Widespread Abundance of Functional Bacterial Amyloid in Mycolata and Other Gram-Positive Bacteria

Abstract

Until recently, extracellular functional bacterial amyloid (FuBA) has been detected and characterized in only a few bacterial species, including Escherichia coli, Salmonella, and the gram-positive organism Streptomyces coelicolor. Here we probed gram-positive bacteria with conformationally specific antibodies and revealed the existence of FuBA in 12 of 14 examined mycolata species, as well as six other distantly related species examined belonging to the phyla Actinobacteria and Firmicutes. Most of the bacteria produced extracellular fimbriae, sometimes copious amounts of them, and in two cases large extracellular fibrils were also produced. In three cases, FuBA was revealed only after extensive removal of extracellular material by saponification, indicating that there is integrated attachment within the cellular envelope. Spores of species in the genera Streptomyces, Bacillus, and Nocardia were all coated with amyloids. FuBA was purified from Gordonia amarae (from the cell envelope) and Geodermatophilus obscurus, and they had the morphology, tinctorial properties, and β-rich structure typical of amyloid. The presence of approximately 9-nm-wide amyloids in the cell envelope of G. amarae was visualized by transmission electron microscopy analysis. We conclude that amyloid is widespread among gram-positive bacteria and may in many species constitute a hitherto overlooked integral part of the spore and the cellular envelope.

The gram-positive bacterial group mycolata (mycolic acid-containing actinomycetes) comprises a number of genera with disease-causing species, including the severely pathogenic organisms Corynebacterium diphtheriae and Mycobacterium tuberculosis. The latter species is the leading cause of death due to a single infectious agent globally (17). Furthermore, mycolata have great environmental and economical impact, since several species (e.g., Gordonia spp.) may lead to unwanted foaming in wastewater treatment plants (10, 27, 43). M. tuberculosis was recently shown to use long entangled pili (MTP) to adhere to endothelium, eventually invading and infecting human and animal tissue (1). MTP's morphology and tinctorial properties are very similar to those of the amyloid-like curli fibrils found in Escherichia coli and Salmonella species (7), although it has not been determined whether they contain the characteristic cross-β structure with β-strands perpendicular to the long fibril axis (44). In higher organisms, amyloid occurs mainly as an aberrant product of protein misfolding in, e.g., neurodegeneration and systemic amyloidosis, but bacteria are adept at turning amyloid to good use. In addition to the two bacteria mentioned above, functional bacterial amyloid (FuBA) has also been reported for streptomycetes (8) and xanthomonads (35). These examples are only the tip of the iceberg. Our recent in situ studies using WO2 antibodies specific for the amyloid conformation (36) in conjunction with 16S rRNA-targeted oligonucleotide probes for identification of the microbes revealed that amyloid-like adhesins are widespread in many phyla in environmental biofilms (29). In view of the occurrence of potential amyloid-like fibrils in one species belonging to the mycolata and the observed link between infection by a mycolata genus (Nocardia) and neurodegenerative Parkinson's disease (13, 25, 26, 47), we have investigated this group of bacteria more closely for the presence of amyloid. Here we show that 12 of 14 different species of mycolata, as well as 6 of 6 other gram-positive bacteria, harbor amyloid. Furthermore, in some cases the amyloid can be visualized only after harsh saponification procedures which remove surrounding lipid molecules, indicating that the amyloid is deeply embedded in the cell envelope. Thus, amyloid may play a hitherto unappreciated central role in the composition of the bacterial envelope in many gram-positive bacteria.

MATERIALS AND METHODS

Organisms and growth conditions.

