in minimal medium at 37 °C either in the absence (control) or in the presence of 20% (v/v) culture supernatant (S) after immunodepletion using an antibody specific for the sequence of Mcc. Aliquots were removed at indicated times and activity of Mcc was measured by CDM. In this experiment the treatment produced significant differences compared to the untreated control (used the same as panel A, top graph). Individual differences were analyzed by the Bonferroni post-test (*P < 0.05; **P < 0.01; ***P < 0.001). In panels A, B and C, activity was measured in duplicate samples and error bars indicate S.D. D: Immunoblots of aliquots removed from culture used in panels A, B and C before (− ) after treatment (+ ) with PK (3 μ g/ml) at indicated time points.
Scientific RepoRts | 7:45720 | DOI: 10.1038/srep45720 Prion-like activity of Mcc ia is conformation dependent. Although the prion activity of mammalian and yeast prions is resistant to harsh procedures that destroy nucleic acids, it is sensitive to agents that alter protein conformation 34 . To gain insight into whether the conversion of Mcc a into Mcc ia is conformation dependent, Mcc producing bacteria were exposed to Mcc ia untreated or subjected to heat or chemical denaturation. For this purpose, Mcc ia seeds either produced in vivo or generated in vitro were denatured by boiling for 15 min, or by incubating with 6 M guanidine hydrochloride (GdnHCl) for 2 h at room temperature (see "Materials and Methods"). Both procedures disrupted the conformation of Mcc ia , as evaluated by dot blot using the A11 conformational antibody (Fig. 5A). However, Mcc could be easily detected by an antibody specific for the C-terminal sequence of Mcc.

Figure 4. Purified in vitro-generated Mcc ia induces the conversion of Mcc a into Mcc ia in vivo. (A) E. coli
VCS257pJEM15 was grown in minimal medium at 37 °C either in the absence (control) or in the presence of purified Mcc ia (2.5, 5 and 10 μ g/ml, final concentration; see "Materials and Methods"). Aliquots were removed at indicated time points and Mcc activity was measured by CDM. Activity was measured in duplicate samples and error bars indicate S.D. The differences in Mcc activity between control and treated samples were highly significant (P < 0.0001) in both variables (time and treatment, as well as the interaction between them) as measured by two-way ANOVA. Post-hoc analysis by the Bonferroni test revealed also significant differences between treatment and controls at times 9 and 12 h (***P < 0.001). (B) Immunoblots of aliquots removed from culture used in panel A before (− ) and after treatment ( + ) with PK (3 μ g/ml) at indicated time points.
Scientific RepoRts | 7:45720 | DOI: 10.1038/srep45720 Mcc producing bacteria were grown in the absence (control) or presence of culture supernatant containing Mcc ia [20% (v/v)] or in vitro-generated, purified Mcc ia (5 μ g/ml) as such or after denaturation either by boiling or by treatment with 6 M GdnHCl. Aliquots from the culture were periodically removed, and Mcc activity and its resistance to PK were tested. The addition of untreated purified, in vitro-generated Mcc ia or culture containing Mcc ia consistently resulted in loss of Mcc activity and early formation of PK resistant Mcc (Fig. 5B-E); however, the addition of purified Mcc ia or culture containing Mcc ia after exposure to boiling and to GdnHCl had no effect either on Mcc activity or on the formation of the PK resistant Mcc as compared to the control (Fig. 5B-E). These results suggest that destruction of Mcc ia conformation abolishes its ability to self-propagate and induce the phenotypic changes to the bacterial culture.
Mcc behaves as a prion in a yeast prion reporter assay. To further explore the prion-like behavior of Mcc, we used a well-established prion reporter assay in yeast, based on [psi -] and [PSI + ] states of the translation termination factor Sup35p 35 . The prion domains in yeast prions including Sup35p are modular and can transfer the prion behaviors to heterologous proteins 36 . We used this feature of yeast prions to generate Mcc-Sup35MC fusion protein, by replacing the N-terminal prion domain of Sup35 with full-length mature Mcc which was fused to the functional MC domain of Sup35p (see "Materials and Methods" for details) (Fig. 6A). The fusion protein was tested for heritable [PSI + ] state using the ADE1 gene containing a premature stop codon. In [PSI + ] cells, Sup35p is assembled into an inactive self-perpetuating prion form which does not take part in premature translation termination and thus cells are able to grow on adenine-deficient medium and produce white colonies (Fig. 6B). Whereas in the case of [psi − ] state, cells do not make functional Ade1, and red colonies appear due to the accumulation of red by-product. Surprisingly, like other yeast prions, Mcc conferred prion behavior to Sup35MC. Indeed, cells expressing Mcc-Sup35MC fusion protein gave rise to white colonies on YEPD plates, which were maintained after repeated streaking (Fig. 6B) in a similar manner as Sup35p in the [PSI + ] state (positive control). The propagation of a prion phenotype for all known yeast prions is dependent on the activity of the heat shock protein 104 (Hsp104) 37 ; therefore, yeast prions can be cured by inhibiting Hsp104 either by chemical inhibition or by genetic manipulation 38,39 . Similar to Sup35p (used as a positive control), the prion form of Mcc-Sup35MC fusion protein could also be cured by transferring a white colony to YEPD plates containing 3 mM of GdnHCl, a known inhibitor of Hsp104 (Fig. 6B). These results clearly suggest that Mcc confers to Sup35MC the ability to exist in two distinct physical and functional states, which are interconvertible and heritable by the prion mechanism.
To further confirm the prion behavior of Mcc-Sup35MC in yeasts, we studied the formation of aggregates by a sedimentation assay. The prion form exists in high assemblies and can be pelleted by high-speed centrifugation, whereas non-prion forms mostly exist in a soluble state 40 (Fig. 6C). These results clearly suggest that change in heritable phenotype is, indeed, due to the distinct physical states of Mcc-Sup35MC fusion protein in yeast.

