PNAS PLUS
Structure of the nonameric bacterial amyloid secretion channel Baohua Caoa,1, Yan Zhaoa,b,1, Yongjun Koua,c, Dongchun Nia, Xuejun Cai Zhanga, and Yihua Huanga,2 a National Laboratory of Biomacromolecules, National Center of Protein Science–Beijing, Institute of Biophysics, Chinese Academy of Sciences, Beijing 100101, China; bSchool of Life Sciences, University of Science and Technology of China, Hefei 230027, Anhui, China; and cUniversity of Chinese Academy of Sciences, Beijing 100101, China
Various strains of bacteria are able to produce a unique class of functional amyloids termed curli, which are critical for biofilm formation, host cell adhesion, and colonization of inert surfaces. Curli are secreted via the type VIII bacterial secretion system, and they share biochemical and structural characteristics with amyloid fibers that have been implicated in deleterious disease in humans. Here, we report the crystal structure of Escherichia coli CsgG, which is an essential lipoprotein component of the type VIII secretion system and which forms a secretion channel in the bacterial outer membrane for transporting curli subunits. CsgG forms a crown-shaped, symmetric nonameric channel that spans the outer membrane via a 36-strand β-barrel, with each subunit contributing four β-strands. This nonameric complex contains a central channel with a pore located at the middle. The eyelet of the pore is ∼12 Å in diameter and is lined with three stacked nine-residue rings consisting of Tyr-66, Asn-70, or Phe-71. Our structure-based functional studies suggest that Tyr-66 and Phe-71 residues function as gatekeepers for the selective secretion of curli subunits. Our study describes in detail, to our knowledge, the first core structure of the type VIII bacterial secretion machinery. Importantly, our structural analysis suggests that the curli subunits are secreted via CsgG across the bacterial outer membrane in an unfolded form. CsgG
| curli | amyloids | outer membrane protein | biofilm
U
nder stressful environmental conditions, some pathogenic Enterobacteriaceae produce a unique class of extracellular fibers called curli (1, 2). These nonbranching, highly aggregative filaments of 4−7 nm in diameter are resistant to degradation by proteases and denaturation by detergents and thus share biochemical and structural characteristics with amyloid fibers (3, 4). Amyloid structures are better known for their involvement in disease-associated processes of protein misfolding and aggregation in humans, such as Alzheimer’s disease (5–7). However, in contrast to these aberrantly folded proteins in pathogenic amyloidosis (6), bacterial curli are “functional” amyloids and are important for biofilm formation, host cell adhesion, and colonization of inert surfaces (8). The curli biogenesis pathway is often termed type VIII secretion system, or nucleation-precipitation secretion pathway, because the fiber growth is dependent on the presence of a nucleator molecule (3, 9). In Escherichia coli, this system is composed of the products of seven curli-specific genes (csg) encoded by two separately transcribed operons, csgBAC and csgDEFG (10). Among the multiple products of the csg gene operons, CsgA forms the major curli subunit and exhibits nucleationdependent oligomerization kinetics in vitro similar to what is observed for pathogenic amyloids from eukaryotic cells (11, 12). CsgB, the minor curli subunit, acts as the nucleator for the aggregation of the secreted CsgA curli subunits into curli fibers and is responsible for anchoring the fibers on the producing cell (13). Both subunits are transported across to the outer membrane (OM) via a specialized pore-forming lipoprotein, CsgG, which acts in concert with the periplasmic and extracellular accessory www.pnas.org/cgi/doi/10.1073/pnas.1411942111
proteins CsgE and CsgF, respectively (14–16). The periplasmic factor CsgE binds directly to CsgA, thereby preventing the latter from prematurely aggregating during transit through the periplasm (15). CsgF, conversely, appears to be exposed to the surface and is critical for CsgB-mediated nucleation of CsgA fibers (16). In addition, CsgD is a transcriptional factor regulating the csgBAC operon (10), presumably coordinating the timely expression of both csg operons. CsgC, the crystal structure of which has recently been reported (17), is a putative oxidoreductase, although its exact function during curli biogenesis remains illdefined. In summary, the curli biogenesis pathway represents a unique class among the bacterial protein secretion machineries, both from a structural as well as a mechanistic perspective. Here, we report the crystal structure of E. coli CsgG, the main transmembrane (TM) component of the type VIII secretion system. This 30-kDa lipoprotein is targeted to the OM, where it was thought to form a translocation channel of 2-nm diameter via oligomerization (14, 17). Our structural analysis reveals a large TM β-barrel structure, with a putative selection filter located at the OM–periplasm interface inside the barrel, representing, to our knowledge, the first core structure of the type VIII bacterial secretion machinery. Results and Discussion Overall Structure of the Nonameric Channel. Full-length E. coli
CsgG protein was overexpressed in E. coli and proteolytically digested, resulting in a variant lacking the N-terminal 34 residues of the mature CsgG protein (i.e., residues 16−49 of the full-length Significance Numerous bacteria produce a unique class of “functional” amyloids termed curli that are important for the fitness of the organism by mediating biofilm formation, host cell adhesion, and colonization on inert surfaces. Here, we report the crystal structure of CsgG, a lipoprotein that forms a secretion channel for curli subunits in the outer membrane. Each CsgG monomer is composed of four β-strands that span the outer membrane. Nine CsgG monomers together form a large, 36-stranded β-barrel with a central secretion channel. The channel is restricted by stacked rings within the pore, formed from inwardly protruding residues. The CsgG structure may provide a template for the development of antibiotics aimed at attenuating biofilm formation. Author contributions: Y.H. designed research; B.C., Y.Z., Y.K., and Y.H. performed research; Y.H. contributed new reagents/analytic tools; B.C., Y.Z., D.N., and Y.H. analyzed data; and X.C.Z. and Y.H. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. Data deposition: The crystallography, atomic coordinates, and structure factors have been deposited in the Protein Data Bank, www.pdb.org (PDB ID code 4Q79). 1
B.C. and Y.Z. contributed equally to this work.
2
To whom correspondence should be addressed. Email: [email protected].
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1411942111/-/DCSupplemental.