All bacteria were grown in liquid shaken cultures (120 rpm, 28°C). The E. coli SM2257 curli-deficient mutant (39) and E. coli SM2258 with upregulated curli production (48) were grown in liquid M63 minimal media (29). The following gram-positive organisms were obtained from Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH (http://www.dsmz.de), were grown in the media indicated, and were tested to determine their abilities to produce FuBA: Corynebacterium flavescens DSM20296 (medium M53), Corynebacterium glutamicum DSM20300 (medium M53), Dietzia maris DSM43672 (medium M65), Dietzia papillomatosis DSM44961 (medium M65), Gordonia amarae DSM43392 (medium M65), Gordonia hydrophobica DSM44015 (medium M535), Mycobacterium avium DSM44156 (medium M645), Mycobacterium phlei DSM43239 (medium M645), Nocardia asteroides DSM43757 (liquid standing culture in medium M65), Nocardia polyresistens DSM44932 (medium M65), Tsukamurella paurometabola DSM20162 (medium M53), Tsukamurella spumae DSM44113 (medium M65), Williamsia maris DSM44693 (medium M65), Williamsia muralis DSM44343 (medium M535), Actinospica acidiphila DSM44926 (medium M987), Bacillus mycoides DSM2048 (medium M1), Enterococcus aquimarinus DSM17690 (medium M92), Geodermatophilus obscurus DSM43160 (medium M65), Streptosporangium cinnabarium DSM44094 (medium M547), and Streptomyces coelicolor DSM40233 (medium M65). The presence of spores was detected by malachite staining (5).

Saponification.

Fifty-milliliter 3-week-old bacterial cultures in stationary phase were pelleted by centrifugation (10,000 × g, 10 min), the growth medium of each culture was removed, and the pellet was resuspended in 0.5% (wt/vol) KOH dissolved in absolute ethanol. Saponification was performed in an incubator (80°C, 200 rpm) using Teflon-sealed Greiner tubes placed in an upright position (3, 22). After 4 days, the remaining bacteria were pelleted and resuspended in phosphate-buffered saline (PBS).

Antibody labeling and fluorescence microscopy of amyloid.

A WO2 labeling (36) immunofluorescence protocol optimized for use with bacterial samples was performed as described previously (29). DAPI (4′,6-diamidino-2-phenylindole) staining was used to compare the relative positions of bacteria to the positions of bound WO2. DAPI staining was performed in the dark for 15 min on slides with air-dried bacteria using 25 μg/ml DAPI in PBS as described previously (18).

Purification of FuBA.

Twenty milliliters of a dense 5-day (28°C, 120 rpm) culture of G. amarae, C. glutamicum, or G. obscurus in exponential phase was used to inoculate 1 liter of M63 medium. After 3 weeks of growth (28°C, 120 rpm), stationary-phase bacteria and extracellular matrix were harvested by centrifugation (16,000 × g, 30 min). The pellet was resuspended in PBS and sonicated (B. Braun Labsonic 1000L rod sonicator) on ice at medium intensity (30 min, 120 W) to liberate FuBA from bacteria. The cells were pelleted by centrifugation (2,000 × g, 4°C, 10 min), and the supernatant was discarded. Pelleting of cells was repeated three times until no cells were visible in the supernatant as determined by phase-contrast microscopy. FuBA and other insoluble contaminants in the supernatant were pelleted in an ultracentrifuge (30,000 rpm, 10°C, 30 min, Sorvall T647.5 rotor), washed once in 10 mM Tris-HCl (pH 8.0), centrifuged, and treated for 10 min with 10 ml of 95°C 2% (wt/vol) sodium dodecyl sulfate (SDS) in 10 mM Tris-HCl (pH 8.0). The solution was cooled, pelleted by centrifugation (100,000 × g, 10°C, 30 min, Sorvall T640.1 rotor), and subjected to a second hot SDS treatment. SDS-treated amyloid material was pelleted, resuspended in SDS-polyacrylamide gel electrophoresis (PAGE) sample buffer, incubated at 95°C for 15 min, and loaded on a 6-cm preparative 12% polyacrylamide gel cast in a Greiner-tube. The recipes used for sample buffer and the polyacrylamide gel were those described by Laemmli (28). A constant current of 20 mA was applied to the preparative SDS-PAGE purification system for 6 h, after which the remaining amyloid material on top of the gel was recovered using careful resuspension with a pipette. The preparative SDS-PAGE purification step was repeated, and finally the amyloid material was washed extensively in MilliQ water.

Electrophoresis.

