Attachment of Streptomyces coelicolor is mediated by amyloidal fimbriae that are anchored to the cell surface via cellulose


  • Wouter De Jong,

    1. Groningen Biomolecular Sciences and Biotechnology Institute (GBB), Department of Microbiology, University of Groningen, Kerklaan 30, 9751 NN Haren, the Netherlands.
    Search for more papers by this author
  • Han A. B. Wösten,

    1. Institute of Biomembranes, Department of Microbiology, University of Utrecht, Padualaan 8, 3584 CH Utrecht, the Netherlands.
    Search for more papers by this author
  • Lubbert Dijkhuizen,

    Corresponding author
    1. Groningen Biomolecular Sciences and Biotechnology Institute (GBB), Department of Microbiology, University of Groningen, Kerklaan 30, 9751 NN Haren, the Netherlands.
      *E-mail; Tel. (+31) 50 3632153; Fax (+31) 50 3632154.
    Search for more papers by this author
  • Dennis Claessen

    1. Groningen Biomolecular Sciences and Biotechnology Institute (GBB), Department of Microbiology, University of Groningen, Kerklaan 30, 9751 NN Haren, the Netherlands.
    Search for more papers by this author

*E-mail; Tel. (+31) 50 3632153; Fax (+31) 50 3632154.


The chaplin proteins ChpA-H enable the filamentous bacterium Streptomyces coelicolor to form reproductive aerial structures by assembling into surface-active amyloid-like fibrils. We here demonstrate that chaplins also mediate attachment of S. coelicolor to surfaces. Attachment coincides with the formation of fimbriae, which are connected to the cell surface via spike-shaped protrusions. Mass spectrometry, electron microscopy and Congo red treatment showed that these fimbriae are composed of bundled amyloid fibrils of chaplins. Attachment and fimbriae formation were abolished in a strain in which the chaplin genes chpA–H were inactivated. Instead, very thin fibrils emerged from the spike-shaped protrusions in this mutant. These fibrils were susceptible to cellulase treatment. This enzymatic treatment also released wild-type fimbriae from the cell surface, thereby abolishing attachment. The reduced attachment of a strain in which the gene of a predicted cellulose synthase was inactivated also indicates a role of cellulose in surface attachment. We propose that the mechanism of attachment via cellulose-anchored amyloidal fimbriae is widespread in bacteria and may function in initiation of infection and in formation of biofilms.


Within the bacterial domain streptomycetes are well known for their complex developmental programme. Streptomycetes grow by means of hyphae that extend at their apices. This feature makes these soil bacteria particularly successful in colonizing and degrading organic material. After a feeding substrate mycelium has been formed, hyphae grow out of the substrate into the air to form chains of hydrophobic spores. These spores are dispersed by wind or insects enabling this organism to colonize a substrate elsewhere.

Formation of the aerial reproductive structures in Streptomyces coelicolor has been studied for several decades (Kelemen and Buttner, 1998; Chater and Horinouchi, 2003; Claessen et al., 2006; Flärdh and Buttner, 2009; de Jong et al., 2009). This has resulted in the identification of a variety of so-called bld (bald) genes, the deletion of which leads to an arrest in morphogenesis. Most of the bld genes encode proteins with an apparent regulatory role, implying that this developmental switch is subject to extensive regulation (Kelemen and Buttner, 1998; Flärdh and Buttner, 2009). For instance, bldA encodes a tRNA required for efficient translation of the rare leucine TTA codon (Leskiw et al., 1991), whereas bldN encodes an extracytoplasmic function sigma factor (Bibb et al., 2000).

Several secreted macromolecules play a pivotal role in development of S. coelicolor. The cslA gene, encoding a cellulose synthase-like protein, was shown to be essential for aerial growth (Xu et al., 2008). CslASC interacts with the DivIVA protein at hyphal tips, where it synthesizes a β-(1-4) glucan. This secreted polysaccharide is thought to maintain the integrity of the hyphal tip that is subject to constant remodelling due to ongoing cell wall synthesis orchestrated by DivIVA (Flärdh, 2003; Xu et al., 2008). SapB is another macromolecule that is secreted during development (Willey et al., 1991). This lantibiotic-like peptide (Kodani et al., 2004) lowers the surface tension of the aqueous environment to enable hyphae to grow into the air (Tillotson et al., 1998).

Like SapB, chaplins are secreted surface-active molecules. This class of proteins comprises eight members (Claessen et al., 2003; Elliot et al., 2003). The chaplins ChpD–H are about 55 amino acids in length, whereas ChpA–C are approximately fourfold larger. ChpA–C consist of two domains similar to the mature forms of ChpD–H followed by a C-terminal sorting signal that anchors them to the peptidoglycan in the cell wall (Elliot et al., 2003; Marraffini et al., 2006). Two of the small chaplins, ChpE and ChpH, are produced by submerged hyphae and have a function similar to SapB. They enable hyphae to grow into the air by lowering the surface tension of the medium (Claessen et al., 2003). These chaplins have another role as well. Together with the other six chaplins they are secreted in the cell walls of aerial hyphae where they provide rigidity and surface hydrophobicity (Claessen et al., 2003; 2004). Chaplins carry out these functions by assembling into amyloid fibrils, which, at least on the surface of aerial hyphae, are organized into a pattern of pair-wise aligned rodlets (Claessen et al., 2003; 2004). This process seems to be co-ordinated by the rodlins RdlA and RdlB, which are produced and secreted by growing aerial hyphae (Claessen et al., 2002). Aerial growth is severely impaired in the absence of chaplins, stressing the important biological function of this class of amyloid proteins during development (Claessen et al., 2004; Capstick et al., 2007; Di Berardo et al., 2008).