PNAS Early Edition | 1 of 6
MICROBIOLOGY
Edited by Scott J. Hultgren, Washington University School of Medicine, St. Louis, MO, and approved November 6, 2014 (received for review June 25, 2014)
protein) and the C-terminal 12 residues (Fig. S1). This truncated form was found to form oligomers, which were subsequently crystallized. The crystal form belongs to the space group P21212, with nine CsgG molecules per crystallographic asymmetric unit. Single-wavelength anomalous dispersion (SAD) data from both selenomethionine (SeMet) and mercury derivatives were used for phase determination. The structural model of the CsgG oligomer was refined at 2.9-Å resolution (Fig. 1 and Table S1). Ninefold noncrystallographic symmetry averaging provided excellent initial phases and high-quality electron-density maps (Fig. S2). The amino acid sequence of the crystallized CsgG construct and the corresponding secondary structure are shown in Fig. 1C. In the refined model, residues 116−128 and 251−265 of each CsgG molecule were omitted because of their crystallographic disorder (Fig. 1 B and C). Structural superposition indicates that the nine subunits are essentially identical with
a root-mean-square deviation of 4,300 Å2 from the two subunits. All three aforementioned modules contribute to these intersubunit interactions. For example, the β2 strand of one subunit interacts with β5 of the neighboring subunit in an antiparallel manner, creating 10 hydrogen bonds between the two main chains as well as two side-chain–side-chain hydrogen bonds (Fig. S4A). In addition, salt bridge bonds appear to contribute to the nonamer formation. In particular, Glu-97 and -107 of one subunit bond with Arg-177 and -98, respectively, of the neighboring subunit (Fig. S4B). Furthermore, the periplasmic region of the channel can be divided into three layers of secondary structures: Helices α1 and α3 line the outer surface of the channel; helix α2 lines the inner surface; and the periplasmic three-strand β-sheet resides between PNAS Early Edition | 3 of 6
the two layers (Fig. 1A, periplasmic view). Interestingly, α2 of each subunit is positioned away from the rest of the periplasmic module, forming extensive interactions with its neighboring subunit (Fig. 1A, periplasmic view), including the abovementioned Glu-107–Arg-98 salt-bridge bond. Such arrangement of α2 increases the interaction surface between two adjacent subunits and may contribute to the overall stability of the nonameric channel. Furthermore, the eyelet loops of the nonamer contribute significantly to intersubunit interactions. As shown in Fig. 3, viewed from the periplasmic space, the pore eyelet is lined with three stacked nine-residue rings consisting of Phe-71, Asn-70, or Tyr66, with the respective side chains pointing to the centers of the corresponding rings (Fig. 3B). On the periplasmic side, nine Phe63 residues form another ring outside of the Phe-71 ring. These residues likely play important roles in stabilizing the conformation of the channel eyelet via hydrophobic interactions. In light of the size of the pore eyelet, it is highly unlikely that fully folded CsgA or CsgB subunits could pass through the eyelet, suggesting that curli subunits are secreted in an unfolded form. TM β-Barrel. In contrast to most β-barrels of known structures of OM proteins, which are largely occluded by long extracellular loops, the TM β-barrel of CsgG lacks extracellular loops, resulting in a barrel cavity of >55,000 Å3 in size, and which is fully open to the extracellular environment (Fig. 2C). Such a feature has not been reported for OM proteins studied so far. Although the precise function of this cavity remains unresolved, it appears that, together with its negatively charged interior, it may create a microenvironment that facilitates the extracellular folding of exported curli subunits. Of the seven proteins encoded by the csg genes, CsgA (151 aa residues; pI 5.0), CsgB (151 aa residues; pI 9.6), and CsgF (119 aa residues; PI 4.5) are potentially exposed to the extracellular space (13, 14). Because of its higher pI, CsgB is the most likely candidate for interaction with the negative charges inside the cavity of the TM β-barrel of the CsgG complex. Consistent with this hypothesis, a triple mutant, E210K/E216K/E218K, had marked decreased curli production (Fig. S5B). We postulate that the negatively charged open cavity
A
B
Y66 F63
N70
F71 Y66 F63
N70
F71
(Side view)
(Periplasmic view)
Fig. 3. Detailed view of the eyelet of the nonameric CsgG channel. (A) Ribbon representation of a CsgG protomer with the side chains of residues Phe-63 (magenta), Tyr-66 (orange), Asn-70 (cyan), and Phe-71 (magenta) represented as sticks (side view from the plane of the OM). Residues Phe-71, Asn-70, and Tyr-66 of CsgG in the structure lie sequentially along the channel axis. (B) A detailed view of the eyelet. The side chains of Tyr-66, Asn-70, and Phe-71 point at the channel axis. The pore eyelet is lined with three stacked nine-residue rings, and each ring is formed by a single identical residue from each CsgG protomer. Nine Phe-63 residues form an external ring (indicated by a dashed circle) outside of the inner ring formed by nine Phe-71 residues.
4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1411942111
of the TM β-barrel of CsgG might play important functions in the secretion of curli subunits. Functional Studies. To assess the functional roles in curli biogenesis of potentially critical residues, a ΔcsgG cell strain was transformed with CsgG variants before assessment of amyloid production. Consistent with previous observations (14), the ΔcsgG mutant of the E. coli K-12 BW25113 strain was defective in curli biogenesis as revealed by negative-staining electron microscopy (Fig. 4A) and produced no detectable levels of curli, as determined by using a Congo red (CR) assay (Fig. S5A). When the native csgG gene (or the native csgG gene with additional six consecutive histidine-codon sequence following the Cys-16 codon of csgG) was reintroduced by plasmid transformation, wildtype (WT) levels of curli production were restored upon induction of csgG gene expression (Fig. 4A and Fig. S5A). Using this complementation assay, we analyzed several sets of mutants for their ability to restore curli production. We first studied the effects of point mutations that are localized in the eyelet of the secretion pore on curli biogenesis. We mutated each of the three residues (Tyr-66, Asn-70, and Phe-71) to 19 other amino acid residues and analyzed their curli production using the CR assay. Of the 19 point mutants of Asn-70, none exhibited decreased curli production compared with the WT (Fig. S5C), suggesting that Asn-70 may not be crucial for curli biogenesis. By contrast, residues Tyr-66 and Phe-71 of CsgG seem to play important roles in curli biogenesis. Eleven of 19 point mutants of Tyr-66 (Y66G, Y66V, Y66I, Y66T, Y66C, Y66P, Y66N, Y66K, Y66R, Y66D, and Y66E) and 9 of 19 point mutants of Phe-71 (F71A, F71G, F71V, F71S, F71T, F71C, F71P, F71D, and F71E) showed noticeable decreased curli production (Fig. 4 B and C and Fig. S5G). It seems that residues of aromatic side chains (Phe, His, Tyr, and Trp), Leu and Met are tolerable at both positions. Conversely, the following types of residues appears to be disfavored at both positions: residues of β-branched side chains (Val, Thr, and Ile), negatively charged side chains (Asp and Glu), and side chains capable forming disulfide bond (Cys) or affecting main-chain conformation (Gly and Pro). To rule out the possibility that the decreased curli production resulted from lower protein expression level of the mutants in the OM, an anti-His Western blot was performed to compare the relative protein expression levels of the mutants to that of the WT. As shown in Fig. S6, the 20 mutants have similar protein expression levels to that of the WT. Because Y66 and Phe-71 of CsgG form tightly interconnected nine-residue rings in the nonamer channel, next we tested whether these mutations affect the assembly of the nonamer. All these mutant proteins were purified by using affinity chromatography, and their stabilities were evaluated by using SDS/PAGE analysis under heating and nonheating conditions. The WT CsgG dissociated to monomers on 12% (wt/vol) SDS/PAGE only by heating samples at 100 °C for 10 min, suggesting that WT CsgG forms stable oligomers that are resistant to SDS denaturing under nonheating conditions. By contrast, except F71A mutation, protein samples of other point mutations showed certain percentages of monomers on SDS/PAGE even under nonheating conditions, indicating reduced stability of the corresponding nonamer (Fig. S7 B–D). Intriguingly, purified mutant proteins with reduced stability exhibited similar size-exclusion profiles to that of the WT CsgG (Fig. S8), indicating that these mutant proteins form nonamers in the OM. Thus, the decreased curli production caused by the mutations at these two positions may result from the decreased selectivity of the pore eyelet to curli subunits. However, we are unable to rule out the possibility that the impaired overall stability of the CsgG nonamer affects the function of the CsgG nonamer. However, at least for the F71A mutant, it seems that mutation did not affect the stability of the nonamer; thus, the decreased curli production is likely to be caused by the Cao et al.