Since FuBA fibrils could not enter a polyacrylamide gel, an extraction protocol with trifluoroacetic acid (TFA) was used to liberate FuBA monomers. Extraction with 100% TFA was performed as described by de Vries et al. (12). Briefly, FuBA was pelleted by centrifugation (13,000 × g, 15 min), the supernatant was removed, 100 μl 100% TFA was added, the material was resuspended thoroughly until the solution was free of aggregates, and the solution was evaporated to dryness using a stream of N₂. The samples were then resuspended in 20 μl of SDS-PAGE sample buffer and subjected to electrophoresis. Samples were not heated prior to electrophoresis since heating was known to induce monomer aggregation. Proteins were visualized by staining with 0.25% (wt/vol) Coomassie brilliant blue R-250 (Sigma).

ThT fluorescence.

FuBA (100 μg/ml) was sonicated for 30 s at medium intensity prior to analysis. Thioflavin T (ThT) was used at a final concentration of 40 μM in PBS, and the ThT emission spectrum from 465 to 600 nm was determined with a PerkinElmer LS55 luminescence spectrofluorometer using excitation at 450 nm, emission and excitation bandwidths of 5 nm, and a scan speed of 200 nm/min. The temperature was kept constant at 25°C, and three spectra were averaged to improve the signal-to-noise ratio.

Secondary structure analysis.

Far-UV circular dichroism (CD) spectra were recorded with a Jasco J-810 spectropolarimeter as described previously (37). Molar ellipticity was calculated based on an average amino acid molecular mass of 110 Da. Fourier transform infrared spectrometry (FTIR) spectra were recorded and analyzed using a 1-μl sample and a Tensor 27 (Bruker) FTIR spectrophotometer as described previously (37).

TEM.

G. amarae was grown for 3 weeks in minimal M63 medium (120 rpm, 28°C) prior to transmission electron microscopy (TEM) analysis. Ten microliters of either a bacterial suspension with an optical density at 650 nm of 1 or purified FuBA (2 mg/ml) from G. amarae was placed on top of carbon-coated, glow-discharged nickel grids for 30 s. The grids were washed on 1 drop of glass-distilled water, stained with 3 drops of 1% (wt/vol) phosphotungstic acid (pH 6.9), and blotted dry. Electron microscopy was performed using a JEOL 1010 TEM at 60 keV. Images were obtained with a Sony XCD-SX900 camera. For size determination, a standard-grid nickel plate (2,160 lines/mm) was used (24).

RESULTS

Production of FuBA in aged mycolata cultures.

Conformationally specific antibodies demonstrated the presence of FuBA in cultures of a wide array of mycolata belonging to different families. Strong binding of WO2 was observed for C. glutamicum, Dietzia spp., Nocardia spp., M. avium, Tsukamurella spp., and Williamsia spp. (Table 1). In cases where sporulation was observed (Bacillus, Nocardia, and S. coelicolor), both aerial hyphae and spores were positive, showing that there was widespread distribution of amyloid.

TABLE 1.

Amyloid prevalence among gram-positive organisms: binding of WO2 to untreated and saponified cells based on immunofluorescence data

Species	WO2 binding to untreated cellsa	WO2 binding to saponified cellsa	WO2 binding locations (untreated)b
Corynebacterineae: Corynebacterium
C. flavescens	+	+++	Within a few large microcolonies
C. glutamicum	+++	+	Within all microcolonies and on single cells
Corynebacterineae: Dietzia
D. maris	+++	+	Within all microcolonies and on single cells
D. papillomatosis	+++	ND	Within all microcolonies and on single cells
Corynebacterineae: Gordonia
G. amarae	−	++
G. hydrophobica	−	−
Corynebacterineae: Nocardia
N. asteroidesc	+++	−	On mycelia and within all microcolonies, spores and aerial hyphae positive
N. polyresistensc	+++	ND	Within all microcolonies and on single cells, spores positive
Corynebacterineae: Mycobacterium
M. avium	+++	ND	Within velvet-like aggregates
M. phlei	−	−
Corynebacterineae: Tsukamurella
T. paurometabola	++	−	Cell microcolonies and large fibrous aggregates
T. spumae	+++	ND	Cell microcolonies and large fibrous aggregates
Corynebacterineae: Williamsia
W. maris	+++	ND	Within all cell microcolonies and on single cells
W. muralis	++	−	Primarily within large cell microcolonies
Non-Corynebacterineae in Actinobacteria
S. coelicolor	−	+++	Only spores positive
S. cinnabarium	+++	−	Within large mycelia and on single cells
G. obscurus	+++	ND	Within cells and huge extracellular matrices
A. acidiphila	++	ND	Primarily within large cell microcolonies
Firmicutes
E. aquimarinus	+	−	On large cellular microcolonies
B. mycoidesc	+	+++	Within cell microcolonies and on large fibrous aggregates, spores positive