Fungi also use amyloid-forming proteins that reduce the surface tension of the aqueous substrate and provide aerial structures with a hydrophobic coating. In many fungi, hydrophobins form these fibrils (Wösten and de Vocht, 2000; Gebbink et al., 2005). However in the maize pathogen, Ustilago maydis, these proteins have been functionally replaced by repellents (Wösten et al., 1996; Teertstra et al., 2006; 2009). Repellents and hydrophobins have also been shown to attach hyphae to a hydrophobic solid (Wösten et al., 1994; Teertstra et al., 2006), which is essential for pathogenicity (Talbot et al., 1996) and possibly also for the degradation of organic substrates (Wösten, 2001).

Here, we show that amyloid fibrils of the chaplins are involved in surface attachment of S. coelicolor. In contrast to the fungal proteins, the amyloid fibrils of the chaplins are organized into attachment structures known as fimbriae. These fimbriae are anchored to the cell wall via cellulose, which provides an important new insight in the role of this polysaccharide in the bacterial domain.


Chaplins are required for attachment and the formation of Streptomyces fimbriae

Previously, we have shown that hyphae of liquid static cultures of streptomycetes attach to polystyrene when grown in gNMMP or mNMMP (Claessen et al., 2002; van Keulen et al., 2003). Attachment is much stronger in mNMMP compared with gNMMP (D. Claessen and G. van Keulen, unpubl. data). Negative staining showed that S. coelicolor formed an extracellular matrix when grown in mNMMP (Fig. 1A), which was absent in gNMMP (Fig. S1). The matrix consisted of 9- to 100-nm-wide fimbriae, which were present throughout culturing (up to at least 15 days). They were associated with the adhering hyphae via spike-shaped protrusions (see arrows in Fig. 1B). Formation of fimbriae did not require the activity of the rodlin proteins, as the ΔrdlAB mutant was shown to make an extracellular matrix indistinguishable from that in the wild-type strain (data not shown). Moreover, attachment of the ΔrdlAB mutant was not affected in mNMMP (data not shown). Formation of fimbriae did depend on the chaplins as concluded from the absence of the typical 9- to 100-nm-wide fibres in the ΔchpABCDEFGH strain (Fig. 1C). Instead, fibrils with a diameter of 9 ± 2 nm emerged from the spike-shaped protrusions on the cell surface (see arrows in Fig. 1D). These thin fibrils resembled those detected on the cell surface when the wild-type strain was grown in gNMMP liquid standing cultures (Fig. S1). Importantly, attachment was greatly reduced in the chp-less mutant (Fig. 1E, Table 1). Compared with the ΔchpABCDEFGH strain, attachment was only partially decreased in the ΔchpABCDH and ΔchpABCDEH strains (Table 1). Both mutants produced fimbriae that were indistinguishable from those produced by the wild-type strain (not shown), although reduced in number (Fig. S2).

Figure 1.

Formation of fimbriae depends on chaplins.
A. Attachment of S. coelicolor in mNMMP medium coincides with the formation of fimbriae in between the adhering hyphae, which are composed of bundled filaments (inlay).
B. Fimbriae emerge from the cell wall of the wild-type strain via spike-shaped protrusions (arrows). In contrast, the ΔchpABCDEFGH strain forms thin fibrils (indicated by arrows in C) that protrude from these structures (see arrows in D).
E. Quantification of biomass of the wild-type (filled circles) and the ΔchpABCDEFGH mutant (open circles) that has attached to the well. The biomass was stained with crystal violet and related to staining of a fixed amount of mycelium from a liquid-shaken culture. The inlay shows crystal violet stained mycelium of a 7-day-old culture of the wild-type strain (i) and the ΔchpABCDEFGH mutant (ii) that resisted washing with water. The scale bar represents 5 μm (A, C), 1.25 μm (B, D) and 100 nm (inlay).

Table 1.  Percentage of the mycelium of Streptomyces strains that is attached to the well of the culture plate.
StrainAttachment (%)
  1. Cultures were grown for 12 days in liquid static cultures.

M14589.3 ± 3.9
ΔchpABCDH67.8 ± 1.2
ΔchpABCDEH65.2 ± 3.6
ΔchpABCDEFGH30.7 ± 13.2
bldN1.5 ± 0.9
cslA (Tn5062)56.0 ± 16.4

Expression of the chaplin genes requires the extracytoplasmic function sigma factor BldN (Elliot et al., 2003). In support of a role for chaplins in attachment, we observed that a bldN mutant was severely affected in adhesion (Table 1). Moreover, the extracellular matrix produced by the bldN mutant resembled that of the chaplin mutant strain (data not shown). Taken together, these results show that formation of the extracellular matrix depends on chaplins and correlates with the capacity of hyphae to attach firmly to the hydrophobic substratum.

