(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Architecture of the biofilm-associated archaic Chaperone-Usher pilus CupE from Pseudomonas aeruginosa [1]

['Jan Böhning', 'Sir William Dunn School Of Pathology', 'University Of Oxford', 'Oxford', 'United Kingdom', 'Adrian W. Dobbelstein', 'Department Of Protein Evolution', 'Max Planck Institute For Biology Tübingen', 'Tübingen', 'Nina Sulkowski']

Date: 2023-06

Abstract Chaperone-Usher Pathway (CUP) pili are major adhesins in Gram-negative bacteria, mediating bacterial adherence to biotic and abiotic surfaces. While classical CUP pili have been extensively characterized, little is known about so-called archaic CUP pili, which are phylogenetically widespread and promote biofilm formation by several human pathogens. In this study, we present the electron cryomicroscopy structure of the archaic CupE pilus from the opportunistic human pathogen Pseudomonas aeruginosa. We show that CupE1 subunits within the pilus are arranged in a zigzag architecture, containing an N-terminal donor β-strand extending from each subunit into the next, where it is anchored by hydrophobic interactions, with comparatively weaker interactions at the rest of the inter-subunit interface. Imaging CupE pili on the surface of P. aeruginosa cells using electron cryotomography shows that CupE pili adopt variable curvatures in response to their environment, which might facilitate their role in promoting cellular attachment. Finally, bioinformatic analysis shows the widespread abundance of cupE genes in isolates of P. aeruginosa and the co-occurrence of cupE with other cup clusters, suggesting interdependence of cup pili in regulating bacterial adherence within biofilms. Taken together, our study provides insights into the architecture of archaic CUP pili, providing a structural basis for understanding their role in promoting cellular adhesion and biofilm formation in P. aeruginosa.

Author summary Many bacteria adhere to surfaces or host cells using filamentous structures termed pili that extend from the bacterial cell and anchor them to their target. Previous studies have characterised various Chaperone-Usher Pathway (CUP) pili, which are common in Gram-negative bacteria. However, little is known about the so-called archaic CUP pili, which are the most widespread type. Archaic CUP pili help maintain the architecture of multicellular bacterial aggregates termed biofilms formed by the pathogen Pseudomonas aeruginosa and many others. In this study, we present a cryo-EM structure of the archaic CUP pilus CupE from P. aeruginosa, providing a structural basis of how the CupE1 protein forms zigzag-shaped, extended pili. By imaging CupE pili on P. aeruginosa cells using electron cryotomography, we show that pili can adopt variable long-range curvature, which may help their role in providing cohesion between cells within the biofilm. Furthermore, structural modelling provides insights into the roles of minor pilin subunits encoded within the cupE operon. These results will help advance our understanding of bacterial pili structure and function.

Citation: Böhning J, Dobbelstein AW, Sulkowski N, Eilers K, von Kügelgen A, Tarafder AK, et al. (2023) Architecture of the biofilm-associated archaic Chaperone-Usher pilus CupE from Pseudomonas aeruginosa. PLoS Pathog 19(4): e1011177. https://doi.org/10.1371/journal.ppat.1011177 Editor: Matthew A. Mulvey, University of Utah, UNITED STATES Received: July 6, 2022; Accepted: February 3, 2023; Published: April 14, 2023 Copyright: © 2023 Böhning et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: The cryo-EM density map of the CupE pilus generated in this study has been deposited in the Electron Microscopy Databank (EMDB) under the accession number EMD-16683. The corresponding atomic coordinates are deposited in the Protein Data Bank (PDB) under accession code 8CIO. The cryo-EM density map of the CupE pilus 111-113 AGA mutant is available under the accession number EMD-16686. Funding: This work was supported by the Wellcome Trust (202231/Z/16/Z to TAMB), the Bert L. & N. Kuggie Vallee Foundation (Scholarship to TAMB), the Leverhulme Trust (Philip Leverhulme Prize to TAMB), the Lister Institute of Preventive Medicine (Lister Prize to TAMB), the UKRI Medical Research Council (MR/K501256/1 and MR/N013468/1: Graduate Studentship to JB; MC_UP_1201/31: core funding to TAMB). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist.

