Differential homotypic and heterotypic interactions of antigen 43 (Ag43) variants in autotransporter-mediated bacterial autoaggregation

Ageorges, Valentin; Schiavone, Marion; Jubelin, Grégory; Caccia, Nelly; Ruiz, Philippe; Chafsey, Ingrid; Bailly, Xavier; Dague, Etienne; Leroy, Sabine; Paxman, Jason; Heras, Begoña; Chaucheyras-Durand, Frédérique; Rossiter, Amanda E.; Henderson, Ian R.; Desvaux, Mickaël

doi:10.1038/s41598-019-47608-4

Download PDF

Article
Open access
Published: 31 July 2019

Differential homotypic and heterotypic interactions of antigen 43 (Ag43) variants in autotransporter-mediated bacterial autoaggregation

Valentin Ageorges¹,
Marion Schiavone^2,3,
Grégory Jubelin¹,
Nelly Caccia¹,
Philippe Ruiz ORCID: orcid.org/0000-0002-5149-9775¹,
Ingrid Chafsey¹,
Xavier Bailly ORCID: orcid.org/0000-0001-6959-7974⁵,
Etienne Dague⁴,
Sabine Leroy¹,
Jason Paxman ORCID: orcid.org/0000-0002-1418-5910⁶,
Begoña Heras⁶,
Frédérique Chaucheyras-Durand^1,2,
Amanda E. Rossiter⁷,
Ian R. Henderson^7,8 &
…
Mickaël Desvaux¹

Scientific Reports volume 9, Article number: 11100 (2019) Cite this article

3056 Accesses
15 Citations
7 Altmetric
Metrics details

Subjects

Abstract

Antigen 43 (Ag43) is a cell-surface exposed protein of Escherichia coli secreted by the Type V, subtype a, secretion system (T5aSS) and belonging to the family of self-associating autotransporters (SAATs). These modular proteins, comprising a cleavable N-terminal signal peptide, a surface-exposed central passenger and an outer membrane C-terminal translocator, self-recognise in a Velcro-like handshake mechanism. A phylogenetic network analysis focusing on the passenger revealed for the first time that they actually distribute into four distinct classes, namely C1, C2, C3 and C4. Structural alignment and modelling analyses demonstrated these classes arose from shuffling of two different subdomains within the Ag43 passengers. Functional analyses revealed that homotypic interactions occur for all Ag43 classes but significant differences in the sedimentation kinetics and aggregation state were present when Ag43^C3 was expressed. In contrast, heterotypic interaction occurred in a very limited number of cases. Single cell-force spectroscopy demonstrated the importance of specific as well as nonspecific interactions in mediating Ag43-Ag43 recognition. We propose that structural differences in the subdomains of the Ag43 classes account for different autoaggregation dynamics and propensities to co-interact.

Proactive vaccination using multiviral Quartet Nanocages to elicit broad anti-coronavirus responses

Article Open access 06 May 2024

Pseudomonas aeruginosa: pathogenesis, virulence factors, antibiotic resistance, interaction with host, technology advances and emerging therapeutics

Article Open access 25 June 2022

Streptomyces umbrella toxin particles block hyphal growth of competing species

Article Open access 17 April 2024

Introduction

In lipopolysaccharidic (LPS) diderm bacteria (archetypical Gram-negative bacteria), nine protein secretion systems (T1SS-T9SS) are currently recognised as enabling protein transport across the outer membrane (OM), whereas the Sec and Tat export pathways enable protein translocation across the cytoplasmic membrane, also called the inner membrane (IM)^1,2,3. Among those, the T5SS represents the simplest and most widely used mechanism for protein secretion in LPS-diderm bacteria^4,5,6,7, where proteins are translocated across the OM via a β-barrel with the assistance of some periplasmic chaperones (e.g. Skp, SurA or DegP), the BAM (β-barrel assembly machinery) and TAM (translocation and assembly module) complexes along the process^8,9,10,11. Based on phylogenetic analyses and structural features, the T5SS is subdivided into the (i) classical autotransporters (T5aSS), (ii) two-partner passenger-translocators (T5bSS), (iii) trimeric autotransporters (T5cSS), (iv) hybrid autotransporters (T5dSS), and (v) inverted autotransporters (T5eSS)^11,12,13,14.

A classical autotransporter (AT) is a modular protein constituted of three main regions, i.e. (i) a N-terminal signal peptide (SP), (ii) a central passenger, and (iii) a translocator at the C-terminus^15,16. As such, AT export via the Sec translocase is mediated by the SP before the passenger is secreted through the β-barrel pore formed by the translocator in the OM. Once secreted, the passenger can have three different fates as it may be (i) displayed at the bacterial cell surface, (ii) cleaved from the translocator and released as an exoprotein into the extracellular milieu, or (iii) processed but remain associated to the cognate translocator¹⁰. The functions of ATs are quite diverse and can be broadly categorised as (i) self-associating ATs (SAATs), (ii) protease ATs, (iii) phosphatase/hydrolase ATs, (iv) lipase/esterase ATs, (v) vacuolating ATs, (vi) adhesin ATs, and (vii) ATs of unknown function^3,10. SAATs promote bacterial aggregation by cell self-association, a phenomenon called autoaggregation^17,18. Besides flocculation and sedimentation of bacterial cells, autoaggregation is an important mechanism with consequences for bacterial virulence and infection due to biofilm formation, protection from environmental stresses, antibiotic resistance and defence against predators¹⁸. The antigen 43 (Ag43) is a prototypical member of the SAATs; this category of ATs also include AIDA (adhesin involved in diffuse adherence) and TibA (toxigenic invasion locus b)¹⁷. Besides homotypic interactions, heterotypic interactions were demonstrated between AIDA-Ag43, AIDA-TibA and Ag43-TibA^17,19. By promoting interspecies cell-to-cell contact, these interactions would have important ecophysiological roles in inducing co-operative multispecies biofilms that allows for more successful microbial colonisation of various niches²⁰.

Besides the basic and common modular architecture of an AT, Ag43 exhibits some additional features. At the N-terminus, Ag43 displays a highly conserved extended SP region (ESPR) that slows the rate of export across the IM through the Sec-YidC pathway to prevent protein misfolding in the periplasm and consequently sustain the secretion-competent state of the passenger across the OM through the translocator^21,22. Of note, the ESPR is not exclusive to Ag43 as it occurs in many different, but not all, ATs. N-terminal to the translocator, Ag43 exhibits a type 1 autochaperone domain (AC1), which is considered as essential for folding of β-helical passengers^3,23. Like most ATs, the Ag43 passenger folds into a single-stranded right-handed parallel β-helix with triangular-shaped coil cross-sections^3,24. Most recently, the crystal structure of the Ag43 from uropathogenic E. coli (UPEC) CFT073 (Ag43a) was solved and revealed the passenger displayed a twisted L-shape β-helical architecture²⁵. The L-shaped conformation of the passenger further appeared as a requirement for driving the formation of cell aggregates and self-association in a Velcro-like handshake mechanism. While the function of Ag43 was primarily investigated in Escherichia coli K12²⁴, its implication in autoaggregation, adhesion and biofilm formation was further confirmed in some pathogenic species, namely enterohaemorrhagic E. coli (EHEC) and UPEC^26,27. The Ag43 is encoded by two identical alleles in EHEC O157:H7 EDL933 but by two different alleles in UPEC CFT073; the UPEC Ag43a variant mediates a strong autoaggregation and biofilm formation phenotypes, whereas the Ag43b variant is a less efficient aggregating and biofilm forming factor²⁷. Phylogenetic analyses had revealed the existence of two subfamilies (SF) of Ag43, namely SF-I and SF-II²⁴.

While homotypic Ag43 interactions have previously been noted, heterotypic interactions of Ag43 variants had not been extensively characterised as yet, it prompted us to investigate these aspects in a systematic study. In a cooperative evolution hypothesis that “birds of a feather flock together”, different combinations of Ag43 variants might select bacterial interactions with siblings to the exclusion of others. From phylogenetic analyses of Ag43 passengers, it first appeared that the grouping was not dichotomous but clustered into four different classes associated to structural differences. Following the expression of representative Ag43 members, different propensities of the variants to induce autoaggregation were evidenced. By monitoring the formation of cell aggregates by epifluorescence microscopy and flow cytometry, homotypic and heterotypic interactions between different Ag43 variants were for the first time identified and characterised. Moreover, biophysical studies by atomic force microscopy (AFM) revealed different dynamics of Ag43 interactions suggesting that structural differences in the N-terminal and C-terminal subdomains of the Ag43 passengers can modulate the association of the different variants.

Results

Ag43 passengers cluster into four distinct phylogenetic classes with combination of two different subdomains

The functional properties of the modular protein Ag43 are determined by the passenger, which protrudes from the bacterial OM upon translocation. To apprehend the diversity of Ag43, a phylogenetic network analysis of the passenger was applied to a dataset of 106 distinct passengers generated from previously reported Ag43 sequences in the literature^24,28 (Supplementary Table S1). The network obtained defined for four main classes of passengers (C1 to C4) supported by robust splits (Fig. 1A).