B
C
Fig. 4. Mutational analysis of the nonameric CsgG channel. (A) Negativestain EM micrographs of E. coli K-12 strain BW25113 WT (Left), ΔcsgG mutant (Center), and BW25113 ΔcsgG mutant transformed with pBAD22-CsgG plasmid (Right). Hair-like surface appendages are curli. (Scale bars: 500 nm.) (B) CR–yeast extract casamino acids (YESCA) plate with E. coli K-12 strain BW25113ΔcsgG mutant transformed with vector control (pBAD22), plasmids encoding WT CsgG or CsgG mutants at position of Tyr-66. Eleven of 19 point mutants (Y66G, Y66V, Y66I, Y66T, Y66C, Y66P, Y66N, Y66K, Y66R, Y66D, and Y66E) have noticeable decreased curli production compared with the WT. (C) CR-YESCA plate with E. coli K-12 strain BW25113ΔcsgG mutant transformed with vector control (pBAD22), plasmids encoding WT CsgG or CsgG mutants at position of Phe-71. Nine of 19 point mutants (F71A, F71G, F71V, F71S, F71T, F71P, F71D, and F71E) interfere with curli production. Reduced
Cao et al.
curli production of CsgG mutants was quantified by using a Thioflavin T (ThT) fluorescence assay as shown in Fig. S5G. All CR-YESCA plates contained 0.01 mM arabinose. Experiments were repeated at least three times. Representative data are shown.
PNAS Early Edition | 5 of 6
PNAS PLUS
impaired substrate selectivity of the CsgG channel to either CsgA or CsgB. Using a similar strategy, we further characterized point mutants of Phe-63 (F63A, F63L, F63T, and F63D). These four point mutants also showed decreases in curli production (Fig. S5 E and G), and mutations seemed to affect the stability of CsgG nonamer as well (Fig. S7B). These results showed that the three aromatic residues Phe-63, Tyr-66, and Phe-71 in the eyelet region of the secretion pore play important roles in curli biogenesis. Because the size of the eyelet pore appears not to be large enough for accommodating periplasmic components such as CsgE (129 aa residues; pI 5.3), residues Tyr-66 and Phe-71 of CsgG themselves may function as determinants for the selective transport of curli subunits CsgA and CsgB. Next, we investigated the functional importance of the two intersubunit salt-bridge bonds (Glu-97–Arg-177 and Arg-98– Glu-107) of the CsgG nonamer in curli biogenesis. Disruption of any of these two salt-bridge bonds severely impaired curli production (Fig. S5 D and G). Interestingly, unlike point mutations at the eyelet region that did not affect protein expression, antiHis Western blot showed that these mutations apparently increased their protein expression levels in the OM (Fig. S6). However, disruption of any of these two salt-bridge bonds caused either much lower protein expression level of the mutant protein (E97A, R98A, R98E, E107A, and E107R) under overexpression conditions or severely affected the stability of the nonameric CsgG channel (E97A) (Fig. S7A). However, even for mutant E97A that appears to be the most unstable mutant also form oligomers in the OM as evaluated from their gel filtration profiles (Fig. S8). These studies strongly suggest that the structurally observed intersubunit electrostatic interactions play an important role in maintaining the stability of the CsgG nonamer in the OM, and again the stability of the nonamer appears to be important for curli biogenesis. We also studied the mutational effects on curli biogenesis of those negatively charged residues that line the inner wall of the CsgG channel. Individual point mutations at residues Asp-164, Glu-200, Asp-210, Glu-216, and Glu-218, which confer the overall negative charge of the lumen of the TM β-barrel, to either a Lys or Ala residue, showed phenotypes similar to the WT in terms of curli production (Fig. S5B). Furthermore, none of the double mutants D210K/E216K, D210K/E218K, or E216K/ E218K had noticeable decreased curli production. Interestingly, a triple mutant, E210K/E216K/E218K, had marked decreased curli production (Fig. S5 B and G). Anti-His Western blot also showed that these mutations did not affect their protein expression levels in the OM (Fig. S6). These observations indicate that the negative charges of the β-barrel lumen may function in a collective manner, with single or double point mutations having only limited impact on curli biogenesis. Together, the results of the complementation assays suggest that the intersubunit salt-bridge bonds play important roles in the stability of CsgG nonamer in the OM. Furthermore, the overall negative charge of the lumen of the TM β-barrel might be important for curli polymerization. Importantly, the three aromatic residues, Phe-63, Tyr-66, and Tyr-71, located at the eyelet region are crucial for recognizing curli subunits and stabilizing the nonameric channel. Because the 22 N-terminal residues of the mature CsgA curli subunit (as well as the homologous CsgB) are sufficient for targeting CsgG (18, 19), we propose that critical information in this peptide fragment of CsgA and CsgB may be recognized by the eyelet filter of CsgG.
