The flag-2 locus is widespread among the Enterobacterales
The complete and draft genomes of 4,028 bacterial strains encompassing the taxonomic diversity of the order Enterobacterales were screened for the presence of flag–2 loci (Supplementary Table S1). A total of 592 (15% of the analysed taxa) strains were observed to possess an orthologous locus (Figure 1; Supplementary Table S1) and these are distributed across a wide taxonomic breadth of the order. As such, flag-2 loci occur in five of the eight families and 31/76 genera included in this study. Exceptions are observed for the families Morganellaceae (7 genera—313 strains), Pectobacteriaceae (7 genera—244 strains) and Thorselliaceae (Thorsellia –1 strain). The highest prevalence can be observed in the family Budvicaceae (7/9 studied taxa) and Yersiniaceae (225/605 strains), while only 13% (316/2,464 taxa) of the family Enterobacteriaceae contained orthologous loci (Figure 1; Table 1). Differences in prevalence at the genus level could also be observed. Notably, flag-2 loci are universally present in several genera, including Citrobacter Clade D (30/30 strains) and Plesiomonas (8/8 strains), while in the two genera with the highest number of flag-2 loci present, Yersinia (222/394 strains) and Escherichia (124/522 strains), 56% and 24% of the evaluated strains encode flag-2 systems, respectively. In some genera, the presence of flag-2 loci represents a rare trait. For example, only two of 151 analysed Pantoea strains contain flag-2 loci. Diversity in terms of flag-2 locus presence can furthermore be observed at the species level. For example, all 100 of the evaluated Y. pestis strains incorporate a flag-2 locus, while it only occurs in 24/100 Escherichia coli strains.
Molecular architecture of the flag–2 loci
The enterobacterial flag–2 loci range in size from ~3.4 to ~81.8 kilobases (average 38.1 kb) and code for between five and 102 (average 43 proteins) proteins (Supplementary Table S2). The discrepancy in size and number of proteins encoded by the loci can largely be attributed to frequent deletions and insertion of non-core genes within the loci. Substantially larger flag-2 loci are observed in Escherichia albertii B156, Citrobacter (Clade A) sp.nov 1 S1285 and three C. rodentium strains. This can be linked to the insertion of prophage elements within the flag-2 loci, contributing on average 37.7 kb of sequence and 54 proteins.
Comparative analysis showed extensive synteny and sequence conservation among the flag–2 loci (Figure 2). Of the 592 strains with flag-2 loci, 461 (77.87% of strains with flag-2 loci) encode an orthologue complement of 39 conserved proteins. One of these conserved proteins is LafA, the flagellin counterpart of the flag-2 system, which is present in multiple copies in 156/592 (26.35%) strains, with up to five copies (Y. pestis Pestoides B—81.81% average amino acid identity) encoded by the flag-2 locus. Multiple copies of the flagellin gene have also been observed in the flag-1 loci of many enterobacteria and have been suggested to contribute to the phenomenon of phase variation [14,17,18]. As flagellin proteins are potent antigens, the phase variable expression of these proteins may enable these organisms to temporarily avoid immune responses in both plant and animal hosts [14,18]. The remaining 38 single-copy orthologues share an average amino acid identity (AAI) of 61.13% among the 461 enterobacteria with complete flag-2 loci.
In accordance with the study on the flag-2 locus of Escherichia sp. nov. 2 strain 042, the flag-2 loci can be subdivided into three distinct gene clusters—Cluster 1–3 (Figure 2) . Cluster 1, comprised of fourteen genes lfhAB-lfiRQPNM-lafK-lfiEFGHIJ, encodes the proteins involved in regulation and assembly of the basal body components and is analogous to the flhAB-fliRQPNMEFGHIJ genes in the flag-1 locus (Figure 3) [7,15]. The encoded orthologues among the 461 complete complement strains share 67.78% AAI. One Cluster 2 protein restricted to the flag-2 loci, LafK, has been predicted to serve as regulator of flagellum biosynthesis  and shares 67.23% AAI among the 461 strains with complete flag-2 loci. Cluster 2 also typically comprises fourteen genes, lfgNMABCDEFGHIJKL, which are orthologous to flgNMABCDEFGHIJKL in the flag–1 locus and encode flagellar structural proteins (Figure 3) . The flag-2 cluster 2 proteins show slightly greater variability than the cluster 1 genes, sharing 61.44% AAI, with four proteins, LfgN (chaperone), LfgM (Anti σ28factor), LfgA (basal body P-ring protein) and LfgL (hook-associated protein) sharing < 50% AAI.