C1 harbours the Ag43 passenger from E. coli K12 MG1655 (UPI00003B25EE), which is among the first identified and the most investigated Ag43²⁴. C2 includes the Ag43 passenger from EHEC O157:H7 EDL933 (UPI00000D07E5), which is the first Ag43 investigated in a pathogenic E. coli²⁶. C3 harbours the Ag43a (UPI00000E4A66) and Ag43b (UPI00000E49D2) from UPEC CFT073, the former being the first and only structurally characterised Ag43 passenger (PDB: 4KH3)²⁵. To date, no Ag43 homologues from C4 have been functionally or structurally characterised. The original SF-I grouping corresponds to the gathering of C1 and C3, whereas SF-II grouping reflects C2 and C4²⁴. The initial clustering correlated with the most obvious difference among the Ag43 passengers, their sizes; while Ag43 passengers from C1 and C3 contain around 500 amino acids, C2 and C4 homologues are smaller with about 430 amino acids (Supplementary Table S1).

Our updated phylogenetic network further revealed a strong ambiguity in the relatedness patterns among Ag43 passengers from C1, C2, C3, and C4, illustrated by the diamond shape of the network with almost no specific divergence. Therefore, a more detailed analysis of subdomains of the passengers was performed to identify the source of this ambiguity. Homologies to pectin lyase fold (IPR011050) overlapping with P22 tailspike protein (IPR012332) and AT adhesin (IPR030930) domains were systematically predicted for all Ag43 passengers, highlighting the β-helix topology encountered in SAATs (Supplementary Table S1). Modelling of the 3D structure of the Ag43 passengers confirmed that they are all predicted to exhibit a right-handed parallel β-helix fold with an L-shaped structure as originally observed in Ag43^{C3-UPI00000E4A6625} (Fig. 2A). Following the analysis of the coil cross-sections along the Ag43 passengers²⁹, it appeared the N-terminal part, namely the SL (for stem of the letter L) region, systematically exhibited triangular (T) β-solenoids of subtype 4 (T4) (Fig. 2 and Supplementary Table S1). A phylogenetic network analysis of the SL subdomain identified two robust and unambiguously differentiated clusters of sequences, SL of type 1 (SL1) and SL2 (Fig. 1B). The two subtypes were related by a single branch, and remaining splits were located within SL2. Sequences from SL1 and SL2 subdomains had mean percentages of similarity of 93 and 88%, respectively, but these percentages fell at 75% when comparing SL1 versus SL2 (Supplementary Table S1). While SL1 clustered subdomains corresponding to Ag43^C1 and Ag43^C4 passengers, SL2 clustered subdomains from Ag43^C2 and Ag43^C3 passengers. Nevertheless, the SL region appeared slightly longer by 15–17 amino acids in Ag43^C1 and Ag43^C3 passengers.

The elbow region joining (EJ) the upper and lower parts of the L-shape was always bordered by β-hairpin motifs with three-rung parallel T5 β-solenoids in-between, which exhibited a highly conserved sequence in the first rung, namely VVLE[N/S]GGRLDVL. The EJ region reveals a remarkably conserved size of 85 amino acids. The phylogenetic network analysis of the EJ subdomain showed a different relatedness pattern compared to the SL subdomain. Two robust sequence subtypes were clustered, namely (i) EJ of type 1 (EJ1), which included Ag43 passenger subdomains from C1 and C3, and (ii) EJ2, which clustered subdomains from the C2 and C4 (Fig. 1C). In average, sequences from EJ1 and EJ2 subdomains had percentages of similarity of 97 and 94%, respectively, but these percentages fell at 53% when comparing EJ1 versus EJ2 (Supplementary Table S1).

The BL (for bottom of the letter L) subdomain in the C-terminal part of the passengers is composed of T5-type β-solenoids (Fig. 2 and Supplementary Table S1). The phylogenetic network analysis of the BL was coherent with the analysis of the EJ subdomain. Two robust sequence subtypes were clustered: (i) BL of type 1 (BL1), which included Ag43 passengers subdomains from C1 and C3, and (ii) BL2, which clustered subdomains from C2 and C4 (Fig. 1D). Sequences from BL1 and BL2 subdomains had percentages of similarity from and above 95 and 93% (mean values of 99 and 98%), respectively, but these percentages fell below 30% (28% in average) when comparing SL1 versus SL2 (Supplementary Table S1). Furthermore, a difference of 53 amino-acids among the length of BL subdomains was observed. It accounted for most of the observed variation in the length of Ag43 passengers between C1/C3 over C2/C4 (Supplementary Table S1). This difference indeed resulted from a truncation at the C-terminus of BL2 compared to BL1.

These data suggest that the two SL phylogenetic clusters and the two coherent phylogenetic clusters of EJ and BL were shuffled during the evolution of Ag43 proteins explaining the observed ambiguity in relatedness pattern of the passengers (Fig. 1A). It resulted in four combinations of Ag43 passengers subdomains: (i) SL1-EJ1-BL1 corresponding to Ag43^C1, (ii) SL2-EJ2-BL2 corresponding to Ag43^C2, (iii) SL2-EJ1-BL1 corresponding to Ag43^C3, and (iv) SL1-EJ2-BL2 corresponding to Ag43^C4.

Autoaggregation occurs through homotypic interactions for all Ag43 variants but the size of the aggregates is significantly much smaller with Ag43^C3

Our new classification of Ag43 delineated the proteins into four classes based on sequence similarity and consequently structural differences. We hypothetised that these variations would result in functional differences. Therefore, we investigated the autoaggregation phenotype of three selected Ag43 homologues from each of the four classes (Fig. 2 and Table 1) when expressed in the non-aggregative E. coli SRUD³⁰. When bacterial cell cultures were left standing statically, all bacterial cells expressing the Ag43 flocculated and sedimented at the bottom of tubes contrary to the control SRUD pBAD (with empty vector pBAB/Myc-HisA, i.e. Ag43⁻) (Fig. 3A,B) (Table 1). No significant differences could be observed for the sedimentation kinetics of bacterial cells expressing Ag43 from C1, C2 and C4 with all three having compact sedimentation pellets. Microscopic observations indicated the flocculation and sedimentation of the bacterial cultures resulted from bacterial cell autoaggregation as no cell aggregates could be observed with SRUD pBAD, control (Fig. 3C).

Table 1 Genotype and phenotype of most relevant E. coli strains used in the study.

Full size table

In contrast, strains expressing the three C3 homologues (Fig. 2), namely Ag43^{C3-UPI00000E4A66}, Ag43^{C3-UPI00000E49D2} and Ag43^{C3-UPI0001EB6ED2}, showed a significant delay in the autoaggregation kinetics compared to other Ag3 expressing strains (Fig. 3A). In addition, the sedimentation pellets were looser and cell aggregates appeared smaller, with the presence of individual single cells (Fig. 3B,C). Within C3, some significant differences were observed as bacterial cultures of SRUD expressing Ag43^{C3-UPI00000E4A66} sedimented significantly quicker than those expressing Ag43^{C3-UPI00000E49D2} or Ag43^{C3-UPI0001EB6ED2}. To confirm this slower autoaggregation phenotype for bacterial cells expressing Ag43^C3, three additional homologues from this class were tested; in each case, autoaggregation was slower (Fig. 3A).

To further characterise the homotypic interactions that are responsible for the autoaggreagtion phenotype, one prototypical (PT) representative was selected from each class (Table 1), namely Ag43^C1-PT (UPI00003B25EE), Ag43^C2-PT (UPI00000D07E5), Ag43^C3-PT (UPI00000E49D2) and Ag43^C4-PT (UPI0001E6D993). Fluorescence microscopy revealed that in the absence of Ag43 (Ag43⁻, i.e. SRUD-Green pBAD and SRUD-Red pBAD) no cell aggregates could be observed (Fig. 4A). In contrast aggregates were detected for cells expressing Ag43 variants from each group (Fig. 4B–E). The combination of Ag43⁻ red-fluorescent-tagged cells mixed with Ag43⁺ green-fluorescent-tagged cells (i.e. SRUD-Green pBAD-Ag43) systematically exhibited only one type of bacterial cell aggregate, consisting uniquely of green-fluorescent-tagged cells where only a few Ag43⁻ cells could be observed as trapped within the aggregates (Fig. 4B–E). When green- and red-fluorescent cells expressing the same Ag43 were mixed equally, both fluorescent cells were found in the aggregates (Fig. 4F–I). These results confirmed that all tested Ag43 proteins, Ag43^C1-PT, Ag43^C2-PT, Ag43^C3-PT or Ag43^C4-PT, enable homotypic interactions inducing bacterial autoaggregation. However, examination of the obtained micrographs seemed to indicate that more individual cells were present when Ag43^C3-PT was expressed (Fig. 4H).