MICROBIOLOGY
A
Concluding Remarks. The biogenesis of bacterial curli is an ideal experimental system to investigate the mechanisms by which bacteria meet the challenges of controlled and ordered amyloid assembly. Understanding the mechanisms of curli biogenesis in bacteria is a prerequisite to the development of therapeutics, e.g., via attenuating biofilm formation, and may provide a paradigm for understanding pathogenic amyloidogenesis in humans (20, 21). The crystal structure of CsgG reveals a protein-secretion channel of distinct architecture—namely, the periplasmic cavity, the 12-Å diameter selectivity pore, and the open 36-strand TM β-barrel. Finally, this structure provides essential clues for functional studies that explore the detailed mechanisms for secretion and polymerization of curli subunits.
BL21(DE3) cells for protein overexpression. The affinity-purified CsgG protein was subsequently subjected to chymotrypsin proteolysis for 3 h at room temperature, followed by size-exclusion chromatography with a Superose 6 10/300 GL column (GE Healthcare). Crystallization was conducted at 16 °C by using the sitting drop vapor diffusion method. CsgG crystals appeared within 1 wk and grew to their final size in ∼2 wk. SeMet-substituted crystals were prepared by using the same protocol as that used for native crystals. Crystals of the mercury derivative were obtained by soaking native crystals in 1 mM CH3HgCl for 2 d at 16 °C. Full experimental details are provided in SI Materials and Methods.
The complete csgG gene was subcloned into the pQLinkN vector (22) with an N-terminal His6-tag that was inserted after Cys-16 of CsgG to facilitate affinity purification. The constructed plasmid was then transformed into E. coli
ACKNOWLEDGMENTS. We thank Drs. Sarah Perrett, Guohong Li, and Zheng Zhou and the entire Y.H. group for technical assistance and valuable discussions. The diffraction data were collected at the Shanghai Synchrotron Radiation Facility and Beijing Synchrotron Radiation Facility of China. This work was supported by Ministry of Science and Technology Grants 2013CB910603 and 2012CB917302 (to Y.H.) and 2011CB910301 (to X.C.Z.); Strategic Priority Research Program of Chinese Academy of Sciences Grant XDB080203 (to Y.H. and X.C.Z.); and National Natural Science Foundation of China Grant 31170698 (to Y.H.).
1. Olsén A, Jonsson A, Normark S (1989) Fibronectin binding mediated by a novel class of surface organelles on Escherichia coli. Nature 338(6217):652–655. 2. Collinson SK, Emödy L, Müller KH, Trust TJ, Kay WW (1991) Purification and characterization of thin, aggregative fimbriae from Salmonella enteritidis. J Bacteriol 173(15):4773–4781. 3. Chapman MR, et al. (2002) Role of Escherichia coli curli operons in directing amyloid fiber formation. Science 295(5556):851–855. 4. Shewmaker F, et al. (2009) The functional curli amyloid is not based on in-register parallel beta-sheet structure. J Biol Chem 284(37):25065–25076. 5. Sunde M, et al. (1997) Common core structure of amyloid fibrils by synchrotron X-ray diffraction. J Mol Biol 273(3):729–739. 6. Chiti F, Dobson CM (2006) Protein misfolding, functional amyloid, and human disease. Annu Rev Biochem 75:333–366. 7. Moreno-Gonzalez I, Soto C (2011) Misfolded protein aggregates: Mechanisms, structures and potential for disease transmission. Semin Cell Dev Biol 22(5):482–487. 8. Barnhart MM, Chapman MR (2006) Curli biogenesis and function. Annu Rev Microbiol 60:131–147. 9. Desvaux M, Hébraud M, Talon R, Henderson IR (2009) Secretion and subcellular localizations of bacterial proteins: A semantic awareness issue. Trends Microbiol 17(4):139–145. 10. Hammar M, Arnqvist A, Bian Z, Olsén A, Normark S (1995) Expression of two csg operons is required for production of fibronectin- and congo red-binding curli polymers in Escherichia coli K-12. Mol Microbiol 18(4):661–670. 11. Cherny I, et al. (2005) The formation of Escherichia coli curli amyloid fibrils is mediated by prion-like peptide repeats. J Mol Biol 352(2):245–252. 12. Wang X, Smith DR, Jones JW, Chapman MR (2007) In vitro polymerization of a functional Escherichia coli amyloid protein. J Biol Chem 282(6):3713–3719.
13. Hammer ND, Schmidt JC, Chapman MR (2007) The curli nucleator protein, CsgB, contains an amyloidogenic domain that directs CsgA polymerization. Proc Natl Acad Sci USA 104(30):12494–12499. 14. Robinson LS, Ashman EM, Hultgren SJ, Chapman MR (2006) Secretion of curli fibre subunits is mediated by the outer membrane-localized CsgG protein. Mol Microbiol 59(3):870–881. 15. Nenninger AA, et al. (2011) CsgE is a curli secretion specificity factor that prevents amyloid fibre aggregation. Mol Microbiol 81(2):486–499. 16. Nenninger AA, Robinson LS, Hultgren SJ (2009) Localized and efficient curli nucleation requires the chaperone-like amyloid assembly protein CsgF. Proc Natl Acad Sci USA 106(3):900–905. 17. Taylor JD, et al. (2011) Atomic resolution insights into curli fiber biogenesis. Structure 19(9):1307–1316. 18. Evans ML, Chapman MR (2014) Curli biogenesis: Order out of disorder. Biochim Biophys Acta 1843(8):1551–1558. 19. Van Gerven N, et al. (2014) Secretion and functional display of fusion proteins through the curli biogenesis pathway. Mol Microbiol 91(5):1022–1035. 20. Kato M, et al. (2012) Cell-free formation of RNA granules: Low complexity sequence domains form dynamic fibers within hydrogels. Cell 149(4):753–767. 21. Li J, et al. (2012) The RIP1/RIP3 necrosome forms a functional amyloid signaling complex required for programmed necrosis. Cell 150(2):339–350. 22. Scheich C, Kümmel D, Soumailakakis D, Heinemann U, Büssow K (2007) Vectors for coexpression of an unrestricted number of proteins. Nucleic Acids Res 35(6):e43. 23. Zhang XJ, Matthews BW (1995) Edpdb—A multifunctional tool for protein-structure analysis. J Appl Cryst 28:624–630.
Materials and Methods
6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1411942111
Cao et al.