^a−, no WO2 binding; +, low level of WO2 binding, and only few microcolonies of cells were positive; ++, intermediate level of WO2 binding, and there was an obvious distinction between labeled highly positive cell microcolonies and unlabeled single cells; +++, strong WO2 binding (i.e., no WO2-negative cells were observed); ND, not done.

^bWO2 DAPI counterstaining allowed assessment of the position of FuBA relative to the cells.

^cN. asteroides, N. polyresistens, and B. mycoides were grown in liquid standing cultures, and massive sporulation was observed; therefore, submerged mycelia and cells, aerial mycelia and cells, and spores were present when the cultures were labeled with WO2.

Unexpectedly, G. amarae did not bind WO2, as shown in Fig. 1A, and the level of fluorescence did not exceed that of the background for G. amarae immunostained solely with secondary antibody (results not shown). However, G. amarae did stain positive with the amyloid-diagnostic fluorescent dye ThT (results not shown). Mycolata possess a protective outer capsule consisting of lipids, polysaccharides, and proteins (15), and these molecules could block access of antibodies (but not of the small molecule ThT) to FuBA present either in the capsule of G. amarae or on the surface of the bacteria. A saponification step to remove lipids was therefore performed with G. amarae prior to WO2 labeling. G. amarae saponified at 80°C bound WO2 to a high degree (Fig. 1B), indicating that removal of lipids by saponification exposes FuBA present in the G. amarae cell envelope. Our immunochemistry data were supported by the observation of fibrillar structures by TEM (Fig. 2A to D). When saponification was performed at higher temperatures, including 37°C (Fig. 2B) and 60°C (Fig. 2C), the bacteria were gradually dissolved and fibril-like structures were observed. After saponification at 80°C, nearly all bacteria had disintegrated, leaving material with a remarkable 9-nm-wide fibril-like morphology as determined by TEM (Fig. 2D). Furthermore, TEM analysis revealed that the purified FuBA from G. amarae had a fibrous morphology (see below) and was able to bind WO2 (see below), suggesting that the purified material was indeed the WO2 binding substances embedded in the capsule of G. amarae.

FIG. 1.

Binding of WO2 to saponified G. amarae reveals the presence of FuBA. (A) WO2 immunofluorescent image of untreated G. amarae, showing the absence of WO2 binding. (B) Saponified G. amarae with a high level of WO2 binding. Bars = 10 μm.

FIG. 2.

Saponification of G. amarae at increasing temperatures reveals gradual liberation of fibril-like substances: TEM micrographs with 1% phosphotungstic acid staining of (A) nonsaponified G. amarae, (B) bacteria saponified for 4 days at 37°C, (C) bacteria saponified for 4 days at 60°C, and (D) bacteria saponified for 4 days at 80°C. Bars in panels A, B, and C represent 0.5 μm; the bar in panel D represents 0.1 μm. The arrows indicate the positions of (A) a dense extracellular matrix and (B to D) fibrillar material.

When immunolabeling was performed with saponified C. flavescens, there was a similar increase in WO2 binding compared to the binding with untreated C. flavescens. In other cases, saponification resulted in a decrease (e.g., C. glutamicum) or complete loss (e.g., N. asteroides) of WO2 binding (Table 1). This is a very good indication that saponification (which includes highly alkaline conditions, which are known to dissolve many types of protein aggregates) by itself is very unlikely to cause formation of amyloid. Altogether, 12 of 14 mycolata species belonging to seven different families bound WO2.