To gain quantitative insight, fluorescent nonaggregative and aggregative E. coli cells were analysed using flow cytometry. SSC-A (side scatter area) and FSC-A (forward scatter area) parameters are commonly used to discriminate different types of cells within a heterogeneous population with respect to granularity and cell size, respectively. Here, SSC-A over FSC-A dot plots allowed differentiating single cells from aggregated cells. As shown in Supplementary Fig. S3, flow cytometry analysis of the non-aggregative E. coli SRUD pBAD revealed a FCS-A/SSC-A pattern that is clearly distinct from the one obtained using the aggregative E. coli SRUD expressing Ag43^C1-PT. Comparison of these 2 samples allowed discriminating single cells from cell aggregates, the latter forming a comet tail in the upper part of the SSC-A over FSC-A representation (Supplementary Fig. S3). In agreement with microscopic examination, no cell aggregates were observed in the absence of Ag43, whereas cells expressing Ag43^C1-PT, Ag43^C2-PT, Ag43^C3-PT and Ag43^C4-PT formed aggregates (Supplementary Fig. S4). Furthermore, when green- with blue-fluorescent-tagged SRUD cells expressing the same Ag43 allele were combined, mixed aggregates formed (Fig. 5A); these represented around 90% of the total aggregates (Fig. 5B). As determined by flow cytometry analyses, quantification of the concentration of single individual cells showed a significant difference for cells expressing Ag43^C3-PT compared to other Ag43 variants (Fig. 5C). Further, the shorter comet tail for E. coli SRUD Ag43^C3-PT was indicative of smaller aggregates (Fig. 5A) and the mean size of the aggregates was confirmed to be significantly smaller for cells expressing Ag43^C3-PT compared to any of the other Ag43 variants (Fig. 5D). To ensure the differences reported for Ag43^C3-PT were not influenced by the protein expression levels, Western blot analyses were performed and indicated the Ag43 variants were expressed at similar levels in E. coli SRUD and displayed at the OM (Fig. S5).

Autoaggregation occurs only for some combinations of heterotypic interactions

To study heterotypic interactions, all six possible combinations of interactions between the prototypical Ag43 proteins from C1 to C4 were tested (Fig. 6). Fluorescent microscopic examinations revealed three major trends (i) large clumps of bacteria containing only red- or green-fluorescent-tagged cells, suggesting only homotypic interactions occur (Fig. 6A,B), (ii) large aggregates containing only one type of fluorescent-tagged cell with smaller aggregates of the other tagged cells associated with the periphery, suggesting a weak propensity for heterotypic interactions (Fig. 6C,D), and (iii) clumps containing both green and red fluorescent cells suggesting the occurrence of heterotypic interactions (Fig. 6E,F).

Flow cytometry analyses confirmed the absence of heterotypic interaction between cells expressing Ag43^C1-PT and Ag43^C2-PT or cells expressing Ag43^C2-PT and Ag43^C4-PT (Fig. 7A) as the percentage of mixed aggregates were systematically estimated at less than 3% (Fig. 7B). Similarly, flow cytometry analyses confirmed the absence of significant heterotypic interactions as observed by epifluorescent microscopy for combinations of cells expressing Ag43^C1-PT/Ag43^C3-PT or Ag43^C2-PT/Ag43^C3-PT, with a percentage of mixed aggregates estimated at less than 5% (Fig. 7A,B). However, combinations of cells expressing Ag43^C1-PT/Ag43^C4-PT or Ag43^C3-PT/Ag43^C4-PT showed significant heterotypic interactions with mixed cell aggregates estimated at just above 50% and 12%, respectively (Fig. 7A,B). In both cases, the percentage of mixed cell aggregates were significantly lower than the corresponding homotypic combinations (Fig. 5B). For all heterotypic combinations, no significant differences could be observed in the concentration of single cells (Fig. 7C). Heterotypic combinations resulted in aggregates with different sizes, some showing a mean size significantly larger than the corresponding homotypic combinations (i.e. Ag43^C1-PT/Ag43^C2-PT, Ag43^C1-PT/Ag43^C3-PT, and Ag43^C1-PT/Ag43^C4-PT) (Figs 5D and 7D). Interestingly, the association of cells expressing Ag43^C3-PT with cells expressing any other class of Ag43 resulted in significantly lower concentration of single individual cells (Figs 5C and 7C) as well as significantly larger cell aggregates (Figs 5D and 7D) when compared to homotypic interactions.

Overall, heterotypic interactions between Ag43 from different classes is rather compromised, since no heterotypic interactions were observed in four out of the six possible combinations of prototypical Ag43 from C1 to C4. When heterotypic interactions occurred, the percentages of mixed aggregates were significantly lower than for homotypic interactions. Nonetheless, the highest percentage of mixed aggregates resulted from cells expressing Ag43^C1-PT/Ag43^C4-PT, both passengers consisting of the SL1 subdomain (Fig. 2B). Conversely, and despite exhibiting the same SL2 subdomain, no heterotypic interactions could be observed for cells expressing Ag43^C2-PT/Ag43^C3-PT. Surprisingly, some heterotypic interactions were observed between cells expressing Ag43^C3-PT/Ag43^C4-PT, which would involve interactions between SL1 and SL2 subdomains.

Ag43-mediated bacterial autoaggregation involves both specific and nonspecific interactions

Our data suggest the Ag43-Ag43 interactions could be stronger or weaker for some pairings. To quantify this and reinforce the hypothesis, single cell-force spectroscopy experiments were carried out to gain further insight in the adhesive forces at play in the autoaggregation process. First, the self-interaction between cells expressing different Ag43 variants was investigated by recording multiple force curves to measure the frequencies of adhesive events and adhesion forces between two E. coli cells (Table 2). For homotypic interactions, Ag43^C2-PT expressing cells exhibited the highest percentage of adhesive events, whereas bacterial cells expressing Ag43^C3-PT exhibited the lowest (approximately 72 and 28%, respectively). For homotypic associations of Ag43^C1-PT cells, the distribution of the adhesion forces show the highest most probable rupture force with 113 pN (Table 2 and Supplementary Fig. S6). Interestingly, adhesion forces for Ag43^C2-PT homotypic interactions showed a bimodal distribution with two maxima at 91 and 223 pN, respectively, which suggest that two different interaction conformations can exist for this protein (Table 2 and Supplementary Fig. S6). Force curves show that multiple adhesion events took place during the self-interaction of Ag43^C1-PT or Ag43^C2-PT expressing cells, whereas single interaction events were mostly recorded for Ag43^C3-PT or Ag43^C4-PT self-interaction (Fig. 8A). For cells expressing Ag43^C1-PT or Ag43^C2-PT but not Ag43^C3-PT or Ag43^C4-PT, statistical analysis of the number of rupture events detected per force curves evidenced a maximum of 10 rupture events, indicating sequential unfolding of these homotypic interactions (Fig. 8B). Using the WLC (worm-like chain) model (allowing to measure the contour length of the unfolded molecule, which is defined as the length of the fully extended protein), the distribution of adhesion forces as a function of the contour length of unfolded Ag43 displayed different type of interactions (Fig. 8C). For Ag43^C2-PT, Ag43^C3-PT and Ag43^C4-PT, the presence of adhesive events associated with unfolding at contour length of >400 nm and rupture forces (50–200 pN) suggests the presence of specific interactions between two cells via multiple electrostatic and/or hydrogen bonds. At the same time, adhesive events with contour length of <100 nm were observed for all Ag43 variants and suggests non-specific hydrophobic interactions. Taking together, these observations could explain the highest frequency of adhesive events measured for Ag43^C2-PT.

Table 2 Percentage of adhesive events and values of the rupture force measured by single-cell force spectroscopy from bacteria expressing different Ag43 variants.

Full size table

As mixed cell aggregates were observed between cells expressing Ag43^C4-PT and (i) Ag43^C1-PT, as well as at lesser extent with (ii) Ag43^C3-PT, but not with (iii) Ag43^C2-PT, these heterotypic interactions were further investigated. A high frequency of adhesive events reaching approximately 74% was recorded between cells expressing Ag43^C4-PT and Ag43^C1-PT (Table 2). Conversely, this frequency remained moderate between cells expressing Ag43^C4-PT and Ag43^C3-PT (approximately 32%) and as expected extremely poor between Ag43^C4-PT and Ag43^C2-PT (approximately 4%), the latter indicative of an absence of significant interactions. Interestingly, the distribution of adhesion forces in heterotypic interactions showed two maxima corresponding to a bimodal distribution indicating the presence of two Ag43 binding conformations (Supplementary Fig. S7). For heterotypic interactions between cells expressing Ag43^C4-PT and Ag43^C1-PT, multiple rupture events were observed with a maximum of 21 rupture events (Fig. 9A,B). Additionally, Ag43^C1-PT and Ag43^C4-PT heterotypic association seemed to also exhibit both non-specific hydrophobic and several specific interactions via multiple electrostatic and/or hydrogen bonds (Fig. 9C). In summary, whenever the occurrence of homotypic or heterotypic interactions, AFM analyses emphasised the important role of both specific and nonspecific interactions in the Ag43-mediated bacterial autoaggregation.

Discussion

Previous phylogenetic analysis had simply divided Ag43 homologues into two distinct clusters, namely SF-I and SF-II²⁴. Our further phylogenetic network analysis on the functional Ag43 passenger domains responsible for bacterial autoaggregation, has now revealed for the first time the existence of four distinct Ag43 phylogenetic classes, C1 to C4. Despite these divisions, structural modelling predicted that all Ag43 passengers form a similar L-shaped turn, the passenger structure themselves being subdivided into SL, EJ and BL subdomains. Ag43 passengers consistently displayed overall right-handed parallel β-helix fold with the exception of the two hairpins (each formed of two antiparallel β-sheets) delimiting the EJ subdomain. While there was evidence of structural variation reported in the literature^30,31, the present investigation evidenced that each SL, EJ and BL subdomain belong to two distinct phylogenetic types, whereby combinations of SL with EJ-BL subdomains divide the four different classes of Ag43 passengers. This new finding represents a major breakthrough in the field as it provides a rational framework for directed functional investigations of Ag43.