Assembly of chaplins into amyloid fibrils on hydrophobic surfaces

The absence of fimbriae in the ΔchpABCDEFGH mutant strain prompted us to investigate whether chaplins are part of the extracellular matrix. MALDI-TOF mass spectrometry on intact fimbriae (see Experimental procedures) revealed masses corresponding to the mature forms of ChpD, ChpE, ChpF and ChpH (Fig. 2). This shows that fimbriae are, at least in part, composed of chaplins.

Figure 2.

Identification of chaplins in fimbriae by MALDI-TOF mass spectrometry. Peaks corresponding to the masses of ChpD, ChpE, ChpF and ChpH (Claessen et al., 2003) are detected in the fimbrial network (see inlay). The scale bar represents 1 μm.

Circular dichroism (CD) was used to study structural changes of chaplins in contact with a hydrophobic surface. Previously, it was shown that purified chaplins (ChpD–H) are unstructured when dissolved in water; however, at a water–air interface, these proteins self-assemble into amyloid fibrils, which is accompanied by the formation of β-sheet structure (Claessen et al., 2003). The structure of water-soluble chaplins (Fig. 3A, dashed line) also changed rapidly upon adding an excess of colloidal Teflon (Fig. 3A, thin solid line). The CD spectrum indicated formation of α-helix (Chang et al., 1978). The conversion towards the α-helical state did not increase the fluorescence of the amyloid specific dye Thioflavin T (ThT; Table 2). The chaplins pelleted together with the Teflon spheres upon centrifugation, indicating that the chaplins were bound to the Teflon (not shown). When the mixture of chaplins associated with the Teflon spheres was heated in the presence of 0.1% Tween-20, thereby promoting lateral interactions between the bound chaplins, the spectrum became indicative for β-sheet structure (Fig. 3A, thick solid line; Sreerama et al., 1999). This treatment increased ThT fluorescence 36-fold (Table 2). This increase was similar to that obtained after vortexing a chaplin solution and showed that all monomers had assembled into amyloid fibrils (Table 2). In the absence of Teflon spheres, a heated solution of chaplin monomers with 0.1% Tween-20 did not increase ThT fluorescence (Table 2). These data show that chaplins can assemble into amyloids on a hydrophobic surface.

Figure 3.

A. Assembly of chaplins on a hydrophobic surface as determined by CD spectroscopy. Spectrum of ChpD–H before (dashed line) and after (solid line) addition of colloidal Teflon and subsequent treatment with 0.1% Tween-20 at 85°C (thick solid line).
B. Assembly of monomeric chaplins in the presence of the assembled form. Mixtures of chaplins (ChpD–H) and rodlins (RdlA + RdlB) were incubated for 60 min with increasing amounts of seeding chaplin fibrils. Note that the rodlins, serving as a loading control, remain soluble.

Table 2.  ThT (3 μM) fluorescence upon interaction with chaplins (14 μg ml−1) in different conformations.
Conformation of chaplinRelative ThT fluorescence
  1. Fluorescence of ThT in the absence of protein was set at 1 and data were corrected for autofluoresence of the Teflon spheres.

Water soluble4.6 (± 0.47)
Water soluble (5 min after addition of 0.1% Tween-20 at 85°C)5.2 (± 0.58)
α-Helical conformation on Teflon0.59 (± 0.069)
β-Sheet conformation on Teflon (5 min after addition of 0.1% Tween-20 at 85°C)166 (± 2.1)
β-Sheet conformation induced by vortexing159 (± 5.7)

Nucleation-driven assembly of chaplins

Amyloid fibrils of ChpD–H (up to 1.5 μg in 25 μl water) were added to 50 μl of a still aqueous solution of their monomers (50 μg ml−1). SDS-PAGE showed that the amount of monomeric chaplins hardly decreased during 1 h incubation when no amyloid fibrils were added (Fig. 3B). In contrast, more and more monomeric chaplin disappeared from solution upon addition of increasing amounts of the assembled form. This coincided with increase in ThT fluorescence showing that the soluble chaplins had assembled into amyloid fibrils (data not shown). Thus, assembly of chaplins becomes independent from a hydrophilic–hydrophobic interface once a nucleus of assembled chaplins is present.