Structure of the nonameric bacterial amyloid secretion channel Baohua Caoa,1, Yan Zhaoa,b,1, Yongjun Koua,c, Dongchun Nia, Xuejun Cai Zhanga, and Yihua Huanga,2 a National Laboratory of Biomacromolecules, National Center of Protein Science–Beijing, Institute of Biophysics, Chinese Academy of Sciences, Beijing 100101, China; bSchool of Life Sciences, University of Science and Technology of China, Hefei 230027, Anhui, China; and cUniversity of Chinese Academy of Sciences, Beijing 100101, China
Various strains of bacteria are able to produce a unique class of functional amyloids termed curli, which are critical for biofilm formation, host cell adhesion, and colonization of inert surfaces. Curli are secreted via the type VIII bacterial secretion system, and they share biochemical and structural characteristics with amyloid fibers that have been implicated in deleterious disease in humans. Here, we report the crystal structure of Escherichia coli CsgG, which is an essential lipoprotein component of the type VIII secretion system and which forms a secretion channel in the bacterial outer membrane for transporting curli subunits. CsgG forms a crown-shaped, symmetric nonameric channel that spans the outer membrane via a 36-strand β-barrel, with each subunit contributing four β-strands. This nonameric complex contains a central channel with a pore located at the middle. The eyelet of the pore is ∼12 Å in diameter and is lined with three stacked nine-residue rings consisting of Tyr-66, Asn-70, or Phe-71. Our structure-based functional studies suggest that Tyr-66 and Phe-71 residues function as gatekeepers for the selective secretion of curli subunits. Our study describes in detail, to our knowledge, the first core structure of the type VIII bacterial secretion machinery. Importantly, our structural analysis suggests that the curli subunits are secreted via CsgG across the bacterial outer membrane in an unfolded form. CsgG
| curli | amyloids | outer membrane protein | biofilm
U
nder stressful environmental conditions, some pathogenic Enterobacteriaceae produce a unique class of extracellular fibers called curli (1, 2). These nonbranching, highly aggregative filaments of 4−7 nm in diameter are resistant to degradation by proteases and denaturation by detergents and thus share biochemical and structural characteristics with amyloid fibers (3, 4). Amyloid structures are better known for their involvement in disease-associated processes of protein misfolding and aggregation in humans, such as Alzheimer’s disease (5–7). However, in contrast to these aberrantly folded proteins in pathogenic amyloidosis (6), bacterial curli are “functional” amyloids and are important for biofilm formation, host cell adhesion, and colonization of inert surfaces (8). The curli biogenesis pathway is often termed type VIII secretion system, or nucleation-precipitation secretion pathway, because the fiber growth is dependent on the presence of a nucleator molecule (3, 9). In Escherichia coli, this system is composed of the products of seven curli-specific genes (csg) encoded by two separately transcribed operons, csgBAC and csgDEFG (10). Among the multiple products of the csg gene operons, CsgA forms the major curli subunit and exhibits nucleationdependent oligomerization kinetics in vitro similar to what is observed for pathogenic amyloids from eukaryotic cells (11, 12). CsgB, the minor curli subunit, acts as the nucleator for the aggregation of the secreted CsgA curli subunits into curli fibers and is responsible for anchoring the fibers on the producing cell (13). Both subunits are transported across to the outer membrane (OM) via a specialized pore-forming lipoprotein, CsgG, which acts in concert with the periplasmic and extracellular accessory www.pnas.org/cgi/doi/10.1073/pnas.1411942111
proteins CsgE and CsgF, respectively (14–16). The periplasmic factor CsgE binds directly to CsgA, thereby preventing the latter from prematurely aggregating during transit through the periplasm (15). CsgF, conversely, appears to be exposed to the surface and is critical for CsgB-mediated nucleation of CsgA fibers (16). In addition, CsgD is a transcriptional factor regulating the csgBAC operon (10), presumably coordinating the timely expression of both csg operons. CsgC, the crystal structure of which has recently been reported (17), is a putative oxidoreductase, although its exact function during curli biogenesis remains illdefined. In summary, the curli biogenesis pathway represents a unique class among the bacterial protein secretion machineries, both from a structural as well as a mechanistic perspective. Here, we report the crystal structure of E. coli CsgG, the main transmembrane (TM) component of the type VIII secretion system. This 30-kDa lipoprotein is targeted to the OM, where it was thought to form a translocation channel of 2-nm diameter via oligomerization (14, 17). Our structural analysis reveals a large TM β-barrel structure, with a putative selection filter located at the OM–periplasm interface inside the barrel, representing, to our knowledge, the first core structure of the type VIII bacterial secretion machinery. Results and Discussion Overall Structure of the Nonameric Channel. Full-length E. coli
CsgG protein was overexpressed in E. coli and proteolytically digested, resulting in a variant lacking the N-terminal 34 residues of the mature CsgG protein (i.e., residues 16−49 of the full-length Significance Numerous bacteria produce a unique class of “functional” amyloids termed curli that are important for the fitness of the organism by mediating biofilm formation, host cell adhesion, and colonization on inert surfaces. Here, we report the crystal structure of CsgG, a lipoprotein that forms a secretion channel for curli subunits in the outer membrane. Each CsgG monomer is composed of four β-strands that span the outer membrane. Nine CsgG monomers together form a large, 36-stranded β-barrel with a central secretion channel. The channel is restricted by stacked rings within the pore, formed from inwardly protruding residues. The CsgG structure may provide a template for the development of antibiotics aimed at attenuating biofilm formation. Author contributions: Y.H. designed research; B.C., Y.Z., Y.K., and Y.H. performed research; Y.H. contributed new reagents/analytic tools; B.C., Y.Z., D.N., and Y.H. analyzed data; and X.C.Z. and Y.H. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. Data deposition: The crystallography, atomic coordinates, and structure factors have been deposited in the Protein Data Bank, www.pdb.org (PDB ID code 4Q79). 1
B.C. and Y.Z. contributed equally to this work.
2
To whom correspondence should be addressed. Email: [email protected].
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1411942111/-/DCSupplemental.