Cluster 3 comprises of the genes lafWZABCDEFSTU, which code for eleven proteins with substantially lower orthology (50.07% AAI) than those in Cluster 1 and 2. These include proteins involved in filament synthesis (LafABCD—orthologues of FliCDST), σ28factor LafS (orthologue of FliA) and the motor proteins LafT and LafU (orthologues of MotA and MotB in the flag-1 locus) (Figure 3). Also within this cluster are genes coding for the proteins LafW and LafZ, which represent a putative hook-associated protein and transmembrane regulator, respectively , orthologues of which are absent from the flag-1 locus. The latter proteins share lower AAI values of 44.57% and 38.89%, respectively.
Gene and en bloc deletion may have resulted in non-functionality of the flag–2 system in some Enterobacterales taxa
While a substantial fraction of the flag-2 loci contain a complement of 39 conserved genes coding for proteins involved in flagellar biosynthesis and functioning, 22.13% of enterobacterial strains are missing at least one of these genes. For example, 22/67 Y. enterocolitica strains are missing the entire Cluster 1 (lfhAB-lfiRQPNM-lafK-lfiEFGHIJ),, while 3/91 Citrobacter Clade A strains lack both Cluster 2 and Cluster 3. Transposition appears to be a major driver of the observed en bloc gene deletions. As such, twenty-five distinct transposase genes are localised within the Enterobacterales flag-2 loci. These belong to a range of different transposase families, including IS1, IS4, IS5, IS110 and Mu transposases and are integrated in diverse locations within the flag-2 loci. The reading frames of individuals genes could also be observed to be disrupted by transposase integration, with lfgF (20 strains) and lfiG (7 strains), being particularly prone. The en bloc deletions and gene disruptions can be envisaged to have a detrimental effect on functionality of the flag-2 system, as has been observed in Escherichia sp. nov 2 strain 042 .
Previous analyses showed that in many Escherichia and Shigella strains, a deletion has occurred within the reading frames of the lfhA and lafU genes which occur at the 5′ and 3′ ends of the flag–2 locus, respectively, resulting in loss of the remaining locus between the lfhA and lafU pseudogene fragments. The presence of direct repeats at the ends of this deletion suggest that this may have resulted through recombination events . Blast analyses of the lafU and lfhA genes and proteins against the 4,028 Enterobacterales strains showed that this occurs in the genomes of 531 (13.18%) of the strains. The lfhA and lafU pseudogenes are primarily found in those taxa where complete flag–2 loci are present. For example, of the 100 E. coli strains analysed, all 76 strains that lack flag–2 loci contain the truncated gene copies. Similarly, 50 (75.76%) of the 66 Citrobacter Clade A strains lacking flag-2 loci show evidence of its deletion. This suggests that the flag–2 locus is likely to have been a far more prevalent feature among the Enterobacterales (27.88%; 1,123/4,028 analysed strains) prior to en bloc deletion of the locus in a substantial number of strains. While large scale deletions are partially responsible for the difference in size and protein complement observed among the enterobacterial flag-2 loci, it can further be attributed to the integration of a substantial set of non-conserved cargo genes within the loci.
The enterobacterial flag-2loci are hotspots for integration of cargo genes
Alignment of the enterobacterial flag-2 loci and comparative analysis of their encoded protein complements revealed that, although extensive synteny and a substantial set of conserved proteins occur among these loci (Figure 2), there are 349 distinct protein coding genes, which are not conserved among all enterobacterial flag-2 loci and which do not form part of the core set involved in flagellar biosynthesis and functioning. As such, they can be considered as cargo genes within the flag-2 loci. A substantial proportion (121 genes; 34.67% of cargo genes) of these genes code for hypothetical proteins and proteins containing domains of unknown function. However, BlastP searches against the NCBI non-redundant protein database and the Conserved Domain Database , identified proteins with a range of non-flagellar related functions within the flag-2 loci. For example, the flag-2 loci of twenty-one Escherichia strains incorporate genes coding for the restriction endonuclease EcoRII (pfam09019; E-value: 8.36E–98; Average size: 401 aa; AAI: 97.1%) and DNA cytosine methylase Dcm (PRK10458; E-value: 0.0; Average size: 474 aa; AAI: 98.6%). These function in cleaving DNA at a specific sequence and methylation of this sequence to prevent restriction and protect the bacterial cell from integration of bacteriophage and plasmid DNA . Four Pragia fontium strains incorporate genes coding for the pilin protein FimA (PRK15303; E-value: 4.13E–03), periplasmic chaperone FimC (PRK09918; E-value: 3.07E–91) and usher protein FimD (PRK15304; E-value: 0.0).