With this new understanding of the diversity of Ag43, the autoaggregation phenotype was reinvestigated based on this updated Ag43^C1, Ag43^C2, Ag43^C3 and Ag43^C4 classification. Following bacterial cell expression of several Ag43 from each of these classes, the most striking difference was based in the significant delay of the sedimentation kinetics for Ag43 proteins belonging to the class 3, compared to the Ag43^C1, Ag43^C2 and Ag43^C4. While phenotypical analyses using different Ag43 have been reported in the scientific literature^30,31, no rationale had been previously described. Further, some contradictory results have also been conveyed²⁴ that could be attributed to differences in critical parameters in the experimental set-up, such as the culture media (e.g. complex over chemically defined) or the bacterial genetic background (e.g. the use of E. coli laboratory strain for Ag43 expression still containing agn43 and/or fim). Here, using the nonaggregative E. coli SRUD strain and the same growth medium, the Ag43a from UPEC CFT073 (UPI00000E4A66) appeared to exhibit a slow sedimentation as originally reported³¹. Together with Ag43b from UPEC CFT073 (UPI00000E49D2) and other Ag43^C3, their sedimentation kinetics were significantly slower than Ag43^C1, Ag43^C2 or Ag43^C4. Further, these aggregation experiments also revealed differences within C3 with Ag43a showing the strongest sedimentations properties in this class compared to other proteins like Ag43b, which showed weaker cell settling properties. Therefore, the latter Ag43 homologue better represents the phenotypical properties of this class and could be considered a prototypical Ag43^C3. Most of the current understanding on Ag43 stems from the work done on the prototypical Ag43^{C1-UPI00003B25EE} from E. coli K12 MG1655, which has a SL1-EJ1-BL1 passenger. To date, only one article has been reported characterising Ag43^{C2-UPI00000D07E5} from EHEC O157:H7 EDL933, which presents a SL2-EJ2-BL2 passenger²⁶. Our work is the first to report the characterisation of Ag43 homologues from C4, such as Ag43^{C4-UPI0001E6D993} from enterotoxigenic E. coli O78:H11 H10407, which exhibits a SL1-EJ2-BL2 passenger³². We show that bacterial cells expressing different Ag43 variants were able to induce autoaggregation through homotypic interactions as supported by microscopic examination and flow cytometry analyses (Fig. 10). However, differences were observed among the four Ag43 classes, with Ag43^C3-PT showing slower sedimentation kinetics, looser autoaggregation pellets and significantly smaller aggregates associated with lower rupture force revealed by AFM analyses. With regard to single-molecule force spectroscopy analysis, it is worth noting that for Ag43^C1-PT similar force curves and rupture forces of 100–200 pN were previously recorded³³.

Considering that heterologous associations have been reported between very different members of the SAATs family, namely Ag43, AIDA-I and TibA^17,19, the paucity of heterotypic interactions between Ag43 variants evidenced in this work was unexpected. While the recognition and involvement of binding interfaces in both homotypic and heterotypic interactions were suggested in the case of co-interacting SAATs, the restricted number of binding pairs respective to the different Ag43 variants suggest the involvement of different binding interfaces. Structural studies of homotypic interactions in Ag43^{C3-UPI00000E4A66} revealed a Velcro-like handshake mechanism via trans-association of the SL subdomains stabilised by multiple hydrogen bonds and electrostatic interactions²⁵. Ag43^C1-PT and Ag43^C4-PT share the SL1 subdomain and we could observe some heterotypic interactions between cells expressing those Ag43 homologues. Nonetheless, the level and force of this heterotypic association are significantly weaker than the corresponding homotypic interactions. Differences at amino acid level between Ag43^C1-PT and Ag43^C4-PT proteins probably result in a reduced number of interactions stabilising the hetero-dimers compared to the homodimers. Additionally, the EJ-BL subdomains could also play an indirect role in the Ag43-Ag43 interaction. Indeed, previous work showed that modification of the EJ subdomain altered the angle between the SL and BL subdomains and the overall L-shaped structure, completely abolishing the ability of cells expressing Ag43 to form aggregates²⁵. The absence of aggregates between cells expressing Ag43^C2-PT and Ag43^C3-PT, despite both having a SL2, suggests this subdomain is less tolerant towards heterotypic associations and may be due to the binding interface regions being more selective in forming interactions. Similarly, no heterotypic interactions could be detected for combinations of Ag43 exhibiting different SL subdomain, namely SL1/SL2, except in the case of cells expressing Ag43^C3-PT and Ag43^C4-PT. Considering the only two observed heterotypic interactions involved Ag43^C4-PT, it is tempting to hypothesise that the amino acid composition of SL1 and the EJ2-BL2 architecture in the passenger are more permissive to heterotypic interactions. Diverse EJ-BL subdomains may induce different bending angles to the L-shape Ag43 structure and modulate the ability to form heterologous associations. However, the molecular modelling approach used here to model the structure of different Ag43 homologues was not able to detect any differences as the angle between SL and BL biased the bending towards the Ag43^{C3-UPI00000E49D2} template. Solving the experimental structures of Ag43 belonging to classes 1, 2 and 4, as well as other SAATs is critical to further inform the role of the different passenger subdomains in inducing autoaggregation and potentially predict heterologous associations among different SAATs.

Autoaggregation is a widespread phenomenon in bacteria but its ecophysiological importance is not fully elucidated. Several direct and side effects of autoaggregation have been reported in the literature, e.g. mechanism of adaptation against different stresses such as host defences as well as promoting cell invasion and colonisation¹⁸. The differential autoaggregation as well as the paucity of heterotypic interactions suggests that the Ag43 variants could play a role in structuring the E. coli population in an ecosystem by rallying or excluding some bacterial strains and thus stratifying the bacterial community. Considering that up to three agn43 alleles can be present in one bacterial genome²⁴, further genomic investigations would be required to characterise the prevalence, distribution and diversity of Ag43, such as the combinations of classes and domain shuffling, among E. coli and within the Bacteria kingdom, as well the correlations with some features of the bacterial strains, such as the origin, the level of virulence or serotype. The reason why E. coli can carry different agn43 alleles in one genome and express so many different variants of Ag43 is still unclear²⁴ but their implication in social behaviour especially through selective altruism following a greenbeard effect is an attractive concept^34,35,36. Considering agn43 can identify the presence of copies of themselves in other cells through Ag43-Ag43 interactions, this might induce some kind of nepotism behaviour toward those individual cells. The diversity of Ag43 variants and multiplicity of alleles could then favour differential and beneficial interactions between individual cells in a bacterial population.

These different aspects would undoubtedly require further investigations to understand how it could contribute to colonisation process and specific adaptation of E. coli within an ecosystem, both considering commensal and pathogenic E. coli strains but also the complexity of bacterial communities, including the microbiota and the rest of the biotope as well as the biocenosis.

Methods

Bacterial strains and culture conditions

Experiments were carried out in an E. coli K12 MG1655 ∆fim ∆agn43 deficient in type 1 fimbriae and Ag43 expression making the cells unable to autoaggregate, hereafter referred as E. coli SRUD (in honour to the inventors Sebastian Reidl and Ulrich Dobrindt)³⁰. The most relevant bacterial strains used in this study are listed in Table 1; the exhaustive list of strains and plasmids is presented in details in Supplementary Table S1. Bacteria were cultured in LB (lysogeny broth) medium^37,38. Briefly, strains were plated on LB agar medium and incubated overnight at 37 °C and a preculture was set up from one colony in LB broth at 37 °C in an orbital shaker (150 rpm). When required, LB medium was supplemented with antibiotics, i.e. ampicillin (100 μg.ml⁻¹), chloramphenicol (25 μg.ml⁻¹) or kanamycin (40 μg.ml⁻¹). Arabinose at a final concentration of 0.2% (w/v) was added in the medium to induce gene expression from pBAD/Myc-HisA vector (ThermoFischer Scientific).

Cloning and expression of genes encoding Ag43 variants

CDS encoding the different Ag43 variants (namely UniParc: UPI003B25EE, UPI000224229A, UPI0001F8D949, UPI00000D07E5, UPI0001F68B55, UPI00005F0EEA, UPI00000E4A66, UPI00000E49D2, UPI0001EB6ED2, UPI000037640F, UPI0001888EFC, UPI000000CD3C, UPI0001E6D993, UPI00092D946C, UPI00092DE8CA) were synthesised de novo by GeneArt (ThermoFischer Scientific) and cloned into the pBAD/Myc-HisA expression vector at the XhoI and HindIII restriction sites. Plasmid constructs were transformed into E. coli SRUD by electroporation using a Gene Pulser II and Pulse Controler Plus (Bio-Rad) (12.5 kV/cm, 200Ω, 25 μF) prior to selection on LB agar supplemented with ampicillin.

To obtain fluorescent bacterial strains, genes encoding for mRuby2, mTagBFP2 and mEmerald fluorescent proteins (GeneBank: AFR60232.1, AIQ82697.1 and AIA24532.1, respectively) were synthesised by GeneArt including a constitutive promoter (iGEM: BBa-J23119) and ribosome binding site (RBS) from pBAD/Myc-HisA, prior to cloning into the pACYC184 AvaI and Bsu36I/Tth111I restriction sites to generate pVaAgRed, pVaAgBlue and pVaAgGreen respectively, whose p15a replication origin belongs to a different incompatibility group from the pBR322 replication origin in pBAD-vector series. Plasmid constructs were transformed by electroporation in E. coli SRUD with or without a pVaAg-vector series and selection was carried on LB agar supplemented with chloramphenicol and when required with the additional supplementation of ampicillin.