Inhibition of attachment, fimbriae formation and chaplin assembly by Congo red

The capacity of chaplins to assemble into amyloid-like fibrils as well as their involvement in the formation of fimbriae were reason to study the effects of the amyloid inhibitor Congo red (CR) (Findeis, 2000; Kuner et al., 2000). S. coelicolor was grown on agar plates in the presence of increasing amounts of CR to assess whether this compound is toxic at high concentrations, as was shown in Acetobacter xylinum (Colvin and Witter, 1983). Neither growth nor differentiation was significantly affected on solid MS agar in the presence of up to 200 μg ml−1 CR (data not shown). Similarly, no effects were observed in liquid shaken cultures at these concentrations (data not shown). However, CR did affect attachment in liquid static cultures. Attachment was already largely abolished at 5 μg ml−1 CR (Fig. 4A and B), which correlated with the absence of fimbriae (Fig. 4C). Instead, thin fibrils with a diameter of about 9 nm were extruded from the spike-shaped protrusions. These fibrils were very similar to those formed in the ΔchpABCDEFGH strain (Fig. 1C) and the bldN mutant (data not shown). The addition of 5 μg ml−1 CR had no effect on the formation of the thin fibrils in the bldN and chpABCDEFGH mutant (data not shown). Notably, the structure of neither the fimbriae nor the thin fibrils was affected when CR was added after their formation (i.e. after 5 days of growth). This indicates that CR does not cause their depolymerization. Accordingly, no decrease in attachment was observed when 5 μg ml−1 CR was added to the medium after the formation of fimbriae (Fig. S3). Taken together, these results demonstrate that CR specifically interferes with formation of fimbriae, thereby affecting attachment.

Figure 4.

The effect of Congo red on attachment, fimbriae formation and chaplin assembly.
A. Addition of increasing amounts of Congo red to static liquid cultures decreases attachment (bottom panel) without affecting growth (top panel).
B. Quantitative assessment of the effect of Congo red (inlay) on attachment.
C. Formation of fimbriae (left) is abolished by the addition of 5 μg ml−1 Congo red (right panel). Instead, thin fibrils emerge from the spike-shaped protrusions (see arrows in inlay, right panel).
D. The assembly of monomeric chaplin induced by seeding with 1 μg of assembled chaplin could be prevented by the addition of increasing amounts of Congo red. Bar represents 2.5 μm or 125 nm (inlay). Error bars represent the standard error of a biological triplicate.

Because chaplins are able to assemble into amyloids when contacting a hydrophobic surface and are part of the fimbriae, we studied the effect of CR on the assembly of chaplins in vitro. Importantly, the assembly process was prevented by the addition of increasing amounts of CR to the solution (Fig. 4D). This shows that this compound can directly interfere with chaplin amyloid formation.

The putative cellulose synthase CslASC is involved in attachment

Previously it was shown that the extracellular matrix produced by several enteric bacteria comprise cellulose, which functions in attachment and biofilm formation (Zogaj et al., 2001; 2003). The operons known to be involved in cellulose biosynthesis in other bacteria are organized in a similar way to a S. coelicolor gene cluster harbouring a gene encoding a cellulose synthase-like protein, called CslASC (Zogaj et al., 2001; Xu et al., 2008). The cslA gene was inactivated by a transposon (see Experimental procedures) that includes a promoter-less eGFP gene. This enables expression analysis of the disrupted gene. Strong expression of eGFP was observed in mycelium that had attached (Fig. 5A), while eGFP expression was weak in mycelium of a shaken culture (data not shown). These data suggest that CslASC plays a role in attachment. Indeed, attachment of the transposon mutant strain was approximately 50% reduced (Fig. 5B, Table 1) compared with the wild-type strain. Similar results were obtained with the XE mutant strain (Fig. 5B and data not shown). This strain has an internal fragment of the cslA gene replaced with a hygromycin-resistance cassette (Xu et al., 2008). Notably, electron microscopy analysis showed that both cslA mutants formed fimbriae that were indistinguishable from and similar in number to those produced by the wild-type strain (Fig. 5C; Fig. S2). This implies that CslASC is not essential for formation of fimbriae.

Figure 5.

The role of cellulose in attachment of S. coelicolor.
A. Expression of the cslA gene in adhering hyphae, the wild-type (left) serving as a control. Light microscopy images are shown at the top, whereas GFP fluorescence is shown at the bottom.
B. Disruption (cslA::Tn5062) or replacement (XE) of cslA reduces attachment.
C. Formation of fimbriae in the cslA (Tn5062) mutant.
D. Calcofluor white stained mycelium (left) of the wild-type (top) and the cslA (Tn5062) mutant strain (bottom). Corresponding bright-field images are shown at the right. Note the bright foci at the edge of wild-type hyphae that are absent in the cslA mutant (arrows).
E. Quantitative and qualitative (inlay) effect of cellulase on attachment. Increasing amounts of cellulase results in detachment.
F. Cellulase interferes with anchoring of fimbriae to the hyphae. The scale bars represent 100 μm (A), 12.5 μm (C), 5 μm (D), and 15 μm (F). Error bars represent the standard error of a biological triplicate.