Introduction Adhesion of bacterial cells to abiotic and biotic surfaces is crucial for the colonization of new environments, including host invasion during infections and biofilm formation [1–5]. Bacterial adhesion is often mediated by proteinaceous, hair-like cell-surface structures known as pili or fimbriae [6,7]. Pili are assembled by repeated interactions of monomeric protein subunits, resulting in a filamentous structure that protrudes from the cell surface to anchor the cell to substrates [8]. Chaperone-Usher Pathway (CUP) pili are among the most widely distributed and best-characterized adhesins in Gram-negative bacteria [9]. CUP systems are typically encoded as single operons and consist of at least three components: a major pilin subunit, a periplasmic chaperone that stabilizes the pilin prior to assembly, and an outer membrane (OM) usher pore protein responsible for translocation and assembly of the pilin [10–13]. Frequently, CUP operons encode additional components, including adhesins that decorate the tip of the pilus distal to the OM [14], regulatory proteins, minor pilin subunits, and additional chaperones. CUP pilin subunits generally consist of an incomplete immunoglobulin-like (Ig-like) β-sandwich fold that lacks the final antiparallel β-strand, but contains an additional N-terminal β-strand extending away from the subunit core [15]. In the assembled pilus, each pilin subunit provides its N-terminal β-strand to the following subunit to complete its Ig-like fold in a process termed donor-strand complementation [16]. Prior to assembly, the missing β-strand within the fold is donated by the chaperone protein [15]. A tip adhesin subunit typically mediates the adhesive function of the pilus, often capping the pilus and mediating specific interactions with host receptors or other abiotic molecules [17–19]. Additional pilin subunits encoded by many CUP operons typically fulfill a specialized structural role within the pilus [20–22] or have a role in terminating pilus assembly [23]. CUP pili are phylogenetically divided into three classes: classical, alternative, and archaic [6]. The best-characterized CUP pili belong to the classical type and include the Fim (Type 1 pilus) [17,24] and Pap (P pilus) [21,25] systems of Escherichia coli, which form stiff, tubular structures important for pathogenicity [24,26]. However, available structural information on non-classical CUP pili is scarce, being limited to a crystal structure of a pilin of the archaic Csu pilus [27]. Moreover, there have been no high-resolution studies of CUP pili in their cellular environment. Archaic CUP pili, the phylogenetically oldest class, are widespread in all proteobacteria, cyanobacteria, and even in some extremophilic phyla such as Deinococcota [6]. The best-characterized archaic CUP pili include the CupE system in P. aeruginosa [28,29] and the Csu system in Acinetobacter baumannii [19,27,30]. Both of these examples are from bacterial species belonging to the ESKAPE class of multidrug-resistant pathogens for which new antibiotics must urgently be developed [31], and targeting antimicrobials against pilins has been suggested as a potential therapeutic avenue [32]. Archaic CUP pili are crucial for the formation of biofilms [28,30], which play an important role in persistent and chronic infections [33,34]. The archaic CupE pilus of P. aeruginosa is thought to have been acquired by horizontal gene transfer and evolved independently of other cup clusters in the genome (cupA-D), all of which belong to the classical type of CUP systems [28]. The cupE gene cluster encodes one major pilin subunit (CupE1) that is the main component of the pilus, two minor pilin subunits (CupE2, CupE3) whose arrangement or function within the pilus is unknown, a chaperone (CupE4), an usher (CupE5), and a tip adhesin protein (CupE6). The cupE operon is activated by a two-component system, PprA-PprB, and plays an important role in micro- and macrocolony formation and in maintaining the three-dimensional shape of the biofilm [28,35]. The expression of all CUP pili in P. aeruginosa is tightly regulated [36], and the CupE pilus is expressed as part of an exopolysaccharide-independent adhesive signature together with the type I secretion system-dependent adhesin BapA, Type IVb pili and extracellular DNA [35]. Hence, its function appears distinct from the other CUP systems in P. aeruginosa (CupA-D), which appear to be part of a different adhesive signature [36–38]. As detailed above, comparatively less is known about the structure and architecture of archaic CUP pili than their classical counterparts, despite their presence in a wide range of species [6]. To bridge this critical knowledge gap, we purified CupE pili that were overproduced in P. aeruginosa cells by deleting the mvaT gene, which encodes a CUP repressor, as well as PA2133, which encodes a phosphodiesterase in the cupA operon. We used electron cryomicroscopy (cryo-EM) and cryotomography (cryo-ET) on natively assembled CupE pili in vitro and in situ to elucidate their structural and architectural properties. We combined our structural and imaging experiments with bioinformatics, revealing important insights into the interdependence and role of CUP pili in biofilm formation of P. aeruginosa.