Immunodetection and subcellular localisation of Ag43 variants

Bacterial cells were grown with arabinose till late exponential phase as described above. Cells were harvested by centrifugation (7 000 g, 5 min, 4 °C), washed and resuspended in PBS (0.01 M, 1 mM PMSF, pH8). Cells were disrupted with glass microbeads using FastPrep (20 s twice at 6 m.s⁻¹). Part of lysed whole cells was reserved and the other part subjected to cell fractionation using a modified sarkosyl (sodium lauroyl sarcosinate) method^39,40. Briefly, unbroken cells and cellular debris were removed by centrifugation (12 000 g, 10 min, 4 °C) before the supernatant was ultracentrifuged (100 000 g, 40 min, 4 °C). The supernatant was retained as the soluble fraction (cytoplasm and periplasm), and the pellet was resuspended in 2% sarkosyl (20 min, room temperature) followed by another ultracentrifugation. The pellet corresponded to the sarcosyl-insoluble fraction gathering the OM. Cold acetone was added to the supernatant (overnight at −20 °C), centrifuged and the pellet corresponding to the sarcosyl-soluble fraction gathering the IM. Protein concentrations were determined using the Bradford method⁴¹.

Proteins from the different fractions were separated by SDS-PAGE. Cell fractions were diluted in Laemmli buffer and heated up at 95 °C for 5 min. From prior testing with increasing amounts of whole cell fractions, 1 µg of total proteins was determined as a non-saturating concentration and equal amounts of each fraction based on the starting material were loaded per lane using Mini-PROTEAN Precast TGX Gels 4–15% (Bio-Rad). Precision Plus Protein Dual Color Standards (Bio-Rad) was used to enable molecular mass estimation as well as transfer control. For Western blotting, proteins were dry transferred from SDS-PAGE on iBlot Transfer Stacks using iBlot 2 Dry Blotting System (Invitrogen). After saturating unoccupied protein-binding sites with blocking agents, membranes were probed with specific anti-Ag43 primary antibodies (dilution 1/5000 in PBS with 0.1% tween and 0.1% milk, 1 h, room temperature). Rabbit Ag43 antibodies were obtained as previously described^42,43 and are crossed reactive between Ag43 variants⁴⁴. Goat anti-Rabbit IgG antibodies conjugated with horseradish peroxidase was used as secondary antibodies (dilution of 1/10 000, 1 h, room temperature) and TMB (3,3′,5,5′-tetramethylbenzidine) was used to detect the enzymatic activity. Enzymatic reaction was stopped by rinsing the membranes with milliQ water. Western blots were imaged under white light conversion screen using E-BOX (VILBER).

Phylogenetic analysis of protein sequences

The Ag43 protein sequences were retrieved from the literature^24,28 and the corresponding passengers recovered by removing the SP, AC1 and translocator regions. For identification of the SPs in Ag43 proteins, the sequences submitted to SignalP v4.1⁴⁵ were deleted from the ESPR (InterPro: IPR024973), which was first identified using InterProScan v5.22⁴⁶. The AC1 domains were identified based on a structural alignment as previously described³. The translocators were identified using the twin HMM (Hidden Markov Model) strategy for AT detection⁴⁷. Finally, truncated and redundant Ag43 passenger sequences were removed, resulting in a dataset of 106 distinct Ag43 passengers (Supplementary Table S1). Then, the amino acid sequences were aligned with T-Coffee in the Expresso mode for automatic incorporation of structural information^48,49 using the solved structure of Ag43 (PDB: 4KH3) as template. The relatedness among protein sequences was studied using a Neighbor-Net phylogenetic network approach with SplitsTree v4.14.8 using the Hamming uncorrected-P distance⁵⁰. The most robust splits of the network were identified by bootstrap using a 95% threshold over 1,000 pseudo-replicates. A similar phylogenetic analysis was repeated for the different passenger subdomains. Pairwise sequence similarities and identities were obtained using Needleman-Wunsch alignment of two sequences with BLAST^51,52 (Supplementary Table S1).

Identification of conserved motifs and structural modelling

Conserved motifs were searched using InterProScan v5.22⁴⁶ and interrogating InterPro (IPR) v61.0 database⁵³. Three-dimensional structures were modelled with Phyre v2.0^54,55 as well as I-TASSER v5.1^56,57. The types of β-solenoid motif present in the β-helices were predicted using PfScan²⁹. In β-helices, rungs were defined as adjacent parallel β-strands. For selected variants as presented in Fig. 2, Ag43 homolog models were checked and modified in the program COOT⁵⁸ using its Ramachandran, geometry and rotamer analysis validation software. The quality of the models were then assessed using ProCheck⁵⁹ with its Ramachandran errors reported as percentages of generously allowed and disallowed amino acids (Supplementary Table S3). Protein structures were visualised using PyMOL v1.7.4^60,61.

Autoaggregation assay

Autoaggregation assays were performed as previously described⁶². Briefly, overnight precultures were diluted 1:100 and grown in LB at 37 °C under orbital shaking. Arabinose was added in early exponential phase (OD_{600 nm} = 0.25) to induce the Ag43 expression from pBAD-vector series (Supplementary Table S2). Cell suspensions were placed in conical tubes at the same OD_{600 nm}, incubated statically and sampled every hour from the top of the tube to measure the OD_{600 nm}. Observations in phase-contrast microscopy were made at the end of the experiments from sediment at the bottom of the tube to confirm the bacterial cell autoaggregation.

Image acquisition and analysis of phase contrast and epifluorescence microscopic observations

Microscopic observations of bacterial cells were performed in phase contrast and phase contrast epifluorescence reflected light. Image acquisition was performed using an Axioplan 2 microscope coupled to an AxioCam MRm camera driven by the AxioVision 4.8 software (Carl Zeiss). For phase contrast microscopy, a 40 × /0.75 objective (Plan-NEOFLUAR) was used and the fluorescence light source was a mercury lamp (HBO100W/2, AttoArc, Carl Zeiss). For fluorescence image acquisition a SYTO9-AO-FITC cube was fitted with a 450–490 nm exciter, 515–565 nm emitter, and 510 nm dichroic mirror as well as a Propidium iodide-Rhodamine (Con A) cube containing 546/12 nm exciter, 590 nm emitter, and 580 nm dichroic mirror (Carl Zeiss).

Quantification of bacterial cell interactions by flow cytometry

To investigate interactions of bacterial cells expressing Ag43 variants, E. coli SRUD-Green (carrying pVaAgGreen) or E. coli SRUD-Blue (carrying pVagAgBlue) were transformed with pBAD/Myc-HisA vectors encoding the different Ag43 variants or empty vector (negative control with non-aggregative E. coli SRUD bacterial cells) (Supplementary Table S2). Overnight E. coli SRUD-Green and E. coli SRUD-Blue precultures were diluted 1:100 and grown as described above in LB until early exponential phase. At this point, samples from these two different bacterial cultures were adjusted at an OD_{600 nm} of 0.25, mixed together at a 50/50 (v/v) ratio with arabinose and grown at 37 °C under shaking until an OD_{600 nm} of 0.5. Conical tubes were then incubated statically for 6 h and bacterial cells fixed with paraformaldehyde (2% final concentration) for 15 min. Cells were collected in the bottom of each tube and diluted in PBS (phosphate-buffered saline).

Samples were filtered on a 40-μm-nylon cell strainer prior to flow cytometry analysis (CytoFLEX, Beckman Coulter). FSC (forward scatter), SSC (side scatter), FITC (fluorescein isothiocyanate; 488 nm exciter, 525/540 nm emitter) and PB450 (Pacific Blue 450 nm; 405 nm exciter, 450/445 nm emitter) axes were set to log display. Areas for FSC and SSC (FSC-A and SSC-A, respectively) were used to discriminate aggregates from single individual cells, whereas heights for FITC and PB450 (FITC-H and PB450-H, respectively) were used to quantify co-interactions within aggregates (Supplementary Fig. S3). At least 3000 events/sec were recorded for each sample and data were analysed with CytExpert software v2.2 (Beckman Coulter). Results were expressed as percentages of mixed aggregates over the total aggregates.

Atomic force microscopy analyses

For cell probe preparation, bacterial cells were handled as for the autoaggregation assay prior to dilution in PBS buffer to an OD_{600 mn} of 0.2. One mL of bacterial suspension was directly deposited onto glass slide pre-coated with polyethylenimin (PEI) as previously described⁶³. Briefly, cleaned glass slides are immersed in 0.2% PEI solution for 1 h, then extensively rinsed with Milli-Q water and finally dried under nitrogen) for 30 min and then the surface was rinsed twice with PBS buffer to get rid of un-adhered free bacteria and remaining residues from growth medium.

A Nanowizard III AFM (JPK-Bruker) was used to attach to atipless silicon nitride cantilevers (NP-010, Bruker) 6 µm borosilicate microspheres using UV curing resin (Norland optical adhesive 63 Glue, Norland Edmund Optics). Sphere attached cantilevers were cured under UV light (λ = 400 nm) for 3 min. The cantilever was then immersed for 1 h in a 0.2% PEI solution and washed. Functionalised cantilevers were calibrated before analysis by the thermal noise method⁶⁴ and spring constants were found to be 0.048 ± 0.011 N/m. Colloidal probe⁶⁵ was brought into contact with an isolated cell for 3 min, and the obtained cell probe was then moved to another cell to perform single cell-force spectroscopy measurements. Suitable parameters for cell attachment are a contact force of 1 nN, a contact time of 2 s and an approach and retract rate of 1 µms⁻¹.