Production of FuBA in nonmycolata cultures.

The abundance of FuBA in mycolata prompted us to examine closely related actinomycetes. S. coelicolor is well known for assembly of chaplins into FuBA that confer hydrophobicity to submerged hyphae, allowing hyphae to grow into the air and form spores (8). This phenomenon has previously been reported to occur only in minimal media or liquid standing cultures, as also shown by the lack of WO2 labeling of S. coelicolor cultivated at 120 rpm in rich media (Fig. 3B). However, when the same S. coelicolor culture was saponified, strong WO2 binding was observed (Fig. 3D), indicating that S. coelicolor contains encapsulated amyloid, like G. amarae. Spores from S. coelicolor showed strong WO2 binding without saponification. The other nonmycolata actinobacteria S. cinnabarium, G. obscurus, and A. acidiphila were all able to bind WO2, as were the distantly related Firmicutes species E. aquimarinus and B. mycoides (Table 1). The fact that 18 of 20 gram-positive organisms examined, belonging to a wide array of species, produce FuBA indicates that FuBA is remarkably widespread among gram-positive organisms.

FIG. 3.

Binding of WO2 to saponified (but not unsaponified) S. coelicolor cultivated in rich, stirred, liquid media reveals the presence of FuBA. (A) Bright-field and (B) WO2 immunofluorescent images of untreated S. coelicolor show no WO2 binding, while (C) bright-field and (D) WO2 immunofluorescent images of saponified S. coelicolor show strong binding of WO2. Bars = 10 μm. The arrows indicate mycelia.

Morphology of the extracellular FuBA.

The appearance of extracellular fibrils binding WO2 was highly variable among the organisms. The cells were visualized by staining with the nucleic acid-binding dye DAPI. C. glutamicum produced FuBA only in cellular microcolonies (Fig. 4A and B), whereas G. obscurus also produced a massive extracellular matrix with few or no cells attached (Fig. 4C and D). WO2 labeling of M. avium revealed a peculiar velvet-like WO2-positive extracellular matrix in the vicinity of cells (Fig. 4E and F), while WO2 labeling of T. spumae exposed large cellular colonies containing fibrils more than 50 μm long (Fig. 4G to H). Immunolabeling of the two Nocardia spp. and B. mycoides showed that both the aerial hyphae and spores bound WO2. Figures 4I and J show spore formation by B. mycoides and the positive WO2 signal.

FIG. 4.

FuBA occurs in various species-specific shapes and sizes. (A, C, E, and G) Bright-field images. (B, D, F, and H) WO2 labeling (green) and DAPI counterstaining (blue). (A and B) C. glutamicum. FuBA is present around all cells. (C and D) G. obscurus. FuBA occurs in large extracellular aggregates. The arrows indicate extracellular material with a high level of amyloid but low cell density. Bars = 10 μm. (E and F) M. avium. Velvet-like substances strongly bind WO2. (G and H) T. spumae. Long (>50 μm) WO2 binding fibrils (arrows) are present. Bars = 10 μm. (I and J) Binding of WO2 to B. mycoides cells and spores. (I) Phase-contrast image of cells and spores (rings). (J) Fluorescence image of the same field, showing a high level of binding of WO2 to both cells and spores. Bars =10 μm.

Biophysical characterization of purified FuBA.

In order to examine the amyloid-like properties of FuBA, a purification protocol was developed. Two bacteria with different FuBA morphologies were chosen for detailed analysis: (i) G. amarae, which produces encapsulated FuBA that requires saponification for exposure, and (ii) G. obscurus, a nonmycolata species with remarkable production of extracellular FuBA, which is accessible to WO2 antibodies without saponification.