Discussion In this study, we present the structure of an archaic CUP pilus, revealing a zigzag architecture with extensive interactions between the donor strand and the complemented subunit. The subunit interface between the globular part of the Ig-like domains has significantly fewer inter-subunit contacts, and imaging of CupE pili in situ shows that they can adopt varied curvatures, which may aid in interactions with the biofilm matrix. This shows that the CupE pilus is not a stiff, tubular assembly like other classical rod-shaped CUP pili such as the P pilus [25], but it is also not highly flexible like the Saf pilus or other FGL-assembled CUP fimbriae [42,43]. The difference in lateral flexibility between the CupE pilus and the closely-related, but stiff, Csu pilus from A. baumannii can be explained by the number of interacting residues at the inter-subunit interface: The Csu pilus, which was characterised in a recent study by overexpression of the corresponding csu operon in E. coli [40] has increased subunit interactions within the pilus as a result of denser packing compared with CupE (S5 Fig). Interestingly, it was found in optical tweezer experiments that the Csu pilus can be extended to almost twice its length along the helical axis by breaking of the subunit interface contacts [40]. Given that CupE shares a similar architecture, it seems likely that it shares this super-elasticity, suggesting it may be a common feature of archaic CUP pili. We propose that the lateral curvature of the CupE pilus observed in the cellular environment may stem from the same properties, i.e., the result of breaking some or all non-donor-strand subunit interface contacts. AlphaFold structure predictions for minor CupE subunits CupE2 and CupE3 suggest that they can also form filaments through donor-strand exchange. While both homofilaments and heterofilaments with CupE1 are predicted to be possible (S7 Fig), only CupE1 homofilaments appear to form the more compact zigzag architecture. However, it is unclear whether the minor pilins CupE2 and CupE3 form homofilaments at certain sections of the pilus, similar to the tips in the classical Type 1 and P pili [9], or whether they are sub-stoichiometrically embedded between CupE1 subunits. Other possibilities are that they functionalize pili or fulfil an undetermined function in pilus assembly. For example, in the Pap system, the PapH subunit terminates pilus assembly [23,50]; whether an analogous protein exists in archaic CUP pili remains to be determined. Structural prediction of the CupE6 adhesin tip subunit also suggests that, in agreement with previous analyses on the Csu system [19], CupE6 contains the same subunit fold and hydrophobic surface at the tip that could interact with other hydrophobic components, thus supporting adhesion. The chemical nature of the substrate of the CupE6 pilus tips in biofilms remains enigmatic, and it is unclear how the highly hydrophobic pilus tip would be stabilized during pilus assembly, presenting an exciting direction for future inquiries. CupE and Csu, the two archetypal archaic CUP pili systems, are both involved in promoting biofilm formation [28,30]. While we study CupE pili on cells, an intrinsic limitation of our system is that single cells expressing CupE do not fully recapture the molecular sociology and crowding conditions within intact biofilms. The interaction partners of the pilus are hence unknown and warrant the focus of future imaging efforts. Since we observed isolated CupE pili forming regular mesh-like arrays in cryo-EM images (S1D Fig), this might indicate that lateral interactions of pili may occur in the crowded conditions of the biofilm matrix, similar to other biofilm matrix fibre systems [51–53]. Further studies on native cellular systems–i.e., biofilms—will be required to determine the exact mode of interaction of CupE pili with other extracellular matrix components and also with cells. Our study finds that deletion of the gene encoding the phosphodiesterase CupA6 in the cupA operon, results in increased expression of the cupE operon. A likely explanation for this is that the cupA operon negatively regulates cyclic di-GMP levels through CupA6, and that deletion of CupA6 causes higher cyclic di-GMP levels, which in turn causes cupE expression. Indeed, we find that cupA and cupE gene clusters mostly co-occur in P. aeruginosa isolates, suggesting both fulfill distinct functions during biofilm development and that their expression may be co-regulated and interdependent. Numerous CUP systems and other adhesins have been identified in P. aeruginosa, and many have been found to be of general importance for biofilm formation. This prompts the overarching question: why does P. aeruginosa have an entire arsenal of different adhesins? Are some adhesin systems co-operative, or are they expressed only under specific conditions—and if so, when? Answering these questions in future studies will greatly improve our understanding of adhesion, biofilm formation, and pathogenicity of P. aeruginosa specifically and Gram-negative bacteria in general.