Single cell-force spectroscopy experiments were carried out with a Nanowizard III atomic force microscope (JPK-Bruker). Measurements were performed at room temperature in PBS buffer at pH 7.2. Force-distance curves were recorded on different cells, using a maximum applied force of 1 nN and a constant approach and retraction speed of 1 µm.s⁻¹. The time of contact was set at 2 s to enhance specific interactions as described previously^66,67. The type of interactions, i.e. specific or non-specific, could be deduced from the patterns of the force-distance curves^68,69,70.

The obtained force versus distance curves were analysed with the JPK Data Processing software on the basis of the worm-like chain (WLC) model, which is suitable to describe the extension of protein. The extension x of the molecule versus the pulling force (or adhesion force) is given by the following equation:

$$F(x)=\,\frac{{k}_{B}T}{{l}_{p}}\,[\frac{1}{4}\,{(1-\frac{x}{{L}_{c}})}^{-2}+\,\frac{x}{{L}_{c}}-\,\frac{1}{4}]$$

where k_B is the Boltzmann constant, T the absolute temperature, l_p is the persistent length and L_c is the contour length. The persistent length gives information about the degree of structural rigidity of the polymer chain and is defined as the longest segment below which the chain can be considered as a rigid rod, the extension of the polymer until the point at which the force necessary to extend further and in our case to break the interaction with the functionalised AFM tip corresponds to the contour length.

Adhesion peaks were generated using Origin 8 software. Statistical differences of AFM data were assessed by analysis of the variance (one-way ANOVA test) using Origin 8. To check the specificity of the measured adhesion forces and avoid artifacts that can be associated with cell probe preparation, control experiments were performed. Use of PEI-coated probes without bacterial cell interacting on PEI-coated glass slide led to a dramatic decrease of the adhesion frequency with only 3% of the force curves presenting an adhesion peak.

Statistical analysis

Statistical analyses were performed using GraphPad Prism v6.01. At least five biological replicates were carried out for all statistical analyses. Error bars on the figures represent the 95% confidence interval for the mean of the independent experiments. Data were statistically analysed following ANOVA one way or two way test, followed by the post-hoc Tukey test, with differences considered significant (p < 0.05, * or lowercase letter), very significant (p < 0.01, ** or bold lowercase letter), highly significant (p < 0.001, *** or capital letter) or very highly significant (p < 0.0001, **** or bold capital letter).

Data Availabiliy

All data generated or analysed during this study are included in this published article and its Supplementary Material files. Other data are available from the corresponding author upon reasonable request.