The optimized protocol was based on principles for purification of fungal hydrophobins (50) and chaplins from S. coelicolor (8) along with our own experiences from purification of amyloids from Escherichia and Pseudomonas species (M. Dueholm, P. Nielsen, and D. Otzen, unpublished results). Briefly, FuBA was liberated from bacteria using prolonged sonication on ice, and contaminating substances were removed using hot 2% SDS treatment followed by preparative SDS-PAGE. TEM analysis revealed that the purified FuBA from G. amarae had a fibrous morphology (Fig. 5A); furthermore, purified FuBA bound WO2 (Fig. 5B), suggesting that the purified material was indeed the WO2 binding substances embedded in the capsule of G. amarae. The TEM analysis, however, also revealed that other residual cell wall components were part of the purified FuBA. In particular, minor parts of the G. amarae capsule with extracellular material (Fig. 2A) were found interspersed between some of the fibers. Treatment with lysozyme to remove potential peptidoglycan in the sample was attempted, but this did not lead to noticeable removal of impurities (data not shown).

FIG. 5.

FuBA is amyloid-like in terms of morphology, WO2 binding, and subunit composition. (A) Phosphotungstic acid-stained electron microscopy image of purified FuBA from G. amarae, revealing the amyloid-like fibrous morphology. Bar = 0.5 μm. (B) WO2 labeling of purified FuBA from G. amarae. Bar = 10 μm. (C) SDS-PAGE of 150 μg untreated (−TFA) and TFA-extracted (+TFA) purified FuBA from G. amarae (left gel) and G. obscurus (right gel).

Nevertheless, purified FuBA bound WO2 to a high degree (Fig. 5B), suggesting that a high proportion of the purified material was indeed the WO2 binding substances embedded in the capsule of G. amarae. SDS-PAGE was performed in order to examine the composition of FuBA from the two species. However, when purified FuBA was loaded, the material was not able to enter the pores of the gel due to FuBA's large size (Fig. 5C); this factor had been used as a purification principle for the preparative SDS-PAGE system. Nevertheless, extraction with 100% TFA (50) resulted in faint protein bands at ∼14 and 8 kDa for G. amarae and at ∼11 and 5 kDa for G. obscurus, indicating the presence of protein components in FuBA. Due the proteins' low solubility, we have not been able to determine the molecular masses of these bands more accurately by mass spectrometry. However, we noted that for both species the molecular masses correspond to those of monomers and dimers, just as we observed for the E. coli FuBA CsgA (M. Dueholm et al., unpublished results) and similar to what has been reported for rodlins and chaplins (8).

The amyloid-diagnostic dye ThT was used to substantiate these indications of the presence of amyloid-like material. A remarkably great (63-fold) increase in ThT emission was observed when purified G. amarae FuBA was added to ThT (Fig. 6A, inset). The emission maximum was close to 482 nm, which is characteristic of amyloid (31). Purified FuBA from G. obscurus was black and concealed the ThT signal (as well as the CD spectrum), probably due to interference. However, 50-fold dilution of the sample resulted in a ThT fluorescence signal that was four times greater than the background value (data not shown). The CD spectrum of purified FuBA from G. amarae had a single minimum at ∼220 nm (Fig. 6A), indicating a β-sheet secondary structure in good agreement with the expectations for the cross-β amyloid fibrils. Finally, the FTIR spectra for purified FuBA from G. amarae (Fig. 6B) and G. obscurus (Fig. 6C) both contained a strong peak in the range from 1,620 to 1,630 cm⁻¹ characteristic of amyloid-like material (32, 51). We unsuccessfully attempted to sequence SDS-PAGE bands of the purified FuBA using trypsin digestion or chemical cleavage after Met, Trp, or Cys coupled with mass spectrometry or Edman degradation. This failure may reflect the small amounts of protein available and/or an unusual amino acid composition.

FIG. 6.

Amyloid-like spectral properties of purified FuBA. (A) Far-UV CD spectrum of purified and sonicated FuBA from G. amarae, revealing a typical β-sheet-rich signature. (Inset) G. amarae FuBA displays a remarkable increase in ThT emission at 482 nm compared to the background signal of ThT in PBS (dotted line). RFU, relative fluorescence units. (B and C) Amide I region of the FTIR spectra of FuBA purified from (B) G. amarae and (C) G. obscurus (solid line). The summation of Lorentz curve fits is shown (dotted line) along with individual fits (dashed lines), and the corresponding peak integrals are also indicated at the top. There is an amyloid-like β-sheet-rich peak at ∼1,625 cm⁻¹. AU, absorbance units.