CslASC-dependent and -independent β-(1-4) glucan formation during attachment

The S. coelicolor cellulose synthase-like protein produces a polysaccharide at hyphal tips, which can be visualized with calcofluor white (Xu et al., 2008). This fluorescent dye specifically stains β-(1-4)-coupled glucans such as cellulose and chitin. Calcofluor white staining of attached S. coelicolor wild-type hyphae revealed the accumulation of β-(1-4) coupled glucans at hyphal tips, consistent with previous results (Xu et al., 2008). However, we also detected bright fluorescent spots at considerable distances from hyphal tips and emerging branches, invariably localized at the outer surface of the adhering hyphae (see arrows in Fig. 5D, top panels). These spots were CslASC-dependent, as they were not observed in stained mycelium of the cslA mutant (Fig. 5D, bottom panels). The wild-type and the cslA mutant strains also showed a weak but detectable staining of the fimbrial network. The stain sometimes extended from the fimbriae to the cell wall of a connecting hypha (Fig. 5D). Taken together, these results demonstrate that during attachment CslASC is not only active at hyphal tips but also subapically. Here it synthesizes a β-(1-4)-coupled glucan involved in attachment. The results also infer that a related glucan that is part of the fimbriae can be synthesized in a CslASC-independent manner.

Cellulose-mediated anchoring of fimbriae

The homology of CslASC to known cellulose synthase proteins (Xu et al., 2008) and the apparent CslASC-independent formation of a β-(1-4)-coupled glucan suggested potential redundancy in biosynthesis of cellulose-like polysaccharides in S. coelicolor. To circumvent redundancy, we decided to study attachment of the wild-type strain in the presence of the cellulose-degrading enzyme cellulase. Strikingly, attachment decreased with increasing amounts of cellulase, and was completely abolished when 1 U of cellulase was added to the well (Fig. 5E). Loss of attachment was also observed when the enzyme was added 9 days after the mycelium had started to attach to the surface (data not shown).

To understand the mechanism by which cellulase inhibited attachment, mycelium grown in the presence of this enzyme was analysed with TEM. Strikingly, cellulase had no discernable effect on formation of the fimbriae. However, the connection of the fimbriae to the mycelium was affected. Fimbriae were abundantly detected in the supernatant dissociated from the surrounding mycelium (Fig. 5F). Notably, hyphae grown in the presence of cellulase still formed the spike-shaped protrusions, indicating that cellulose is not required for their formation. Taken together, these results show that cellulose has a role in anchoring fimbriae to the adhering hyphae.

Extrusion of cellulose at spike-shaped protrusions

The importance of cellulose in attachment, and the identification of thin fibrils emerging from the cell wall-associated protrusions in strains that were no longer able to synthesize or assemble chaplins (see above), were reason to analyse whether these thin fibrils consist of cellulose. We therefore analysed what the effect of the addition of cellulase was on formation of these fibrils in the bldN mutant (Fig. 6). Strikingly, the thin fibrils were no longer detected in the presence of cellulase, whereas 5 μg ml−1 CR had no effect. Identical results were obtained with the thin fibrils formed by the ΔchpABCDEFGH strain and wild-type strain in the presence of 5 μg ml−1 CR (data not shown). These results show that the spike-shaped protrusions are sites for cellulose biosynthesis, serving as an anchoring platform for the fimbriae.

Figure 6.

The thin fibrils formed by the bldN mutant consist of cellulose. (A) The bldN mutant forms thin fibrils emerging from spike-shaped protrusions (arrow heads) whose formation is not affected by low levels of CR (B). In the presence of CR and cellulase these fibrils are no longer formed (C). Note the material that is expelled from the spike-shaped protrusions (inlay). The scale bar represents 5 μm (A–C) or 2 μm (inlay). Arrowheads point to the spike-shaped protrusions.


Chaplins of the filamentous bacterium S. coelicolor were identified as proteins that function in the formation of spore-forming aerial hyphae (Claessen et al., 2003; Elliot et al., 2003). Chaplins self-assemble into surface-active amyloid fibrils that enable hyphae to escape the aqueous environment to grow into the air and that provide aerial hyphae and spores with a hydrophobic coating (Claessen et al., 2003). The amyloid fibrils of the chaplins are organized by the rodlin proteins into pair-wise aligned rodlets at the surface of the aerial structures (Claessen et al., 2002; 2003; 2004; Di Berardo et al., 2008). Here, we show that the amyloid fibrils of chaplins also function in attachment of hyphae to a hydrophobic surface. To this end, these fibrils are organized into fimbriae that are anchored to the cell wall via cellulose.