References

Desvaux, M., Parham, N. J., Scott-Tucker, A. & Henderson, I. R. The general secretory pathway: a general misnomer? Trends in microbiology 12, 306–309, https://doi.org/10.1016/j.tim.2004.05.002 (2004).
Article CAS PubMed Google Scholar
Desvaux, M., Hébraud, M., Talon, R. & Henderson, I. R. Secretion and subcellular localizations of bacterial proteins: a semantic awareness issue. Trends in microbiology 17, 139–145, https://doi.org/10.1016/j.tim.2009.01.004 (2009).
Article CAS PubMed Google Scholar
Rojas-Lopez, M. et al. Identification of the Autochaperone Domain in the Type Va Secretion System (T5aSS): Prevalent Feature of Autotransporters with a β-Helical Passenger. Frontiers in microbiology 8, 2607, https://doi.org/10.3389/fmicb.2017.02607 (2017).
Article PubMed Google Scholar
Yen, M. R. et al. Protein-translocating outer membrane porins of Gram-negative bacteria. Biochimica et biophysica acta 1562, 6–31, https://doi.org/10.1016/S0005-2736(02)00359-0 (2002).
Article CAS PubMed Google Scholar
Abby, S. S. et al. Identification of protein secretion systems in bacterial genomes. Scientific reports 6, 23080, https://doi.org/10.1038/srep23080 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Henderson, I. R. & Lam, A. C. Polymorphic proteins of Chlamydia spp.–autotransporters beyond the Proteobacteria. Trends in microbiology 9, 573–578, S0966-842X(01)02234-X (2001).
Desvaux, M., Khan, A., Beatson, S. A., Scott-Tucker, A. & Henderson, I. R. Protein secretion systems in Fusobacterium nucleatum: genomic identification of Type 4 piliation and complete Type V pathways brings new insight into mechanisms of pathogenesis. Biochim Biophys Acta 1713, 92–112, https://doi.org/10.1016/j.bbamem.2005.05.002 (2005).
Article CAS PubMed Google Scholar
Leyton, D. L., Rossiter, A. E. & Henderson, I. R. From self sufficiency to dependence: mechanisms and factors important for autotransporter biogenesis. Nat Rev Microbiol 10, 213–225, https://doi.org/10.1038/nrmicro2733 (2012).
Article CAS PubMed Google Scholar
Selkrig, J. et al. Discovery of an archetypal protein transport system in bacterial outer membranes. Nat Struct Mol Biol 19(506–510), S501, https://doi.org/10.1038/nsmb.2261 (2012).
Article CAS Google Scholar
Henderson, I. R., Navarro-Garcia, F., Desvaux, M., Fernandez, R. C. & Ala’Aldeen, D. Type V protein secretion pathway: the autotransporter story. Microbiology and molecular biology reviews: MMBR 68, 692–744, https://doi.org/10.1128/MMBR.68.4.692-744.2004 (2004).
Article CAS PubMed PubMed Central Google Scholar
Leo, J. C., Grin, I. & Linke, D. Type V secretion: mechanism(s) of autotransport through the bacterial outer membrane. Philosophical transactions of the Royal Society of London. Series B, Biological sciences 367, 1088–1101, https://doi.org/10.1098/rstb.2011.0208 (2012).
Article CAS PubMed PubMed Central Google Scholar
Desvaux, M., Parham, N. J. & Henderson, I. R. Le système de sécrétion de type V chez les bactéries Gram-négatives (Type V secretion system in Gram-negative bacteria). Biofutur 237, 34–37 (2003).
Google Scholar
Desvaux, M., Parham, N. J. & Henderson, I. R. Type V protein secretion: simplicity gone awry? Curr Issues Mol Biol 6, 111–124 (2004).
CAS PubMed Google Scholar
Tsai, J. C. et al. The bacterial intimins and invasins: a large and novel family of secreted proteins. PloS one 5, e14403, https://doi.org/10.1371/journal.pone.0014403 (2010).
Article ADS CAS PubMed PubMed Central Google Scholar
Desvaux, M., Parham, N. J. & Henderson, I. R. The autotransporter secretion system. Research in microbiology 155, 53–60, https://doi.org/10.1016/j.resmic.2003.10.002 (2004).
Article CAS PubMed Google Scholar
Drobnak, I. et al. Of linkers and autochaperones: an unambiguous nomenclature to identify common and uncommon themes for autotransporter secretion. Molecular microbiology 95, 1–16, https://doi.org/10.1111/mmi.12838 (2015).
Article CAS PubMed Google Scholar
Klemm, P., Vejborg, R. M. & Sherlock, O. Self-associating autotransporters, SAATs: functional and structural similarities. Int J Med Microbiol 296, 187–195, https://doi.org/10.1016/j.ijmm.2005.10.002 (2006).
Article CAS PubMed Google Scholar
Trunk, T., Khalil, H. S. & Leo, J. C. Bacterial autoaggregation. Aims Microbiol 4, 140–164, https://doi.org/10.3934/microbiol.2018.1.140 (2018).
Article PubMed PubMed Central Google Scholar
Sherlock, O., Schembri, M. A., Reisner, A. & Klemm, P. Novel roles for the AIDA adhesin from diarrheagenic Escherichia coli: cell aggregation and biofilm formation. Journal of bacteriology 186, 8058–8065, https://doi.org/10.1128/JB.186.23.8058-8065.2004 (2004).
Article CAS PubMed PubMed Central Google Scholar
Kjaergaard, K., Schembri, M. A., Ramos, C., Molin, S. & Klemm, P. Antigen 43 facilitates formation of multispecies biofilms. Environ Microbiol 2, 695–702, https://doi.org/10.1046/j.1462-2920.2000.00152.x (2000).
Article CAS PubMed Google Scholar
Desvaux, M. et al. A conserved extended signal peptide region directs posttranslational protein translocation via a novel mechanism. Microbiology 153, 59–70, https://doi.org/10.1099/mic.0.29091-0 (2007).
Article CAS PubMed Google Scholar
Jong, W. S. et al. YidC is involved in the biogenesis of the secreted autotransporter hemoglobin protease. The Journal of biological chemistry 285, 39682–39690, https://doi.org/10.1074/jbc.M110.167650 (2010).
Article CAS PubMed PubMed Central Google Scholar
Oliver, D. C., Huang, G., Nodel, E., Pleasance, S. & Fernandez, R. C. A conserved region within the Bordetella pertussis autotransporter BrkA is necessary for folding of its passenger domain. Molecular microbiology 47, 1367–1383, https://doi.org/10.1046/j.1365-2958.2003.03377.x (2003).
Article CAS PubMed Google Scholar
van der Woude, M. W. & Henderson, I. R. Regulation and function of Ag43 (flu). Annual review of microbiology 62, 153–169, https://doi.org/10.1146/annurev.micro.62.081307.162938 (2008).
Article CAS PubMed Google Scholar
Heras, B. et al. The antigen 43 structure reveals a molecular Velcro-like mechanism of autotransporter-mediated bacterial clumping. Proceedings of the National Academy of Sciences of the United States of America 111, 457–462, https://doi.org/10.1073/pnas.1311592111 (2014).
Article ADS CAS PubMed Google Scholar
Torres, A. G. et al. Characterization of Cah, a calcium-binding and heat-extractable autotransporter protein of enterohaemorrhagic Escherichia coli. Molecular microbiology 45, 951–966, https://doi.org/10.1046/j.1365-2958.2002.03094.x (2002).
Article CAS PubMed Google Scholar
Ulett, G. C. et al. Functional analysis of antigen 43 in uropathogenic Escherichia coli reveals a role in long-term persistence in the urinary tract. Infection and immunity 75, 3233–3244, https://doi.org/10.1128/IAI.01952-06 (2007).
Article CAS PubMed PubMed Central Google Scholar
Vo, J. L. et al. Autotransporter Adhesins in Escherichia coli Pathogenesis. Proteomics 17, https://doi.org/10.1002/pmic.201600431 (2017).
Article Google Scholar
Kajava, A. V. & Steven, A. C. The turn of the screw: variations of the abundant beta-solenoid motif in passenger domains of Type V secretory proteins. J Struct Biol 155, 306–315, https://doi.org/10.1016/j.jsb.2006.01.015 (2006).
Article CAS PubMed Google Scholar
Reidl, S., Lehmann, A., Schiller, R., Salam Khan, A. & Dobrindt, U. Impact of O-glycosylation on the molecular and cellular adhesion properties of the Escherichia coli autotransporter protein Ag43. Int J Med Microbiol 299, 389–401, https://doi.org/10.1016/j.ijmm.2009.01.001 (2009).
Article CAS PubMed Google Scholar
Klemm, P., Hjerrild, L., Gjermansen, M. & Schembri, M. A. Structure-function analysis of the self-recognizing Antigen 43 autotransporter protein from Escherichia coli. Molecular microbiology 51, 283–296, https://doi.org/10.1046/j.1365-2958.2003.03833.x (2004).
Article CAS PubMed Google Scholar
Crossman, L. C. et al. A commensal gone bad: complete genome sequence of the prototypical enterotoxigenic Escherichia coli strain H10407. Journal of bacteriology 192, 5822–5831, https://doi.org/10.1128/JB.00710-10 (2010).
Article CAS PubMed PubMed Central Google Scholar
Jacquot, A. et al. The dynamics and pH-dependence of Ag43 adhesins’ self-association probed by atomic force spectroscopy. Nanoscale 6, 12665–12681, https://doi.org/10.1039/c4nr03312d (2014).
Article ADS CAS PubMed Google Scholar
Gardner, A. & West, S. A. Greenbeards. Evolution 64, 25–38, https://doi.org/10.1111/j.1558-5646.2009.00842.x (2010).
Article PubMed Google Scholar
West, S. A. & Gardner, A. Altruism, spite, and greenbeards. Science 327, 1341–1344, https://doi.org/10.1126/science.1178332 (2010).
Article ADS CAS PubMed Google Scholar
West, S. A., Griffin, A. S. & Gardner, A. Social semantics: altruism, cooperation, mutualism, strong reciprocity and group selection. J Evol Biol 20, 415–432, https://doi.org/10.1111/j.1420-9101.2006.01258.x (2007).
Article CAS PubMed Google Scholar
Bertani, G. Studies on lysogenesis. I. The mode of phage liberation by lysogenic Escherichia coli. Journal of bacteriology 62, 293–300 (1951).
CAS PubMed PubMed Central Google Scholar
Bertani, G. Lysogeny at mid-twentieth century: P1, P2, and other experimental systems. Journal of bacteriology 186, 595–600, https://doi.org/10.1128/jb.186.3.595-600.2004 (2004).
Article CAS PubMed PubMed Central Google Scholar
Filip, C., Fletcher, G., Wulff, J. L. & Earhart, C. F. Solubilization of the cytoplasmic membrane of Escherichia coli by the ionic detergent sodium-lauryl sarcosinate. Journal of bacteriology 115, 717–722 (1973).
CAS PubMed PubMed Central Google Scholar
Huang, C. Z. et al. Systematic identification of the subproteome of Escherichia coli cell envelope reveals the interaction network of membrane proteins and membrane-associated peripheral proteins. J Proteome Res 5, 3268–3276, https://doi.org/10.1021/pr060257h (2006).
Article CAS PubMed Google Scholar
Bradford, M. M. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72, 248–254, https://doi.org/10.1016/0003-2697(76)90527-3 (1976).
Article CAS PubMed Google Scholar
Owen, P., Caffrey, P. & Josefsson, L. G. Identification and partial characterization of a novel bipartite protein antigen associated with the outer membrane of Escherichia coli. Journal of bacteriology 169, 3770–3777, https://doi.org/10.1128/jb.169.8.3770-3777.1987 (1987).
Article CAS PubMed PubMed Central Google Scholar
Henderson, I. R., Meehan, M. & Owen, P. Antigen 43, a phase-variable bipartite outer membrane protein, determines colony morphology and autoaggregation in Escherichia coli K-12. FEMS microbiology letters 149, 115–120, S0378-1097(97)00065-7 (1997).
Roche, A., McFadden, J. & Owen, P. Antigen 43, the major phase-variable protein of the Escherichia coli outer membrane, can exist as a family of proteins encoded by multiple alleles. Microbiology 147, 161–169, https://doi.org/10.1099/00221287-147-1-161 (2001).
Article CAS PubMed Google Scholar
Nielsen, H. Predicting Secretory Proteins with SignalP. Methods Mol Biol 1611, 59–73, https://doi.org/10.1007/978-1-4939-7015-5_6 (2017).
Article CAS PubMed Google Scholar
Jones, P. et al. InterProScan 5: genome-scale protein function classification. Bioinformatics 30, 1236–1240, https://doi.org/10.1093/bioinformatics/btu031 (2014).
Article CAS PubMed PubMed Central Google Scholar
Celik, N. et al. A bioinformatic strategy for the detection, classification and analysis of bacterial autotransporters. PloS one 7, e43245, https://doi.org/10.1371/journal.pone.0043245 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
Armougom, F. et al. Expresso: automatic incorporation of structural information in multiple sequence alignments using 3D-Coffee. Nucleic acids research 34, W604–608, https://doi.org/10.1093/nar/gkl092 (2006).
Article CAS PubMed PubMed Central Google Scholar
Di Tommaso, P. et al. T-Coffee: a web server for the multiple sequence alignment of protein and RNA sequences using structural information and homology extension. Nucleic acids research 39, W13–17, https://doi.org/10.1093/nar/gkr245 (2011).
Article CAS PubMed PubMed Central Google Scholar
Huson, D. H. & Bryant, D. Application of phylogenetic networks in evolutionary studies. Molecular biology and evolution 23, 254–267, https://doi.org/10.1093/molbev/msj030 (2006).
Article CAS PubMed Google Scholar
Needleman, S. B. & Wunsch, C. D. A general method applicable to the search for similarities in the amino acid sequence of two proteins. Journal of molecular biology 48, 443–453, https://doi.org/10.1016/0022-2836(70)90057-4 (1970).
Article CAS PubMed Google Scholar
Tatusova, T. A. & Madden, T. L. Blast 2 sequences - a new tool for comparing protein and nucleotide sequences. FEMS microbiology letters 174, 247–250, https://doi.org/10.1111/j.1574-6968.1999.tb13575.x (1999).
Article CAS PubMed Google Scholar
Finn, R. D. et al. InterPro in 2017-beyond protein family and domain annotations. Nucleic acids research 45, D190–D199, https://doi.org/10.1093/nar/gkw1107 (2017).
Article CAS PubMed Google Scholar
Kelley, L. A., Mezulis, S., Yates, C. M., Wass, M. N. & Sternberg, M. J. The Phyre2 web portal for protein modeling, prediction and analysis. Nat Protoc 10, 845–858, https://doi.org/10.1038/nprot.2015.053 (2015).
Article CAS PubMed PubMed Central Google Scholar
Kelley, L. A. & Sternberg, M. J. Protein structure prediction on the Web: a case study using the Phyre server. Nat Protoc 4, 363–371, https://doi.org/10.1038/nprot.2009.2 (2009).
Article CAS PubMed Google Scholar
Zhang, Y. I-TASSER server for protein 3D structure prediction. BMC Bioinformatics 9, 40, https://doi.org/10.1186/1471-2105-9-40 (2008).
Article CAS PubMed PubMed Central Google Scholar
Yang, J. et al. The I-TASSER Suite: protein structure and function prediction. Nat Methods 12, 7–8, https://doi.org/10.1038/nmeth.3213 (2015).
Article CAS PubMed PubMed Central Google Scholar
Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr D Biol Crystallogr 60, 2126–2132, https://doi.org/10.1107/S0907444904019158 (2004).
Article CAS PubMed Google Scholar
Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Thornton, J. M. PROCHECK - a program to check the stereochemical quality of protein structures. J. Appl. Cryst. 26, 283–291, https://doi.org/10.1107/S0021889892009944 (1993).
Article CAS Google Scholar
Yuan, S., Chan, H. C. S., Filipek, S. & Vogel, H. PyMOL and Inkscape Bridge the Data and the Data Visualization. Structure 24, 2041–2042, https://doi.org/10.1016/j.str.2016.11.012 (2016).
Article CAS PubMed Google Scholar
Rigsby, R. E. & Parker, A. B. Using the PyMOL application to reinforce visual understanding of protein structure. Biochem Mol Biol Educ 44, 433–437, https://doi.org/10.1002/bmb.20966 (2016).
Article CAS PubMed Google Scholar
Chagnot, C. et al. In vitro colonization of the muscle extracellular matrix components by Escherichia coli O157:H7: the influence of growth medium, temperature and pH on initial adhesion and induction of biofilm formation by collagens I and III. PloS one 8, e59386, https://doi.org/10.1371/journal.pone.0059386 (2013).
Article ADS CAS PubMed PubMed Central Google Scholar
Vadillo-Rodriguez, V. et al. Comparison of atomic force microscopy interaction forces between bacteria and silicon nitride substrata for three commonly used immobilization methods. Applied and environmental microbiology 70, 5441–5446, https://doi.org/10.1128/AEM.70.9.5441-5446.2004 (2004).
Article CAS PubMed PubMed Central Google Scholar
Hutter, J. L. & Bechhoefer, J. Calibration of Atomic-Force Microscope Tips. Rev Sci Instrum 64, 3342–3342, https://doi.org/10.1063/1.1144449 (1993).
Article ADS CAS Google Scholar
Ducker, W. A., Senden, T. J. & Pashley, R. M. Direct Measurement of Colloidal Forces Using an Atomic Force Microscope. Nature 353, 239–241, https://doi.org/10.1038/353239a0 (1991).
Article ADS CAS Google Scholar
Beaussart, A. et al. Quantifying the forces guiding microbial cell adhesion using single-cell force spectroscopy. Nature Protocols 9, 1049–1055, https://doi.org/10.1038/nprot.2014.066 (2014).
Article CAS PubMed Google Scholar
Zeng, G. H., Muller, T. & Meyer, R. L. Single-Cell Force Spectroscopy of Bacteria Enabled by Naturally Derived Proteins. Langmuir: the ACS journal of surfaces and colloids 30, 4019–4025, https://doi.org/10.1021/la404673q (2014).
Article CAS Google Scholar
Gaboriaud, F. & Dufrene, Y. F. Atomic force microscopy of microbial cells: application to nanomechanical properties, surface forces and molecular recognition forces. Colloids Surf B Biointerfaces 54, 10–19, https://doi.org/10.1016/j.colsurfb.2006.09.014 (2007).
Article CAS PubMed Google Scholar
Ahimou, F., Denis, F. A., Touhami, A. & Dufrêne, Y. F. Probing microbial cell surface charges by atomic force microscopy. Langmuir, https://doi.org/10.1021/la026273k (2002).
Article CAS Google Scholar
Alsteens, D., Dague, E., Rouxhet, P. G., Baulard, A. R. & Dufrene, Y. F. Direct measurement of hydrophobic forces on cell surfaces using AFM. Langmuir 23, 11977–11979, https://doi.org/10.1021/la702765c (2007).
Article CAS PubMed Google Scholar