Identification of the Mcc prion domain (PrD).
To study the region of Mcc responsible for its prion-like activity, we compared the sequence of Mcc with that of human PrP. As shown in Fig. 7A, a remarkable sequence similarity was found between the central region of mature Mcc (amino acids  with the internal hydrophobic core region of PrP (amino acids 111-133). Indeed, a 57% sequence identity and a 74% similarity was observed in these regions between these proteins from two greatly distant organisms. Importantly, various pieces of evidence suggest that this region of PrP plays an important role in the formation of PrP Sc (refs 41 and 42). To analyze whether this region of Mcc might be the domain responsible for prion activity, we investigated if this sequence can form self-replicating amyloid aggregates that can induce the conversion of full-length Mcc a into Mcc ia in vitro and in vivo. For this purpose, we used a synthetic peptide comprising the residues 12-37 of mature Mcc. The reason for using this sequence instead of Mcc  was to increase the peptide solubility and easiness to handling. To examine whether Mcc(12-37) forms amyloid aggregates in vitro, we performed a ThT binding assay and analyzed the aggregates' morphology by electron microscopy. Purified synthetic Mcc(12-37) at a concentration of 228 μ g/ml was allowed to aggregate and aliquots were tested for ThT binding. As shown in Fig. 7B, Mcc(12-37) formed amyloid following a nucleation-dependent mechanism, characterized by a 12 h lag phase. Interestingly, preformed aggregates of Mcc  were capable of seeding the aggregation of both soluble Mcc(12-37) and full-length purified Mcc a , reducing their lag phase (Fig. 7B). The amyloid nature of Mcc(12-37) aggregates was confirmed by transmission electron microscopy (Fig. 7C). We