Materials and methods Construction of P. aeruginosa mutants P. aeruginosa deletion mutants were created as described previously utilising the suicide plasmid pKNG101 [54]. Briefly, to engineer gene deletions in the PAO1 strain, 500 bp DNA fragments of the 5’ (upstream) and 3’ (downstream) ends of the gene of interest were obtained by PCR using PAO1 chromosomal DNA as a template. The upstream fragment was amplified with the oligonucleotides P1 and P2 while the downstream fragment was amplified using P3 and P4 (S2 Table). A third PCR step using P1 and P4 resulted in a DNA fragment with the flanking region of the gene of interest. The gene fragment was then cloned into pCR-BluntII-TOPO (Invitrogen), the sequence confirmed and sub-cloned into the pKNG101 suicide vector (S3 Table). The pKNG-derivatives were maintained in E. coli strain CC118λpir (for strain descriptions, see S4 Table) and conjugated into PAO1 using E. coli 1047 harbouring the conjugative plasmid pRK2013. pKNG101 was conjugated into P. aeruginosa as described in [54]. After homologous recombination, colonies were streaked onto agar containing 20% (w/v) sucrose and grown at room temperature for 48 hours to select for colonies that have lost the plasmid backbone. Gene deletions were verified by PCR using external primers P5 and P6 (S2 Table). Chromosomal substitution of TTTTSST to AGATSST in CupE1 was achieved by reintroduction of the previously deleted cupE1-2 fragment into PAO1 ΔpilA ΔfliC ΔmvaT ΔcupA6 ΔcupE1-2. In brief, the cupE1-2 DNA fragment was amplified using primers called ΔcupE1-2 P1 and P4 (S2 Table), subcloned into pCR-BluntII-TOPO (Invitrogen) and subjected to site directed mutagenesis using CupE1-AGATSST Fw and Rev primers (S2 Table). After the sequence was confirmed, the fragment was cloned into pKNG101 and conjugation was conducted as described above. Mutation was verified by PCR using external primers ΔcupE1-2 P5 and P6 and clones with cupE1-2 fragment reintroduced were sequenced. Isolation of CupE pili An overnight culture of P. aeruginosa PAO1 ΔmvaT ΔcupA6 ΔpilA ΔfliC, grown in lysogeny broth (LB) medium at 37°C with agitation at 180 revolutions per minute (rpm), was used to plate lawns on agar plates, and incubated overnight at 37°C. Bacterial lawns were scraped and resuspended in 1x phosphate-buffered saline (PBS). The resulting suspension was vortexed for 90 seconds to promote dissociation of pili from the cell surface. Cells were then centrifuged at 4,500 relative centrifugal force (rcf) for 20 minutes, and the supernatant centrifuged again 3–4 times at 16,000 rcf to remove remaining cells and cellular debris. 500 mM NaCl and 3% (w/v) PEG-6000 were added to the supernatant, and pili were precipitated on ice for 1 hour. Precipitated pili were collected via centrifugation for 30 minutes at 16,000 rcf. For cryo-EM, precipitated pellets were combined and precipitated again in the same manner and resuspended in 1x PBS to produce the final product. Negative stain electron microscopy 2.5 μl of sample was applied to a glow-discharged carbon support grid (TAAB), blotted, washed three times with water, and stained using three 20 μl drops of 2% (w/v) uranyl acetate and allowed to air-dry. Negatively stained grids were imaged on a Tecnai T12 microscope. Cryo-EM and cryo-ET sample preparation For cryo-EM grid preparation of purified CupE pili, 2.5 μl of the sample was applied to a freshly glow-discharged