Download references

Acknowledgements

This work was supported in part by INRA (Institut National de la Recherche Agronomique) and by the CoMBa (“Colonisation de la Matrice extracellulaire Bacterienne”) project funded by the “Région Auvergne” FRI (“Fond Régional Innovation”) cluster IRP (“Institut de Recherche Pharmabiotique”). Support from the Australian Research Council (ARC) (DP150102287, DP180102987, FT130100580) is also acknowledged. Valentin AGEORGES is a PhD Research Fellow granted by the “Conseil Régional d’Auvergne – FEDER (Fonds Européen de Développement Régional)”. The authors thank Ulrich Dobrindt for providing with E. coli K12 MG1655 ∆fim ∆agn43. The authors are grateful to the INRA MIGALE bioinformatics platform (http://migale.jouy.inra.fr) for providing computational resources as well as to the INRA MEDiS microscopy platform for access to the epifluorescence microscope.

Author information

Authors and Affiliations

Université Clermont Auvergne, INRA, UMR454 MEDiS, 63000, Clermont-Ferrand, France
Valentin Ageorges, Grégory Jubelin, Nelly Caccia, Philippe Ruiz, Ingrid Chafsey, Sabine Leroy, Frédérique Chaucheyras-Durand & Mickaël Desvaux
Lallemand Animal Nutrition, 31702, Blagnac, Cedex, France
Marion Schiavone & Frédérique Chaucheyras-Durand
LISBP, Université de Toulouse, CNRS UMR5504, INRA UMR792, INSA, 31077, Toulouse, Cedex, France
Marion Schiavone
LAAS-CNRS, Université de Toulouse, 31077, Toulouse, Cedex, France
Etienne Dague
INRA UR346 Epidémiologie Animale, 63122, Saint Genès Champanelle, France
Xavier Bailly
Department of Biochemistry and Genetics, La Trobe Institute for Molecular Science, La Trobe University, Bundoora, Victoria, 3086, Australia
Jason Paxman & Begoña Heras
Institute of Microbiology and Infection, University of Birmingham, B152TT, Birmingham, United Kingdom
Amanda E. Rossiter & Ian R. Henderson
Institute for Molecular Biosciences, University of Queensland, St. Lucia, Queensland, 4067, Australia
Ian R. Henderson

Authors

Valentin Ageorges
View author publications
You can also search for this author in PubMed Google Scholar
Marion Schiavone
View author publications
You can also search for this author in PubMed Google Scholar
Grégory Jubelin
View author publications
You can also search for this author in PubMed Google Scholar
Nelly Caccia
View author publications
You can also search for this author in PubMed Google Scholar
Philippe Ruiz
View author publications
You can also search for this author in PubMed Google Scholar
Ingrid Chafsey
View author publications
You can also search for this author in PubMed Google Scholar
Xavier Bailly
View author publications
You can also search for this author in PubMed Google Scholar
Etienne Dague
View author publications
You can also search for this author in PubMed Google Scholar
Sabine Leroy
View author publications
You can also search for this author in PubMed Google Scholar
Jason Paxman
View author publications
You can also search for this author in PubMed Google Scholar
Begoña Heras
View author publications
You can also search for this author in PubMed Google Scholar
Frédérique Chaucheyras-Durand
View author publications
You can also search for this author in PubMed Google Scholar
Amanda E. Rossiter
View author publications
You can also search for this author in PubMed Google Scholar
Ian R. Henderson
View author publications
You can also search for this author in PubMed Google Scholar
Mickaël Desvaux
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

V.A. and M.D. conceptualise the overarching aims of the research study. V.A., M.S., E.D., G.J., B.H., I.R.H. and M.D. conceived and designed the experiments. V.A., M.S., N.C., P.R., I.C., X.B. and A.E.R. performed the experiments and data acquisition. V.A., M.S., G.J., P.R., I.C., X.B., S.L., J.P., B.H., F.C.D., I.R.H. and M.D. analysed and interpreted the data. M.D. had management as well as coordination responsibility for the execution of the research work. M.D. and B.H. contributed to the acquisition of the financial supports and resources leading to this publication. V.A., M.S., G.J., X.B., E.D., S.L., J.P., B.H., F.C.D., I.R.H. and M.D. wrote the article, including drafting and revising critically the manuscript for important intellectual content.

Corresponding author

Correspondence to Mickaël Desvaux.

Ethics declarations

Competing Interests

F.C.D. and M.S. are employees of the Lallemand SAS Company, which provided support in the form of salaries and partly financed their work. All other authors declare no competing interests.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Table S1

Figure S1

Figure S2

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Ageorges, V., Schiavone, M., Jubelin, G. et al. Differential homotypic and heterotypic interactions of antigen 43 (Ag43) variants in autotransporter-mediated bacterial autoaggregation. Sci Rep 9, 11100 (2019). https://doi.org/10.1038/s41598-019-47608-4

Download citation

Received: 05 March 2019
Accepted: 18 June 2019
Published: 31 July 2019
DOI: https://doi.org/10.1038/s41598-019-47608-4

This article is cited by

Unveiling the impact of cryptic plasmids curing on Escherichia coli Nissle 1917: massive increase in Ag43c expression
- Qi Lin
- Zhuo Jiang
- Zuo-ming Nie
AMB Express (2024)
Distinct types of multicellular aggregates in Pseudomonas aeruginosa liquid cultures
- Gavin Melaugh
- Vincent A. Martinez
- Rosalind J. Allen
npj Biofilms and Microbiomes (2023)
The non-attached biofilm aggregate
- Kasper N. Kragh
- Tim Tolker-Nielsen
- Mads Lichtenberg
Communications Biology (2023)
Variation of Antigen 43 self-association modulates bacterial compacting within aggregates and biofilms
- Julieanne L. Vo
- Gabriela C. Martínez Ortiz
- Begoña Heras
npj Biofilms and Microbiomes (2022)

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.