PNAS Early Edition | 1 of 6
MICROBIOLOGY
Edited by Scott J. Hultgren, Washington University School of Medicine, St. Louis, MO, and approved November 6, 2014 (received for review June 25, 2014)
protein) and the C-terminal 12 residues (Fig. S1). This truncated form was found to form oligomers, which were subsequently crystallized. The crystal form belongs to the space group P21212, with nine CsgG molecules per crystallographic asymmetric unit. Single-wavelength anomalous dispersion (SAD) data from both selenomethionine (SeMet) and mercury derivatives were used for phase determination. The structural model of the CsgG oligomer was refined at 2.9-Å resolution (Fig. 1 and Table S1). Ninefold noncrystallographic symmetry averaging provided excellent initial phases and high-quality electron-density maps (Fig. S2). The amino acid sequence of the crystallized CsgG construct and the corresponding secondary structure are shown in Fig. 1C. In the refined model, residues 116−128 and 251−265 of each CsgG molecule were omitted because of their crystallographic disorder (Fig. 1 B and C). Structural superposition indicates that the nine subunits are essentially identical with
a root-mean-square deviation of 4,300 Å2 from the two subunits. All three aforementioned modules contribute to these intersubunit interactions. For example, the β2 strand of one subunit interacts with β5 of the neighboring subunit in an antiparallel manner, creating 10 hydrogen bonds between the two main chains as well as two side-chain–side-chain hydrogen bonds (Fig. S4A). In addition, salt bridge bonds appear to contribute to the nonamer formation. In particular, Glu-97 and -107 of one subunit bond with Arg-177 and -98, respectively, of the neighboring subunit (Fig. S4B). Furthermore, the periplasmic region of the channel can be divided into three layers of secondary structures: Helices α1 and α3 line the outer surface of the channel; helix α2 lines the inner surface; and the periplasmic three-strand β-sheet resides between PNAS Early Edition | 3 of 6
the two layers (Fig. 1A, periplasmic view). Interestingly, α2 of each subunit is positioned away from the rest of the periplasmic module, forming extensive interactions with its neighboring subunit (Fig. 1A, periplasmic view), including the abovementioned Glu-107–Arg-98 salt-bridge bond. Such arrangement of α2 increases the interaction surface between two adjacent subunits and may contribute to the overall stability of the nonameric channel. Furthermore, the eyelet loops of the nonamer contribute significantly to intersubunit interactions. As shown in Fig. 3, viewed from the periplasmic space, the pore eyelet is lined with three stacked nine-residue rings consisting of Phe-71, Asn-70, or Tyr66, with the respective side chains pointing to the centers of the corresponding rings (Fig. 3B). On the periplasmic side, nine Phe63 residues form another ring outside of the Phe-71 ring. These residues likely play important roles in stabilizing the conformation of the channel eyelet via hydrophobic interactions. In light of the size of the pore eyelet, it is highly unlikely that fully folded CsgA or CsgB subunits could pass through the eyelet, suggesting that curli subunits are secreted in an unfolded form. TM β-Barrel. In contrast to most β-barrels of known structures of OM proteins, which are largely occluded by long extracellular loops, the TM β-barrel of CsgG lacks extracellular loops, resulting in a barrel cavity of >55,000 Å3 in size, and which is fully open to the extracellular environment (Fig. 2C). Such a feature has not been reported for OM proteins studied so far. Although the precise function of this cavity remains unresolved, it appears that, together with its negatively charged interior, it may create a microenvironment that facilitates the extracellular folding of exported curli subunits. Of the seven proteins encoded by the csg genes, CsgA (151 aa residues; pI 5.0), CsgB (151 aa residues; pI 9.6), and CsgF (119 aa residues; PI 4.5) are potentially exposed to the extracellular space (13, 14). Because of its higher pI, CsgB is the most likely candidate for interaction with the negative charges inside the cavity of the TM β-barrel of the CsgG complex. Consistent with this hypothesis, a triple mutant, E210K/E216K/E218K, had marked decreased curli production (Fig. S5B). We postulate that the negatively charged open cavity
A
B
Y66 F63
N70
F71 Y66 F63
N70
F71
(Side view)
(Periplasmic view)
Fig. 3. Detailed view of the eyelet of the nonameric CsgG channel. (A) Ribbon representation of a CsgG protomer with the side chains of residues Phe-63 (magenta), Tyr-66 (orange), Asn-70 (cyan), and Phe-71 (magenta) represented as sticks (side view from the plane of the OM). Residues Phe-71, Asn-70, and Tyr-66 of CsgG in the structure lie sequentially along the channel axis. (B) A detailed view of the eyelet. The side chains of Tyr-66, Asn-70, and Phe-71 point at the channel axis. The pore eyelet is lined with three stacked nine-residue rings, and each ring is formed by a single identical residue from each CsgG protomer. Nine Phe-63 residues form an external ring (indicated by a dashed circle) outside of the inner ring formed by nine Phe-71 residues.
4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1411942111
of the TM β-barrel of CsgG might play important functions in the secretion of curli subunits. Functional Studies. To assess the functional roles in curli biogenesis of potentially critical residues, a ΔcsgG cell strain was transformed with CsgG variants before assessment of amyloid production. Consistent with previous observations (14), the ΔcsgG mutant of the E. coli K-12 BW25113 strain was defective in curli biogenesis as revealed by negative-staining electron microscopy (Fig. 4A) and produced no detectable levels of curli, as determined by using a Congo red (CR) assay (Fig. S5A). When the native csgG gene (or the native csgG gene with additional six consecutive histidine-codon sequence following the Cys-16 codon of csgG) was reintroduced by plasmid transformation, wildtype (WT) levels of curli production were restored upon induction of csgG gene expression (Fig. 4A and Fig. S5A). Using this complementation assay, we analyzed several sets of mutants for their ability to restore curli production. We first studied the effects of point mutations that are localized in the eyelet of the secretion pore on curli biogenesis. We mutated each of the three residues (Tyr-66, Asn-70, and Phe-71) to 19 other amino acid residues and analyzed their curli production using the CR assay. Of the 19 point mutants of Asn-70, none exhibited decreased curli production compared with the WT (Fig. S5C), suggesting that Asn-70 may not be crucial for curli biogenesis. By contrast, residues Tyr-66 and Phe-71 of CsgG seem to play important roles in curli biogenesis. Eleven of 19 point mutants of Tyr-66 (Y66G, Y66V, Y66I, Y66T, Y66C, Y66P, Y66N, Y66K, Y66R, Y66D, and Y66E) and 9 of 19 point mutants of Phe-71 (F71A, F71G, F71V, F71S, F71T, F71C, F71P, F71D, and F71E) showed noticeable decreased curli production (Fig. 4 B and C and Fig. S5G). It seems that residues of aromatic side chains (Phe, His, Tyr, and Trp), Leu and Met are tolerable at both positions. Conversely, the following types of residues appears to be disfavored at both positions: residues of β-branched side chains (Val, Thr, and Ile), negatively charged side chains (Asp and Glu), and side chains capable forming disulfide bond (Cys) or affecting main-chain conformation (Gly and Pro). To rule out the possibility that the decreased curli production resulted from lower protein expression level of the mutants in the OM, an anti-His Western blot was performed to compare the relative protein expression levels of the mutants to that of the WT. As shown in Fig. S6, the 20 mutants have similar protein expression levels to that of the WT. Because Y66 and Phe-71 of CsgG form tightly interconnected nine-residue rings in the nonamer channel, next we tested whether these mutations affect the assembly of the nonamer. All these mutant proteins were purified by using affinity chromatography, and their stabilities were evaluated by using SDS/PAGE analysis under heating and nonheating conditions. The WT CsgG dissociated to monomers on 12% (wt/vol) SDS/PAGE only by heating samples at 100 °C for 10 min, suggesting that WT CsgG forms stable oligomers that are resistant to SDS denaturing under nonheating conditions. By contrast, except F71A mutation, protein samples of other point mutations showed certain percentages of monomers on SDS/PAGE even under nonheating conditions, indicating reduced stability of the corresponding nonamer (Fig. S7 B–D). Intriguingly, purified mutant proteins with reduced stability exhibited similar size-exclusion profiles to that of the WT CsgG (Fig. S8), indicating that these mutant proteins form nonamers in the OM. Thus, the decreased curli production caused by the mutations at these two positions may result from the decreased selectivity of the pore eyelet to curli subunits. However, we are unable to rule out the possibility that the impaired overall stability of the CsgG nonamer affects the function of the CsgG nonamer. However, at least for the F71A mutant, it seems that mutation did not affect the stability of the nonamer; thus, the decreased curli production is likely to be caused by the Cao et al.