Cargo genes are found interspersed throughout the flag-2 loci, usually in single or two gene clusters. However, two regions appear to be particularly prone to integration of cargo genes. The first variable region (VR1) occurs between the flag-2 gene clusters 1 and 2 (between lfiJ and lfgN),, while the second (VR2) occurs at the 5′ end of cluster 3 (between lafW and lafZ) (Figure 2). VR1 occurs in the flag–2 loci of 382/592 (64.53%) enterobacterial strains and is particularly prevalent in members of the family Budviciaceae (7/7 strains), Enterobacteriaceae (310/316 strains) and Hafniaceae (28/28 strains), but are more restricted among the flag-2 loci of the Erwiniaceae (4/8 strains) and Yersiniaceae (33/225 strains; 14.67%). The VR1 regions vary in size between 0.7 and 18.9 kb (average size: 5.9 kb) and code for between one and twenty-three (average proteins: 5) proteins (Supplementary Table S3). This region shows evidence of having been derived through horizontal gene transfer, with G+C content deviations of –15.1 to +4.1% (average –2.6%) and –12.5 to +4.2% (average –3.6%) from the genomic and flag–2 G+C content, respectively. A total of 154 distinct proteins are encoded by the VR1 regions of the enterobacterial flag-2 loci. VR2 is more prevalent compared to VR1, occurring in 478/592 (80.74%) of all Enterobacterales. This includes the flag-2 loci of all Budviciaceae and Hafniaceae and 93.78%, 72.47% and 37.5% of the Yersiniaceae, Enterobacteriaceae and Erwiniaceae,respectively. This variable region is typically smaller than VR1, ranging in size from 0.4 to 5.5 kb (average size: 0.6 kb) and coding for between one and six (average: 1 protein) distinct proteins. This region also shows evidence of horizontal acquisition, with an even more pronounced G+C deviation of –13.6 to +8.7% (average: –4.7%) and –15.4 to +5.1% (average: –4.7%) from the genomic and flag-2 locus G+C contents, respectively. VR2 codes for twenty-seven distinct proteins, with the majority of these (21/27; 77.78%) being hypothetical proteins or those containing domains of unknown function (Supplementary Table S3). By contrast, many of the proteins in VR1 share orthology with proteins involved in glycosylation and modification of the flagellar filament.
The flag-2 variable region 1 (VR1) encodes the machinery for glycosylation, methylation and modification of the flagellum
The most prevalent genes among the (520/1826 total VR1 proteins; 28.48%) enterobacterial flag-2 VR1 regions are those that code for twenty-two distinct glycosyltransferases. Glycosyltransferase enzymes catalyze glycosidic bond synthesis resulting in the covalent attachment of a glycan to proteins or other sugars and are likely to contribute towards glycosylation of the main structural protein of the flagellum, flagellin [14,15]. This posttranslational modification has been linked to a wide range of phenotypes, including surface recognition, adhesion, biofilm formation, antigen masking from immune response and virulence [14,15]. Flagellin glycosylation is a relatively common feature among the Enterobacteriaceae and the flag–1 flagellar systems of 307/2,000 enterobacteria (15.4%) were predicted to be glycosylated . Here, 341/592 (57.60%) of the flag-2 loci contain genes coding for glycosyltransferases which are predicted to be involved in flagellin glycosylation. BlastP comparison against the CAZy database , using the dbCAN pipeline , classified these flag-2 proteins into five distinct glycosyltransferase families. Ten distinct proteins belong to the GT2 family (GT2–1 to GT2–10), members of which transfer a wide array of saccharides including mannose, galactose, N-acetylglucosamine, glucose and their derivatives [14,24]. The most common of these GT2 glycosyltransferases is GT2–1, which occurs in 222 Enterobacteriaceae belonging to eight distinct genera. The VR1 of seventeen strains in six genera of Enterobacteriaceae and Erwiniaceae incorporate a gene coding for a glycosyltransferase of the GT4 family, which likewise transfer a broad range of sugars, including glucose, mannose and glucosamine . A further 70 strains encode five distinct glycosyltransferase orthologues (GT9–1 to GT9–5) of the GT9 family, which incorporates lipopolysaccharide N-acetylglucosaminyltransferases and heptosyltransferases . Another represented glycosyltransferase family, GT25 comprises four distinct orthologues (GT25–1 to GT25–4) in a total of twenty-six taxa and includes galactosyltransferases and proteins involved in lipopolysaccharide biosynthesis. Finally, fourteen strains incorporate two different types of GT32 (GT32–1 and GT32–2) family glycosyltransferases with a purported role in mannose, galactose and glucosamine saccharide transfer .