Formation of amyloidogenic pili

Several pathogenic and non-pathogenic microorganisms have been shown to produce amyloidogenic fibrils that function in adhesion to abiotic or biotic substrates, such as the fibrils of the SC3 hydrophobin of the filamentous fungus Schizophillum commune (Wösten et al., 1994), repellents of the phytopathogenic fungus U. maydis (Teertstra et al., 2006; 2009), curli and tafi of various members of the Enterobacteriaceae (Chapman et al., 2002; Zogaj et al., 2003) and the recently identified Mycobacterium tuberculosis pili (Alteri et al., 2007). In addition, amyloid adhesins were shown to be abundantly present in various types of natural biofilms (Larsen et al., 2007; 2008), indicating that they are important components of the extracellular matrix of these bacterial communities. However, almost nothing is known about the mechanism of amyloidogenic fimbriae formation in Gram-positive bacteria. In fact, details about the formation of Gram-positive pili in general have only recently started to emerge (Telford et al., 2006; Mandlik et al., 2008; Proft and Baker, 2009). Pili of Gram-negative bacteria are typically formed by non-covalent interactions between pilin subunits. In contrast, pilins of Gram-positives are covalently polymerized by the activity of sortase enzymes. Their genes are typically present near those of the pilin substrate (Mandlik et al., 2008). Sortases recognize pilins by their C-terminal cell wall sorting signal, mostly containing a conserved LPXTG motif followed by a hydrophobic domain and a positively charged tail (Marraffini et al., 2006). Strikingly, the three large chaplin variants, ChpA–C, also have a sortase recognition sequence (although with a variation of the LPXTG motif, namely LAXTG), but are not located close to any of the seven sortase homologues on the S. coelicolor chromosome (Pallen et al., 2001). Our work demonstrates that these three chaplins are not essential for formation of the fimbriae as these structures were still produced in the ΔchpABCDH mutant strain. This shows that the S. coelicolor fimbriae are atypical Gram-positive pili that resemble the M. tuberculosis pili, which are also polymerized in a sortase-independent manner (Alteri et al., 2007). However, co-assembly of the small and large chaplins and the covalent coupling of ChpA–C to the cell wall could contribute, in addition to cellulose (see below), to anchoring of the fimbriae to the cell surface (see Fig. 7).

Figure 7.

The role of chaplins and cellulose in fimbriae formation in S. coelicolor. CslA is involved in the extrusion of cellulose fibrils from the spike-shaped protrusions along the cell wall of adhering hyphae. Chaplin monomers are secreted and assemble into amyloids when contacting either assembled chaplin fibrils or a hydrophobic surface. In addition, assembly may be triggered by the activity of a (unknown) nucleator. The assembled chaplin amyloids interact with the cellulose fibrils and each other, leading to fimbriae formation. Possibly, the large chaplins (ChpA–C) may contribute to the covalent coupling of fimbriae to the cell wall.

The best-studied amyloidogenic pili are curli. Escherichia coli has a specific nucleation-precipitation machinery for the assembly of curli, which might help to prevent self-assembly of monomers inside the cell, and accelerate polymerization of curli on the cell surface in vivo. The major constituent of curli fibrils is the CsgA protein, which in vitro is capable of self-assembling into fibrils indistinguishable from those observed on surfaces of wild-type E. coli cells. However, CsgB and CsgF are involved in nucleation of CsgA in vivo (Hammar et al., 1996; Chapman et al., 2002; Hammer et al., 2007; Nenninger et al., 2009). How is the assembly process of chaplins initiated? Although we cannot exclude the involvement of a nucleator (Fig. 7), as observed for curli, chaplins can assemble without one. When mixtures of small chaplins were dried down on a surface, a regular pattern of fibrils was formed (Claessen et al., 2003). Under these conditions, the protein concentration will increase over time due to evaporation of the solvent, leading to self-assembly, as observed for other amyloid-forming proteins (Wösten et al., 1993). A similar process can happen when chaplins bind to a hydrophobic surface. A certain amount of bound monomers could result in the formation of an amyloid nucleus, which then induces other chaplins to adopt the amyloidal end-state.

When chaplins were incubated with Teflon spheres, they initially did not form amyloid fibrils. Instead, the protein adopted a conformation rich in α-helix. This so called α-helical state probably represents an intermediate of the assembly process analogous to that observed for other amyloid-forming proteins (de Vocht et al., 1998; 2002; Giacomelli and Norde, 2003). This intermediate seems to proceed to the amyloid state when the local concentration of protein is increased as was recently shown for the SC3 hydrophobin (K. Scholtmeijer and H.A.B. Wösten, unpublished). Alternatively, the conversion to the amyloid state is obtained by promoting lateral interactions between the bound protein (e.g. chaplins and hydrophobin) by treating the coated surface with diluted detergent at high temperature. It may also be that cellulose induces the conversion to the amyloid conformation (see below).

A novel role for cellulose in fimbrial anchoring and attachment

Various bacteria produce cellulose during formation of biofilms (Zogaj et al., 2001; Römling, 2002), and adherence to plant tissues (Matthysse et al., 1981; Matthysse and McMahan, 1998). For instance, the extracellular matrix produced by Salmonella enterica comprises, in addition to curli fimbriae, cellulose and one or more other polysaccharides (White et al., 2003). Cellulose was shown to be tightly associated with the curliated structures on the cell surface, but was not required for their formation. These data are consistent with our results in S. coelicolor. In addition, we observed detachment of the fimbriae from the cell surface by enzymatic treatment with cellulase, revealing an important novel role for cellulose in fimbrial anchoring (Fig. 7). Cellulose fibrils were shown to emerge from spike-shaped protrusions along the hyphal cell wall in the absence of chaplins, indicating that their formation and anchoring occur at these sites. A role for CslASC in this process is envisaged as this protein was shown to polymerise a β-(1-4) glucan along the periphery of adhering hyphae. However, other, yet unknown proteins also contribute to this process as fimbriae of the cslA mutant still stained with calcofluor white.