B
C
Fig. 4. Mutational analysis of the nonameric CsgG channel. (A) Negativestain EM micrographs of E. coli K-12 strain BW25113 WT (Left), ΔcsgG mutant (Center), and BW25113 ΔcsgG mutant transformed with pBAD22-CsgG plasmid (Right). Hair-like surface appendages are curli. (Scale bars: 500 nm.) (B) CR–yeast extract casamino acids (YESCA) plate with E. coli K-12 strain BW25113ΔcsgG mutant transformed with vector control (pBAD22), plasmids encoding WT CsgG or CsgG mutants at position of Tyr-66. Eleven of 19 point mutants (Y66G, Y66V, Y66I, Y66T, Y66C, Y66P, Y66N, Y66K, Y66R, Y66D, and Y66E) have noticeable decreased curli production compared with the WT. (C) CR-YESCA plate with E. coli K-12 strain BW25113ΔcsgG mutant transformed with vector control (pBAD22), plasmids encoding WT CsgG or CsgG mutants at position of Phe-71. Nine of 19 point mutants (F71A, F71G, F71V, F71S, F71T, F71P, F71D, and F71E) interfere with curli production. Reduced
Cao et al.
curli production of CsgG mutants was quantified by using a Thioflavin T (ThT) fluorescence assay as shown in Fig. S5G. All CR-YESCA plates contained 0.01 mM arabinose. Experiments were repeated at least three times. Representative data are shown.
PNAS Early Edition | 5 of 6
PNAS PLUS
impaired substrate selectivity of the CsgG channel to either CsgA or CsgB. Using a similar strategy, we further characterized point mutants of Phe-63 (F63A, F63L, F63T, and F63D). These four point mutants also showed decreases in curli production (Fig. S5 E and G), and mutations seemed to affect the stability of CsgG nonamer as well (Fig. S7B). These results showed that the three aromatic residues Phe-63, Tyr-66, and Phe-71 in the eyelet region of the secretion pore play important roles in curli biogenesis. Because the size of the eyelet pore appears not to be large enough for accommodating periplasmic components such as CsgE (129 aa residues; pI 5.3), residues Tyr-66 and Phe-71 of CsgG themselves may function as determinants for the selective transport of curli subunits CsgA and CsgB. Next, we investigated the functional importance of the two intersubunit salt-bridge bonds (Glu-97–Arg-177 and Arg-98– Glu-107) of the CsgG nonamer in curli biogenesis. Disruption of any of these two salt-bridge bonds severely impaired curli production (Fig. S5 D and G). Interestingly, unlike point mutations at the eyelet region that did not affect protein expression, antiHis Western blot showed that these mutations apparently increased their protein expression levels in the OM (Fig. S6). However, disruption of any of these two salt-bridge bonds caused either much lower protein expression level of the mutant protein (E97A, R98A, R98E, E107A, and E107R) under overexpression conditions or severely affected the stability of the nonameric CsgG channel (E97A) (Fig. S7A). However, even for mutant E97A that appears to be the most unstable mutant also form oligomers in the OM as evaluated from their gel filtration profiles (Fig. S8). These studies strongly suggest that the structurally observed intersubunit electrostatic interactions play an important role in maintaining the stability of the CsgG nonamer in the OM, and again the stability of the nonamer appears to be important for curli biogenesis. We also studied the mutational effects on curli biogenesis of those negatively charged residues that line the inner wall of the CsgG channel. Individual point mutations at residues Asp-164, Glu-200, Asp-210, Glu-216, and Glu-218, which confer the overall negative charge of the lumen of the TM β-barrel, to either a Lys or Ala residue, showed phenotypes similar to the WT in terms of curli production (Fig. S5B). Furthermore, none of the double mutants D210K/E216K, D210K/E218K, or E216K/ E218K had noticeable decreased curli production. Interestingly, a triple mutant, E210K/E216K/E218K, had marked decreased curli production (Fig. S5 B and G). Anti-His Western blot also showed that these mutations did not affect their protein expression levels in the OM (Fig. S6). These observations indicate that the negative charges of the β-barrel lumen may function in a collective manner, with single or double point mutations having only limited impact on curli biogenesis. Together, the results of the complementation assays suggest that the intersubunit salt-bridge bonds play important roles in the stability of CsgG nonamer in the OM. Furthermore, the overall negative charge of the lumen of the TM β-barrel might be important for curli polymerization. Importantly, the three aromatic residues, Phe-63, Tyr-66, and Tyr-71, located at the eyelet region are crucial for recognizing curli subunits and stabilizing the nonameric channel. Because the 22 N-terminal residues of the mature CsgA curli subunit (as well as the homologous CsgB) are sufficient for targeting CsgG (18, 19), we propose that critical information in this peptide fragment of CsgA and CsgB may be recognized by the eyelet filter of CsgG.