Given that distinct sugars can be incorporated by glycosyltransferases belonging to the different GT families, the type of sugars incorporated in the flagellin glycan cannot solely be determined on the basis of the type of glycosyltransferase present. Proteins that catalyse the synthesis of these sugar moieties may be localized in genomic locations other than the flag-2 locus. However, the VR1 region of 35 Citrobacter Clade A strains encodes orthologues of the sialyltransferase PM0188 (pfam11477; E-value: 2.54E–16) of Pasteurella multocida, suggesting that the flagellin glycan of the latter strains may incorporate neuraminic acid . Orthologues of lipoteichoic acid synthase LtaS (LtaS1 and LtaS2; cd16015; 30 strains across three genera in the Enterobacteriaceae and Hafniaceae) and CDP-glycerol phosphotransferase TagB (TagB1 to TagB8; COG1887; 65 strains across thirteen genera in the Enterobacteriaceae and Hafniaceae) are encoded on the VR1 regions of flag-2 loci. Additionally this region encodes a glycerol–3-phosphate cytidylyltransferase TagD (cd02171; E-value: 4.09E–46) in 235 strains across ten genera and three families. These proteins are central to the synthesis of teichoic acid, phosphodiester-linked polyol glycopolymers that form a major part of the cell wall of most Gram-positive bacteria . The presence of orthologues of genes involved in this function in VR1 suggest that these glycopolymers form part of the flagellin glycan of the enterobacterial flag-2 system.
While the exact glycan sugar moieties of the flagellin glycans are difficult to determine, the presence of orthologues of diverse proteins which may modify or substitute the glycan chains suggest the flag-2 flagellin glycan is heavily decorated as has been observed in the primary flagellar system of many enterobacteria as well as Gram-positive bacteria [14,18]. Among the VR1-encoded proteins are orthologues of the acetyltransferases NeuD (Neuraminic acid acetyltransferase; TIGR03570; E-value: 2.65E–61; Citrobacter Clade A werkmannii AK–8), RimI (N-acetyltransferase; pfam00583; E-value: 1.69E–11; Escherichia sp. nov 1 E1642), OptS (O-phosphoseryl acetylase; PRK06253; E-value: 5.94E–03; two Escherichia albertii strains) and WbbJ (maltose O-acetyltransferase; cd04647; E-value: 3.33E–25; twelve Citrobacter Clade A strains), the pyruvyl transferase WcaK (pfam04230; E-value: 2.59E–16; fourteen Citrobacter Clade A strains) and transaminase WecE (dTDP–4-amino–4,6-dideoxygalactose transaminase; COG0399; E-value: 4.48E–103; twenty-nine Enterobacteriaceae and Hafniaceae strains). Furthermore, fifteen distinct methyltransferases belonging to the FkbM (FkbM1 to –3; TIGR01444), Mtf11 (Mtf11–1–3; pfam08241), Mtf12 (pfam08242), Mtf23 (Mtf23–1 to –4; pfam13489), Mtf24 (Mtf24–1 and –2; pfam13578) and Mtf25 (Mtf25–1 and –2; pfam13649) families are encoded in the flag-2 VR1 regions. This suggests that the flag-2 flagellin glycans are decorated with methyl, acetyl, amino and pyruvyl groups. An additional N-lysine methylase FliB has been observed to be relatively common among the flag-1 loci and flagellin methylation has been suggested to play a role in virulence in Salmonella enterica [14,27]. Orthologous proteins, previously termed LafV (LafV1 to LafV10), occur in the flag-2 VR1 regions of 342/592 (57.58%) enterobacteria advocating that flagellin methylation is also a common phenomenon in the flag-2 flagellar system. While a range of functions have been ascribed to flagellin glycosylation and methylation for the primary flagellar system, its prevalence among the flag-2 loci suggests similar important roles in the latter system, but this needs to be confirmed experimentally.