Discussion
The capability of proteins to self-propagate and transmit biological information in a similar manner as genetic material is a recently recognized concept. Prions were first identified as proteinaceous infectious agents responsible for various catastrophic neurodegenerative disorders known as prion diseases or TSEs 4 . In TSEs, a naturally occurring protein undergoes conformational changes leading to the formation of a misfolded and aggregated form, termed PrP Sc . Like a typical infectious micro-organism, PrP Sc can be transferred from individual-to-individual by various routes of administration and replicate in the new host forming more PrP Sc . PrP Sc self-replication involves the catalytic conversion of the normal isoform of the prion protein (PrP C ) through a seeding/nucleation mechanism, in which the PrP Sc aggregate binds PrP C and promotes its misfolding by  principle 8 . More importantly, the prion mechanism of transmission of biological information by propagation of protein conformational changes has been shown not to be restricted to diseases, but to operate in diverse organisms to modulate the biological function of certain proteins 8,17 . The seminal discovery of self-perpetuating prion proteins in yeast and other fungus has proven that prion-like conformational changes can also have a beneficial outcome 16 . Unlike prion protein, yeast prions are generally nonpathogenic and produce distinct phenotypes with different physiological functions, providing a competitive advantage in adverse conditions 11 . Later, several other beneficial prions have been discovered that use prion-like conformational changes to propagate biological signals. For example, the self-perpetuating fiber-like polymers of MAVS regulate mammalian antiviral immune defense 19 and a translation regulator of the invertebrate Aplysia forms self-perpetuating polymers that help to maintain long-term potentiation in sensory neurons 18 . These findings clearly suggest that diverse organisms harness prion-like conformational changes in several unrelated proteins to regulate biological function. However, it is currently unclear how frequent and universal the use of the prion principle is for modulating the biological activity of proteins.
Here, we report a bacterial protein that exhibits prion-like characteristics and could be considered as the first example of a prokaryotic prion. However, in contrast with mammalian prions, the prion form of Mcc is not toxic and the non-prion form is the toxic one. This apparent discrepancy may not be so, because evidence accumulated in the past several years have shown that the formation of large amyloid aggregates in TSEs as well as other amyloid-related disorders (e.g. Alzheimer's or Parkinson's diseases) may actually be a protective mechanism to decrease the amount of small oligomers which are the real toxic species 43 . Another contrasting feature that differentiates Mcc from other prions is that Mcc ia is located in the extracellular space and thus the conformational information inherent to Mcc is dissociated from the organism and not passed on to neighboring or daughter cells. Because of this, the biological changes are not transmitted to another organism, but to the environment in which bacteria live. Currently, the definition of prions is changing in the field thanks to many reports showing prion-like properties of various other proteins. Thus, at this time is yet unclear whether prions need to be associated to cells or changing the biological properties of cells upon transmission. For example, many reports have provided evidences for the Alzheimer's amyloid-beta protein behaving like a prion [44][45][46] , and this is also an extracellular protein which arguably changes the environment in the brain (not necessarily the cells). Bacteria living in the gut and releasing Mcc into the medium may produce a very similar result as amyloid-beta in the brain of Alzheimer's patients.
One of the surprising features of the Mcc prion behavior is its sequence similarity with the central region of PrP. The sequence 16-37 of Mcc shows 57% identity and 74% similarity with the PrP sequence spanning residues 111-133 (Fig. 7A). Various pieces of evidence have shown that this region of PrP plays an essential role in PrP conversion and its pathological properties, including: (a) The synthetic peptide comprising the sequence 106-126 of PrP has been widely used to model prion neurotoxicity [47][48][49]  Our findings indicate that Mcc exhibits several characteristics of prions; however, the functional relevance of the Mcc prion mechanism is unclear. Mcc producing bacteria secrete Mcc a that kills competing bacteria, helping them to occupy a spatial niche in a given ecosystem. However, when Mcc producing bacteria have prevailed, Mcc a is no longer needed. We can speculate that under these conditions, the prion-like mechanism may operate to encapsulate Mcc a into Mcc ia , protease resistant aggregates that may serve as a reservoir for Mcc a . Later if the conditions of the ecosystem change and competition arise, Mcc ia aggregates may undergo conformational changes that release Mcc a thus providing a competitive advantage. We have previously shown that Mcc ia amyloid aggregates can indeed release Mcc a upon changing environmental conditions 52 . In summary, our findings indicate that the prion mechanism is not restricted to eukaryotic organisms, but is likely an ancient process of regulation of protein function by rapid self-propagation of changes in protein conformation. Thus, the prion principle is a conserved process occurring all the way from bacteria to humans.

Materials and Methods
Chemicals were obtained from Sigma (St. Louis, Mo), unless otherwise stated.

Purification of Microcin E492 (Mcc).
Mcc was purified from the culture supernatants as described earlier 27 . In brief, E. coli VCS257 cells harboring pJEM15 plasmid were grown in M9 minimal medium containing 0.2% glucose, 0.2% sodium citrate, 1 g/liter casamino acid, 1 mg/liter thiamine and 100 mg/liter ampicillin to an optical density of 1.2 at 600 nm at 37 °C with shaking. Bacterial cells and debris were removed by centrifugation at 4000 rpm for 10 min. The resultant supernatant was passed through a Sep-Pak C18 cartridge (Waters). The cartridge was sequentially washed with 65% methanol, and 25% acetonitrile. Finally, the bound Mcc was eluted with 50% acetonitrile, and lyophilized. Lyophilized powder of Mcc (purified active Mcc) was stored at − 20 °C until used. Under these conditions, the preparation contains highly purified Mcc ( > 90%) as evaluated by silver stained gels 53 . Transmission electron microscopy (TEM). An aliquot of 10 μ l of reaction sample was placed onto Formvar-coated 200-mesh copper grids for 5 min, washed at least three times with distilled water, and then negatively stained with 2% uranyl acetate for 1 min. Grids were examined by electron microscopy (H-7600, Hitachi, Japan) operated at an accelerating voltage of 80 kV.