Quantifoil R 2/2 Cu/Rh 200 mesh grid and plunge-frozen into liquid ethane using a Vitrobot Mark IV (ThermoFisher) at 100% humidity at an ambient temperature of 10°C. For tomography sample preparation of PAO1 ΔpilA ΔfliC ΔmvaT ΔcupA6, a bacterial lawn from an overnight LB agar plate incubated at 37°C without antibiotics was resuspended in PBS, and 10 nm Protein-A-gold beads (CMC Utrecht) were added as fiducials prior to plunge-freezing. Cryo-EM and cryo-ET data collection Cryo-EM data was collected in a Titan Krios G3 microscope (ThermoFisher) operating at an acceleration voltage of 300 kV, fitted with a Quantum energy filter (slit width 20 eV) and a K3 direct electron detector (Gatan). Images were collected in super-resolution counting mode using a physical pixel size of 1.092 Å/pixel for helical reconstruction of CupE pili and 3.489 Å/pixel for cellular tomography data. For helical reconstruction of CupE, movies were collected as 40 frames, with a total dose of 45–46 electrons/Å2, using a range of defoci between -1 and -2.5 μm. For the wild-type CupE pilus dataset, 11,584 movies were collected; for the 111-113 AGA dataset, 4,665 movies were collected. Cryo-ET tilt series of PAO1 ΔpilA ΔfliC ΔmvAT ΔcupA6 cells were collected using a dose-symmetric tilt scheme as implemented in SerialEM [55], with a total dose of 121 electrons/Å2 per tilt series and defoci of -8 to -10 μm, and with ±60° tilts of the specimen stage at 1° tilt increments. Cryo-EM processing Helical reconstruction of CupE pili was performed in RELION 3.1 [56–58]. Movies were motion-corrected and Fourier-cropped using the RELION 3.1 implementation of MotionCor2 [59], and CTF parameters were estimated using CTFFIND4 [60]. Initial helical symmetry of CupE pili was estimated through indexing of layer lines and counting the number of visible subunits along the pilus. Three-dimensional classification was used to identify a subset of particles that supported refinement to 3.5 Å resolution. For final refinement, CTF multiplication was used for the final polished set of particles [57,61,62]. Symmetry searches were used during reconstruction, resulting in a final rise of 33.12 Å and a right-handed twist per subunit of 214.56°. Resolution was estimated using the gold-standard Fourier Shell Correlation (FSC) method as implemented in RELION 3.1. Local resolution measurements were also performed using RELION 3.1. Model building and refinement Manual model building of the CupE1 subunit was performed in Coot [63] as follows. A homology model based on the structure of A. baumannii CsuA/B (RCSB 6FQA) was calculated using MODELLER [64] and this homology model was fit into the cryo-EM density as a rigid body. Residues of the homology model that were inconsistent with the density, including the N-terminal donor strand, were deleted and manually rebuilt. The initially built model was subjected to real-space refinement against the cryo-EM map within the Phenix package [65,66]. Five subunits of CupE1 were built and used for final refinement. Non-crystallographic symmetry between individual CupE1 subunits was applied for all refinement runs. Model validation including map-vs-model resolution estimation was performed in Phenix. Tomogram reconstruction Tilt series alignment via tracking of gold fiducials was performed using the etomo package as implemented in IMOD [67]. Tomograms were reconstructed with WBP in IMOD or SIRT