DISCUSSION

Amyloid production in gram-positive bacteria.

Previously, FuBA has been detected and characterized only in the gram-negative bacterial taxa E. coli, Salmonella, and Xanthomonas and the gram-positive genus Streptomyces (7, 8, 36). However, our recent in situ immunolabeling experiments with WO2, Congo red, and ThT in natural bacterial habitats and wastewater treatment plants have revealed the widespread presence of FuBA in several other bacterial phyla (29). The report of FuBA in 18 mycolata and other gram-positive bacteria in this study substantiates these observations and shows that the ability to produce amyloid is much more common in bacteria than previously expected. The WO2 antibody has been shown to bind only to amyloid and not to other kinds of protein aggregates (36), and our previous work documented that WO2 binds to curli-producing E. coli but not to E. coli strains in which curli production has been knocked out (29, 30). In addition, we were able to purify FuBA to such an extent that only two bands (possibly a monomer and a dimer) were revealed by SDS-PAGE. We also observed a small amount of impurities (Fig. 2A) which may correspond to other cell wall components arising from the simple nature of our purification protocol and the inability of these materials to penetrate the SDS gel. However, the purified material showed ThT binding, had a fibrous morphology, and contained β-sheet-rich structures according to CD and FTIR. Therefore, we believe that it is unlikely that WO2 binds to a nonamyloid component in the 18 gram-positive strains that were positive for WO2. Only 2 of the 20 species tested did not produce FuBA under the growth conditions used. It is well known that amyloids are not expressed by all strains of E. coli and Salmonella (4, 39) and also that the growth conditions are important, particularly nutrient stress, which can promote curli production, adhesion, and biofilm formation (4, 39). Hence, it is possible that the two FuBA-negative species could produce FuBA under other growth conditions. The growth conditions in this study were optimized for G. amarae (medium, temperature, age of culture) and used for all species. The evidence that supports the widespread presence of FuBA suggests that this type of fibril is a multifunctional tool used by several bacterial species for purposes as diverse as formation of part of the cellular envelope, coating of spores, and as extracellular fibrils.

Morphology of extracellular amyloid.

We do not have any evidence that FuBA is a prerequisite for flocculation, but we suggest that FuBA affects the properties of the resulting biofilm, although other experiments will have to elucidate this. Thus, gram-positive FuBA may act in a fashion similar to that of the extracellular curli fibrils from E. coli and Salmonella, which are known to facilitate adhesion and bacterial survival by initiating biofilm formation (4, 39). Several mycolata, including G. amarae (27), have been implicated in biofilm formation and operational problems in wastewater treatment plants (2). There have been several publications describing extracellular mycolata fibrils, but until very recently these fibrils have not been linked to amyloid (9, 15, 23, 34). Scanning electron microscopy of aging M. tuberculosis, Mycobacterium smegmatis, and Mycobacterium paratuberculosis cultures has revealed the same extracellular entangled fibers that seem to associate the bacteria and lead to formation of biofilms (9, 15, 23, 34). Alteri et al. (1) showed that M. tuberculosis produces curli-like fimbriae that have crucial roles in infection. In the present work, FuBA was detected in the closely related pathogenic organisms M. avium and N. asteroides, and it therefore seems compelling that the MTP of M. tuberculosis are amyloid. Understanding the basis of adherence is the first step in combating tuberculosis, especially in light of the extremely resistant M. tuberculosis strains found recently (19).

We observed a striking amount of extracellular FuBA produced by G. obscurus, M. avium, and T. spumae. FuBA production at this level has not been described previously for bacteria. The observation of extracellular fibrils that are more than 50 μm long is very interesting and suggests that these fibrils play a central role in the development of the three-dimensional architecture of biofilms. It also suggests that extracellular self-assembly may take place, similar to that of the chaplins of S. coelicolor (8). The velvet-like pattern observed for clumps of M. avium has been described previously (14, 23, 33), where fibrillar surface substances allow cells to form cellular networks and floating biofilms. The identity of these structures is not known, but as this study shows, they may at least in part consist of amyloids, perhaps in combination with other extracellular polymers. The same type of loosely attached material has been described for the pathogen Mycobacterium lepraemurium and the obligate pathogen Mycobacterium leprae (16). Whether the large structures observed for T. spumae are similar to the so-called “honeycombs” recently observed in biofilms of Staphylococcus epidermidis (42) remains to be investigated.