The observed spike-shaped protrusions are morphologically reminiscent of cellulosomes that are present on the cell surface of anaerobic cellulolytic bacteria such as Clostridium thermocellum and Ruminococcus albus (Felix and Ljungdahl, 1993). Cellulosomes are large, multi-component complexes consisting of tens of polypeptides, which are responsible for the binding to, and hydrolysis of cellulose (Felix and Ljungdahl, 1993). It is tempting to speculate that the Streptomyces protrusions are cellulosome variants designated to synthesize, anchor and expel cellulose. Anchoring of cellulose could be mediated by one of the many cellulose-binding proteins encoded in the S. coelicolor genome, such as CbpC (Walter and Schrempf, 2008), or homologues of AbpS of Streptomyces reticuli (Walter et al., 1998). In this respect, it is interesting to note that CbpC is a designated sortase substrate.

The identification of cellulose fibrils in the absence of chaplins indicates that cellulose acts as a scaffold for the bundling of chaplin amyloid fibrils into fimbriae (Fig. 7). This implies that cellulose has a high affinity for assembled amyloids. However, cellulose could also directly contribute to the formation of chaplin amyloid fibrils, not only during attachment, but also during formation of aerial hyphae. A role for glycans in inducing amyloid formation has been observed with disease-associated amyloids, where proteoglycans and heparin-like polymers promote amyloid fibrillogenesis (van Horssen et al., 2003; Bellotti and Chiti, 2008). The formation of cellulose at the periphery of adhering hyphae and at the hyphal tip during aerial growth could thus contribute to the co-ordination in space and time of the chaplin assembly process.

Experimental procedures

Strains, plasmids and culture conditions

The E. coli and Streptomyces strains used in this study are shown in Table 3. E. coli was grown at 37°C in LB medium with or without antibiotics. Streptomyces strains were grown at 30°C on R5 or MS agar medium (Kieser et al., 2000) or in liquid YEME (Kieser et al., 2000), mNMMP or gNMMP medium (van Keulen et al., 2003). Briefly, mNMMP and gNMMP are minimal media containing either mannitol (25 mM) or glucose (25 mM) and casamino acids (0.25%) as the carbon source respectively.

Table 3.  Strains used in this study.
Strain or plasmidDescriptionReference or source
  1. S. coelicolor-derived rodlin and chaplin mutant strains were created in the M145 genetic background.

S. coelicolor strains
 M145Wild-type SCP1- SCP2-Kieser et al. (2000)
 ΔrdlABrdlAB::aac(3)IVClaessen et al. (2004)
 ΔchpABCDHchpAD::aac(3)IV chpB::vph chpCH::aadAClaessen et al. (2003)
 ΔchpABCDEHchpAD::scar chpB::vph chpCH::aadA chpE::aac(3)IVClaessen et al. (2003)
 ΔchpABCDEFGHchpAD::scar chpB::vph chpCH::aadA chpE::scar chpF::scar chpG::aac(3)IVClaessen et al. (2004)
 J2177bldN::hygBibb et al. (2000)
 cslA (Tn5062)M145 sco2836::Tn5062This work
 XEM145 cslASC::hygXu et al. (2008)
E. coli strains
 DH5αF-Φ80lacZΔM15 Δ(lacZYA-argF)U169 recA1 endA1 hsdR17(rK-, mK-)
phoA supE44 thi-1 gyrA96 relA1λ-
Hanahan (1983)
 ET12567F-dam-13::Tn9 dcm-6 hsdM hsdR recF143 zjj-202::Tn10 galK2 galT22
ara14 lacY1 xyl-5 leuB6 thi-1 tonA31 rpsL136 hisG4 tsx78 mtl-1 glnV44
MacNeil et al. (1992)
 BW25113ΔaraBADAH33ΔrhaBADLD78 lacIqrrnB3ΔlacZ4787 hsdR514 rph-1Datsenko and Wanner (2000)

For growth in static liquid cultures, mycelium was harvested from 1 ml of a 2-day-old YEME liquid shaken culture in early stationary phase. The dispersed mycelium was washed twice with 0.1 M NaH2PO4/K2HPO4 buffer (pH 6.8) and taken up in 1 ml of the same buffer. This suspension was diluted (1:1000) in fresh gNMMP or mNMMP medium and subsequently used to fill 25-well flat-bottomed polystyrene plates (Greiner Bio-One) with 4 ml per well.

Attachment assay

Attachment of hyphae in static liquid cultures in 25-well plates was quantified as follows: 100 μl crystal violet solution (0.5%; Acros Organics) was added to each well after 2–10 days of growth and left at room temperature for 10 min. For gNMMP cultures, non-adherent cells were removed by washing three times with 5 ml water using a 25 ml glass pipette attached to a motorized pipettor. For mNMMP cultures, plates were vigorously washed with running tap water (tap diameter 1.7 cm; distance to the tap 40 cm; water flow 9 l min−1). After drying at 50°C, crystal violet associated with the attached biomass was solubilized in 5 ml 10% SDS on a platform rocking shaker for 30 min. The OD570 of 200 μl aliquots was determined in a microtiter plate reader. If necessary, dilutions were made in 10% SDS. For total biomass quantification, mycelium from single wells was collected and 100 μl crystal violet solution was added. After 30 min, the mycelium was centrifuged for 10 min at 4300 g and washed three times with 15 ml water. After drying the mycelium, crystal violet was solubilized with 5 ml 10% SDS and processed as described above. The relative attachment values were calculated by dividing the value of the OD570 of the attached mycelium by the value of the OD570 of the total biomass.