in Tomo3D [68,69]. Deconvolution of tomograms using the tom_deconv.m script [70] was performed for visualisation purposes. Subtomogram averaging Subtomogram averaging of pili on cells was performed in RELION 4 [41], employing helical reconstruction [56]. A cylindrical reference was used to avoid bias. Helical symmetry was applied to enhance the signal during particle alignment. The map presented in Figs 3 and S6 is unsymmetrized. Data visualisation and quantification Cryo-EM images were visualized in IMOD. Fiji [71] was used for bandpass and Gaussian filtering, followed by automatic contrast adjustment. Atomic structures and tomographic data were displayed in ChimeraX [72]. Segmentation of tomograms was performed manually in IMOD. Quantification of cell surface filaments in the ΔcupA6 mutant was performed through manual annotation in 30 randomly acquired negative stain EM images targeted on cells located at low magnification. Atomic models are shown in perspective view, except for S3C Fig, which is shown in orthographic view. Hydrophobic surfaces were calculated in ChimeraX using the in-built mlp function. Difference maps were calculated using EMDA [73] with maps lowpass-filtered to the same resolution (4.2 Å). Angles between subunits (S6E Fig) were determined by spline-interpolating points every 33.1 Å along the path of the pilus, as segmented in IMOD, calculating normal lines at each point, and determining the angles between the normal of each point. Bioinformatic analysis Sequence data were downloaded from the Pseudomonas Genome Database version 20.2 [46] and filtered to exclude incomplete genomes. Searches were performed against every single strain using PSI-BLAST [74], with the CupA3 (PA2130 NCBI locus tag), CupB3 (PA4084), CupC3 (PA0994), and CupE5 (PA4652) proteins from the reference strain P. aeruginosa PAO1 as queries. Since the CupD system is missing in the strain PAO1, the CupD3 usher (PA14_59735) from strain UCBPP-PA14 was used as the query. To obtain unambiguous assignment of genes to CUP proteins, output data was further filtered with custom scripts and probable sequencing errors were corrected. Structure predictions were performed with AlphaFold-Multimer version 2.1.1 [44,45], with sequences from the reference genome PAO1 as queries. Filaments were predicted without the signal peptide, which was predicted using SignalP [75]. The multiple sequence alignments (MSAs) used for the structure inference were built with the standard AlphaFold pipeline and the “reduced_dbs” preset. Template modelling was enabled and structures were inferred with eight MSA recycling iterations and all five different model parameter sets. After prediction, the models were ranked by the pTM score and only the highest-ranking model was selected. The PAE-value-plot for each structure is shown in S8 Fig. All predictions were performed using the high-performance computer “Raven”, operated by the Max-Planck Computing & Data Facility in Garching, Munich, Germany. The multiple sequence alignment shown in S4 Fig was obtained by first calculating an initial alignment using PROMALS3D [76] in default settings and then curating it manually.

Acknowledgments The authors would like to thank Dr. Thomas Clamens for help with strain generation.

[END]
---
[1] Url: https://journals.plos.org/plospathogens/article?id=10.1371/journal.ppat.1011177

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.