Amyloid in the cell envelope.

The cell wall in many mycolata is thought to consist of an outer layer consisting of mycolic acids, lipids, proteins, and polysaccharides and an inner electron-dense cell wall core consisting of peptidoglycan and arabinogalactan (40, 45). Our results show that some species of mycolata contain amyloids in the cell envelope not accessible to WO2, and this strongly indicates a previously uncharacterized function of FuBA. However, analysis of some nonmycolata and even bacteria belonging to the distant phylum Firmicutes revealed cell envelope amyloids. Thus, the presence of amyloids seems to be a more universal property of many gram-positive bacteria, so more detailed studies are needed to reveal the exact nature of these amyloids in the cell envelope.

Fibril-like structures were visible in G. amarae samples after saponification at different temperatures, indicating that the amyloids were not produced during the treatment. TEM images revealed that FuBA fibrils that were 9 nm wide were predominantly close to or integrated into the cell wall, whereas a minor fraction was distant from the cells. Several authors have used freeze fracture electron microscopy to show that the outer capsule surrounding intraphagosomal M. avium and M. lepraemurium consists of a multilaminar structure (41, 46). Each lamella of the M. avium coat is made up of parallel straight fibrils that are 5 nm wide. This structure is very similar to that of amyloid hydrophobins on the surface of Schizophyllum commune. Perhaps FuBA in the capsule could be partially responsible for the amazing survival of specific pathogenic mycolata species inside macrophages. This would be analogous to the silk moth chorion, where a lamellar ultrastructure of packed amyloid fibrils protects the developing embryo against temperature variations, mechanical pressure, proteases, bacteria, etc. (21). If a layer of lamellar amyloid is also present in the G. amarae envelope, this could explain this bacterium's remarkable resistance to permeabilization and disruption (27).

Very interestingly, the results of this study also show that all bacterial species that formed spores under the conditions tested produced spores coated with amyloids. Spores covered by amyloids have been described for various fungi, where they are known as hydrophobins (20). This coating facilitates the dispersal of the spores by wind and enhances their attachment to surfaces and possibly also their pathogenic properties (20). Some amyloids (hydrophobins) seem to have an important role in helping fungal conidia avoid clearance by neutrophils and macrophages in the early stages of infection. The presence of amyloid-like structures on spores of Bacillus atrophaeus has also recently been observed in detailed atomic force microscopy studies (38). Spores from B. mycoides are very hydrophobic, as verified by atomic force microscopy force measurements (6), and this could be due to the presence of amyloids. Biofilm formation by Bacillus cereus also takes place primarily at the air-liquid surface, where the bacteria sporulate and may be dispersed (49). Our results indicate that spores produced by many spore-forming gram-positive bacteria also are covered by amyloids, which promote wind dispersal, surface attachment, and pathogenicity, and this may also explain the extreme resistance of the spores to environmental stresses.

S. coelicolor is assumed to produce amyloid only in connection with formation of aerial hyphae and spore formation in liquid standing or solid medium (8), but the saponification and immunofluorescence analysis revealed that the amyloid was an integrated part of the cell wall also in nonsporulating cells. Saponification is known to remove substances (especially lipids) from the outer layers of the mycolata capsule (3) and could thus also remove embedding molecules from S. coelicolor, making amyloid accessible for WO2 binding. This is in agreement with a recent atomic force microscopy study (11), which showed that fibrous material is present on the surface of S. coelicolor before the onset of aerial hypha formation. Thus, amyloid may be formed in the cell envelope prior to the formation of aerial fungal hyphae not only in S. coelicolor but also in other sporulating species, such as Nocardia and Bacillus species.