The enterobacterial flag-2 locus displays a complex evolutionary history of vertical transmission and horizontal acquisition
The relatively low prevalence of flag-2 loci among the Enterobacterales suggests that they may have been acquired through horizontal gene transfer (HGT) in a select set of taxa. Comparison of a phylogeny on the basis of 32 flag-2 proteins conserved among 87 taxa representative of six families and 27 genera and a house-keeping protein phylogeny shows that there is congruence at the genus and family level for some taxa (Figure 4). For example, Yersinia and Rouxiella (Yersiniaceae),, Hafnia and Obesumbacterium (Hafniaceae) and Pragia and Budvicia (Budviciaceae) form cohesive clades in both phylogenies (Figure 4). This suggests the flag-2 loci have been maintained via vertical transmission through speciation events in these evolutionary lineages. However, a more complex evolutionary history can be ascribed to the flag–2 loci of some taxa. The flag-2 locus of Leminorella grimontii ATCC 33999 (Budviciaceae) clusters with those of the Enterobacteriaceae Citrobacter Clade A and C and Escherichia, while that of Chania multitudinisentens RB–25 clusters with Pluralibacter (Enterobacteriaceae)..Similarly, the flag-2 loci of Pantoea brennerii IF5SW-P1 and Pantoea allii LMG 24248 cluster distinctly with Enterobacter and Lelliottia spp. and not with the other taxa, Erwinia and Izhakiella,of the Erwiniaceae. Moreover, the Enterobacteriaceae form three distinct clades when considering their flag–2 loci (Figure 4). This suggests that flag-2 loci have been derived through HGT events in some of these taxa. Genomic regions derived through recent HGT events are often typified by G+C contents that vary substantially from the rest of the genome . In general, the flag–2 loci of all 592 enterobacterial strains have a G+C content only marginally above (average G+C deviation = 0.5%; range: –4.0 to +5.0%) that of the remainder of the genome (Supplementary Table S2). This may be attributed in part to the distinct G+C content of the VR1 and VR2 regions. Exclusion of these regions, however, resulted in even more pronounced G+C deviations for the flag-2 loci (average G+C deviation: +1.1%; range: –4.0 to +5.8%) suggesting horizontal acquisition in a more substantive set of taxa. This is particularly evident in the Enterobacteriaceae, where the flag-2 loci have an average G+C content 2.1% (range –3.1 to + 5.8%) above that of the rest of the genome. It is furthermore evident in those taxa placed in distinct family clades in the flag-2 tree, including L. grimontii ATCC 33999 (G+C deviation: +2.5%), C. multitudinisentens RB–25 (G+C deviation: +2.4%) and P. allii LMG 24248 (G+C deviation: –3.6%). In one of the Enterobacteriaceae clades in the flag-2 tree, G+C deviations are pronounced in Escherichia (G+C deviation: +4.8%), Pseudocitrobacter (G+C deviation: +2.8%) and Siccibacter (-–2.9%). By contrast, the flag–2 loci in taxa belonging to CitrobacterClade A have average G+C deviations of 0.0%, suggesting that the latter may represent ancestral flag-2 loci which have been derived through HGT in the former strains. However, among the flag–2 loci of Escherichia spp. (124 strains) and Yersinia spp. (223 strains) G+C deviations of +2.6 to +5.8% and –3.0 to +1.0% could be observed, indicating complex evolutionary histories for the flag-2 loci even within these apparently stable lineages.
Analysis of the flag-1 protein complement of forty-one motile species across eleven bacterial phyla showed extensive sequence similarity between the twenty-four core proteins conserved among all taxa . It has thus been postulated that a few genes, or even a single precursor, may have given rise to the full complement of genes required for the synthesis of the primary flagellar system through gene duplications, gene fusions and the recruitment of novel genes [29,30]. It is plausible that similar evolutionary processes may have given rise to the extant flag-2 loci, which may explain the complex evolutionary histories of these loci among the Enterobacterales.