MICROBIOLOGY
A
Concluding Remarks. The biogenesis of bacterial curli is an ideal experimental system to investigate the mechanisms by which bacteria meet the challenges of controlled and ordered amyloid assembly. Understanding the mechanisms of curli biogenesis in bacteria is a prerequisite to the development of therapeutics, e.g., via attenuating biofilm formation, and may provide a paradigm for understanding pathogenic amyloidogenesis in humans (20, 21). The crystal structure of CsgG reveals a protein-secretion channel of distinct architecture—namely, the periplasmic cavity, the 12-Å diameter selectivity pore, and the open 36-strand TM β-barrel. Finally, this structure provides essential clues for functional studies that explore the detailed mechanisms for secretion and polymerization of curli subunits.
BL21(DE3) cells for protein overexpression. The affinity-purified CsgG protein was subsequently subjected to chymotrypsin proteolysis for 3 h at room temperature, followed by size-exclusion chromatography with a Superose 6 10/300 GL column (GE Healthcare). Crystallization was conducted at 16 °C by using the sitting drop vapor diffusion method. CsgG crystals appeared within 1 wk and grew to their final size in ∼2 wk. SeMet-substituted crystals were prepared by using the same protocol as that used for native crystals. Crystals of the mercury derivative were obtained by soaking native crystals in 1 mM CH3HgCl for 2 d at 16 °C. Full experimental details are provided in SI Materials and Methods.
The complete csgG gene was subcloned into the pQLinkN vector (22) with an N-terminal His6-tag that was inserted after Cys-16 of CsgG to facilitate affinity purification. The constructed plasmid was then transformed into E. coli
ACKNOWLEDGMENTS. We thank Drs. Sarah Perrett, Guohong Li, and Zheng Zhou and the entire Y.H. group for technical assistance and valuable discussions. The diffraction data were collected at the Shanghai Synchrotron Radiation Facility and Beijing Synchrotron Radiation Facility of China. This work was supported by Ministry of Science and Technology Grants 2013CB910603 and 2012CB917302 (to Y.H.) and 2011CB910301 (to X.C.Z.); Strategic Priority Research Program of Chinese Academy of Sciences Grant XDB080203 (to Y.H. and X.C.Z.); and National Natural Science Foundation of China Grant 31170698 (to Y.H.).
1. Olsén A, Jonsson A, Normark S (1989) Fibronectin binding mediated by a novel class of surface organelles on Escherichia coli. Nature 338(6217):652–655. 2. Collinson SK, Emödy L, Müller KH, Trust TJ, Kay WW (1991) Purification and characterization of thin, aggregative fimbriae from Salmonella enteritidis. J Bacteriol 173(15):4773–4781. 3. Chapman MR, et al. (2002) Role of Escherichia coli curli operons in directing amyloid fiber formation. Science 295(5556):851–855. 4. Shewmaker F, et al. (2009) The functional curli amyloid is not based on in-register parallel beta-sheet structure. J Biol Chem 284(37):25065–25076. 5. Sunde M, et al. (1997) Common core structure of amyloid fibrils by synchrotron X-ray diffraction. J Mol Biol 273(3):729–739. 6. Chiti F, Dobson CM (2006) Protein misfolding, functional amyloid, and human disease. Annu Rev Biochem 75:333–366. 7. Moreno-Gonzalez I, Soto C (2011) Misfolded protein aggregates: Mechanisms, structures and potential for disease transmission. Semin Cell Dev Biol 22(5):482–487. 8. Barnhart MM, Chapman MR (2006) Curli biogenesis and function. Annu Rev Microbiol 60:131–147. 9. Desvaux M, Hébraud M, Talon R, Henderson IR (2009) Secretion and subcellular localizations of bacterial proteins: A semantic awareness issue. Trends Microbiol 17(4):139–145. 10. Hammar M, Arnqvist A, Bian Z, Olsén A, Normark S (1995) Expression of two csg operons is required for production of fibronectin- and congo red-binding curli polymers in Escherichia coli K-12. Mol Microbiol 18(4):661–670. 11. Cherny I, et al. (2005) The formation of Escherichia coli curli amyloid fibrils is mediated by prion-like peptide repeats. J Mol Biol 352(2):245–252. 12. Wang X, Smith DR, Jones JW, Chapman MR (2007) In vitro polymerization of a functional Escherichia coli amyloid protein. J Biol Chem 282(6):3713–3719.
13. Hammer ND, Schmidt JC, Chapman MR (2007) The curli nucleator protein, CsgB, contains an amyloidogenic domain that directs CsgA polymerization. Proc Natl Acad Sci USA 104(30):12494–12499. 14. Robinson LS, Ashman EM, Hultgren SJ, Chapman MR (2006) Secretion of curli fibre subunits is mediated by the outer membrane-localized CsgG protein. Mol Microbiol 59(3):870–881. 15. Nenninger AA, et al. (2011) CsgE is a curli secretion specificity factor that prevents amyloid fibre aggregation. Mol Microbiol 81(2):486–499. 16. Nenninger AA, Robinson LS, Hultgren SJ (2009) Localized and efficient curli nucleation requires the chaperone-like amyloid assembly protein CsgF. Proc Natl Acad Sci USA 106(3):900–905. 17. Taylor JD, et al. (2011) Atomic resolution insights into curli fiber biogenesis. Structure 19(9):1307–1316. 18. Evans ML, Chapman MR (2014) Curli biogenesis: Order out of disorder. Biochim Biophys Acta 1843(8):1551–1558. 19. Van Gerven N, et al. (2014) Secretion and functional display of fusion proteins through the curli biogenesis pathway. Mol Microbiol 91(5):1022–1035. 20. Kato M, et al. (2012) Cell-free formation of RNA granules: Low complexity sequence domains form dynamic fibers within hydrogels. Cell 149(4):753–767. 21. Li J, et al. (2012) The RIP1/RIP3 necrosome forms a functional amyloid signaling complex required for programmed necrosis. Cell 150(2):339–350. 22. Scheich C, Kümmel D, Soumailakakis D, Heinemann U, Büssow K (2007) Vectors for coexpression of an unrestricted number of proteins. Nucleic Acids Res 35(6):e43. 23. Zhang XJ, Matthews BW (1995) Edpdb—A multifunctional tool for protein-structure analysis. J Appl Cryst 28:624–630.
Materials and Methods
6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1411942111
Cao et al.