Aggregation of purified Mcc
Congo red (CR) binding. Amyloid-like nature of Mcc ia was examined by binding of the amyloid specific dye Congo red by the spectroscopic band shift assay 27 . Purified Mcc a was allowed to aggregate in 50 mM PIPES (pH 6.5) containing 500 mM NaCl at 37 °C with vigorous shaking for 24 h. An aliquot was incubated with 12.5 μ M CR for 30 min at 37 °C. The absorbance spectrum was recorded from 400 nm to 700 nm.
Proteinase K (PK) digestion and Immunoblotting. Samples were incubated with the indicated concentrations of PK at 37 °C for 30 min. The reaction was stopped by boiling of the sample in NuPAGE LDS buffer at 100 °C for 10 min. Then, samples were resolved by NuPAGE 4-12% Bis-Tris gels (Invitrogen). Proteins were electrophoretically transferred to nitrocellulose membranes (Amersham Biosciences, Germany). Membranes were blocked with 5% w/v nonfat dry milk in Tris-buffered saline-Tween 20 (TBS-T, 20 mM Tris, pH 7.2, 150 mM NaCl and 0.05% (v/v) Tween 20) at room temperature for 2 h. After blocking, the membranes were probed with anti-Mcc antibody (1:3000) and the anti-rabbit horseradish peroxidase-conjugated secondary antibodies (1:5000). The blots were visualized using enhanced chemiluminescence plus western blotting detection kit (Amersham Biosciences, Piscataway, NJ).

Dot blot analysis.
Five μ l of each reaction was spotted onto nitrocellulose membranes (Amersham Biosciences, Germany), and air dried for 30 min at room temperature. Finally, blots were blocked and visualized as described above (see PK digestion and Immunoblotting).
Mcc activity assay. Activity of Mcc in each sample was measured by critical dilution method (CDM) as described previously 55 . In brief, samples were centrifuged at 16,500 × g for 10 min at room temperature. The resultant supernatant was serially diluted in minimal medium, and 5 μ l aliquots were laid onto Luria Broth (LB) agar plates overlaid with Mcc-sensitive E.coli BL21 (DE3) p11α 2 cells. After 16 h of incubation at 37 °C, the activity of Mcc in the last dilution that gave detectable inhibition (approximately 2-times above the background levels) was determined and expressed in arbitrary units/ml 55 . Immunodepletion of Mcc ia . Dynabeads sheep anti-rabbit antibody (Life Technologies, Oslo, Norway) were conjugated with anti-Mcc antibody overnight at 4 °C as per manufacturer's instructions. Following conjugation, beads were washed and incubated with culture supernatant containing Mcc ia overnight at 4 °C, beads were removed with a magnet and the resulting supernatants were used for seeding.
Fourier-Transform Infrared spectroscopy (FTIR). FTIR experiments were conducted in an FT/IR-4100 spectrometer from JASCO. Purified Mcc a (400 μ g/ml) was used as such or aggregated for 24 h, as described previously. Protein slurry was then placed on the top of a diamond PRO450-S Attenuated Total Reflectance unit from JASCO adapted to the FT/IR-4100 system. System parameters included 4.0 cm −1 resolution and an accumulation of 80 scans per sample. The data was processed using Cosine anodization and Mertz phase correction. The data was also corrected for ATR and carbon dioxide vapor absorption. Data fitting and secondary structure calculations of samples were analyzed, after buffer subtraction, by the Secondary Structure Estimation 4000 software from JASCO. Each sample was measured in triplicates to confirm the secondary structure.
Sedimentation assay. A single colony of each strain expressing Sup35p or Mcc-Sup35MC was inoculated in 3 ml liquid YEPD alone or supplemented with 3 mM GdnHCl, and grown overnight at 30 °C with shaking. Cultures grown overnight were diluted into fresh media to a density of OD 600 = 0.05 and grown to OD 600 = 0.6 before harvesting. Cells were resuspended in buffer A (50 mM Tris-Hcl, pH 7.5; 50 mM KCl; 10 mM MgCl 2 ) containing 10 mM PMSF and protease inhibitor (Roche). Cells were lysed with a bead beater using glass beads and were briefly spun at 3,000 rpm for 3 min to sediment cell debris. Lysates were incubated with SDS (final SDS concentration 1%) and triton X-100 (final concentration 0.5%) and centrifuged for 1 h at 40,000 rpm. Supernatant and pellet fractions were recovered. The total cell extracts, soluble, and pellet fractions of each sample were analyzed by SDS-PAGE and immunoblot analysis using an antibody, anti-Sup35C and C-terminal specific Mcc antibody.
Statistical Analysis. The significance of differences between the Mcc bacteriotoxic activity under different conditions was analyzed by two-way analysis of variance (ANOVA) using time and treatment as the variables. To assess differences in specific time points the data was analyzed by the Bonferroni post-test. Student's t-test was used to analyze differences of activity between untreated, purified Mcc a and Mcc ia (Fig. 1E). The level of significance was set at P < 0.05.