Construction of the cslA (Tn5062) mutant strain

Cosmid StE20 carrying the Tn5062 transposon in the SCO2836 gene (Bishop et al., 2004) was transferred by conjugation to S. coelicolor M145 to disrupt the cslA gene. Aparamycin-resistant ex-conjugants were screened for kanamycin sensitivity. Inactivation of the cslA gene was verified by PCR and Southern analyses (data not shown).

Isolation of fimbriae

The S. coelicolor wild-type strain was grown in 4 ml mNMMP medium supplemented with 0.25 U ml−1 cellulase from Aspergillus niger (Sigma-Aldrich). After 10 days of growth, the culture was mixed by pipetting and allowed to stand for 10 min. Two millilitres of the supernatant, containing the fimbriae, was centrifuged at 10 700 g for 10 min. The pellet was washed twice with water and taken up in 50 μl water.

Purification of chaplins from S. coelicolor

Chaplins were extracted with trifluoroacetic acid (TFA) from SDS-treated cell walls of sporulating cultures of the S. coelicolorΔrdlAB strain (Claessen et al., 2003), as described (Wösten et al., 1993; Claessen et al., 2002). TFA extracts were taken up in water (50–200 μg ml−1) and, if necessary, adjusted to pH 7 with diluted ammonia.

Gel electrophoresis

SDS-PAGE was done in 16% gels as described (Laemmli, 1970). Pre-stained broad range molecular weight markers of Fermentas were used. After separation, proteins were stained with the Bio-Rad Silver Stain Plus kit, according to instructions of the manufacturer.

Maldi-TOF mass spectrometry of fimbriae

Purified fimbriae were analysed with an Axima Performance Maldi-TOF mass spectrometer (Shimadzu Biotech) using a sinapinic acid matrix that had been dissolved in a mixture of 40% acetonitril/0.1% TFA.

Fluorescence microscopy

Fluorescence of GFP and calcofluor white was monitored with a Zeiss Axioskop 50 wide-field fluorescence microscope. All images in a given figure were taken on the same day with the same excitation and camera gain. In case of GFP, samples were analysed using a 470/40 nm bandpass filter, with a 495 nm beamsplitter and a 525/50 nm emission bandpass filter, while for calcofluor white stained samples, a 365/12 nm excitation filter, with a 395 nm beamsplitter and a 397 nm long-pass filter was used. For calcofluor white staining, adhering colonies were carefully removed from the polystyrene microtiter plate with a pipet, mounted on an agarose-covered glass slide, and stained for 5 min with a 0.1% (w/w) solution of the dye (Sigma-Aldrich).

Electron microscopy

For negative staining, mycelium was transferred to Formvar-coated nickel grids. After extensive washing with water, staining was done for 2 s with 2% uranyl acetate. Samples were analysed with a Philips CM12 transmission electron microscope, connected to a MegaView III CCD camera (Soft Imaging System).

Quantification of fimbriae

To quantify fimbriae, random EM images were taken (see below). Positions of the fimbriae were indicated on transparencies, which were subsequently digitized using a Canon 8800F scanner. Scanned images were loaded into ImageJ (1.42). Average pixel densities (from 6 to 10 scanned images) were determined and used as a measure for the number of fimbriae produced by each strain.

Circular dichroism

The CD spectra were recorded over the wavelength region 190–250 nm on an Aviv 62A DS Circular dichroism spectrometer, using a 5 mm quartz cuvette. The temperature was kept constant at 25°C and the sample compartment was flushed with a continuous stream of N2. Spectra represent the average of three scans using a bandwidth of 1 nm, a step width of 1 nm and a 5 s averaging per point. The spectra were corrected by using a reference solution without the protein. Typically, a protein concentration of 50–200 μg ml−1 was used.

To determine the secondary structure of chaplins interacting with a hydrophobic support, an amount of colloidal Teflon was added such that the protein in solution could cover 10% of the surface of the solid (de Vocht et al., 1998). Spectra were taken before and after heating at 85°C in the presence of 0.1% Tween.

Fluorescence spectroscopy

Amyloid fibrils of chaplins were stained with 3 μM of the fluorescent dye ThT. Fluorescence was followed at 482 nm (excitation = 450 nm) on an Aminco-Bowman series 2 luminescence spectrometer (SLM-Aminco·).


The authors would like to thank Quirijn van Dijk and Anna Machowska for their assistance. Furthermore, we are indebted to Paul Dyson for providing the mutated cosmid used in this study. This work was financially supported by grants from the Northern Netherlands collaboration initiative (SNN EZ/KOMPAS RM119) and the Dutch Science Foundation NWO (Project 816.02.009). D.C. is supported by a Marie Curie Reintegration grant (FP7-PEOPLE-ERG-230944).