The flag-2 flagellar system may represent a multi-functional organ among the Enterobacterales.
A broad range of functions have been ascribed to the flag-1 flagellar system, including swimming and swarming motility, adhesion, biofilm formation, host invasion and colonization . By contrast, little is known about the function(s) of the secondary flagellar system among the Enterobacterales. The deletion (between lfhA and lafU) in Escherichia or frameshift mutation in Escherichia sp. nov. 2 strain 042 was deemed to inactivate the flag-2 system, rendering it non-functional . Phage integration within both the flag-1and flag-2 loci of C. rodentium has been suggested to have resulted in their non-motile phenotype . This is in line with the concept that synthesis and rotation of the primary (flag–1) system, represents a significant metabolic burden on the bacterial cell, accounting for 2.1% of the overall energy requirement in E. coli, where the flag-2 system may not provide a selective advantage to the cell [15,32]. However, several lines of evidence suggest the flag-2 system plays several important roles in other enterobacterial taxa. The flag-2 flagella of Plesiomonas shigelloides have been shown to be essential for swarming motility . Here we have observed flag-2 loci in the genomes of members of the family Budviciaceae, two strains of Budvicia aquatica andfour strains of P. fontium) and Erwiniaceae (Tatumella pytseos ATCC 33301), which lack the flag-1 flagellar system but have nevertheless been described as being capable of flagellar motility . By contrast, L. grimontii ATCC 33999 (Budviciaceae)and Rouxiella chamberiensis 13033 (Enterobacteriaceae),, which lack a flag-1 locus but retain a flag-2 locus, have been described as non-motile . The flag-2 locus of the latter strain is, however, missing the lafZ gene.
While the flag-2 loci of many of the enterobacterial taxa contain deletions or genes with disrupted reading frames, a substantial number (77.87% of strains containing flag-2 loci) appear to encode the full complement of proteins for the synthesis and functioning of this flagellar system. Many of these have functional flag-1 loci, and as such the flag-2systemmay play roles other than in motility in these taxa. The flag-1 system has been implicated in the secretion of virulence factors by pathogenic bacteria, including the phospholipase YplA in Y. enterocolitica, the lipase XlpA and antibacterial xenocin in the entomopathogen Xenorhabdus and invasion antigen (Cia) protein in Campylobacter spp. [4,34,35]. Analysis of the proteins encoded in the VR1 and VR2 regions of the flag-2 loci revealed a number of putative secretion targets among the cargo proteins. The VR1 region of the Escherichia fergusonii YH17130 flag-2 locus incorporates a gene coding for a Type VI secretion system (T6SS) effector protein VgrG (COG3501; E-value: 0.0). This gene is flanked by genes coding for a 1,556 amino acid Rhs-domain containing protein (COG3209; E-value: 8.53E–44) and polymorphic toxin immunity protein Imm26 (pfam15428; E-value: 1.7E–15). Rhs-domain proteins represent toxins that have been found to be involved in inter-bacterial competition and are also secreted via the T6SS, where the Imm26 protein may serve as a protective mechanism against autotoxicity . Furthermore, the VR1 and VR2 regions in the flag-2 locus of two and three Lelliottia spp., respectively encode orthologues of two distinct Haemolysin co-regulated (Hcp) proteins (Hcp1; pfam05638; E-value: 1.48E–30; Hcp2; COG3157; E-value: 1E–40), which likewise serve as secretion effector proteins . The presence of distinct secreted effector proteins in a number of strains imply a putative role for the flag-2 system in secretion. Furthermore, within the VR1 regions of the Escherichia sp. nov. 1 strain E1642 and Lelliottia aquatilis 6331–17 are genes coding for an orthologue of pesticin (Pst; CD16903; E-value: 1.88E–73), a phage lysozyme-like bacteriocin from Y, pestis and an orthologue of the RNase toxin Ntox44 (pfam15607; E-value: 4.52e–16) [38,39], respectively, ascribing a role in antibacterial activity or inter-bacterial competition to the flag-2 system. Finally, four P. fontium (Budviciaceae) incorporate a gene coding for an 879 aa orthologue (96.6% AAI) of the autotransporter MisL (PRK15313; E-value: 6.51E–97). In S. enterica serovar Typhimurium this protein plays a role in adherence, aggregation and biofilm formation .