Evolution of the Chaperone/Usher Assembly Pathway: Fimbrial Classification Goes Greek

Abstract

Summary: Many Proteobacteria use the chaperone/usher pathway to assemble proteinaceous filaments on the bacterial surface. These filaments can curl into fimbrial or nonfimbrial surface structures (e.g., a capsule or spore coat). This article reviews the phylogeny of operons belonging to the chaperone/usher assembly class to explore the utility of establishing a scheme for subdividing them into clades of phylogenetically related gene clusters. Based on usher amino acid sequence comparisons, our analysis shows that the chaperone/usher assembly class is subdivided into six major phylogenetic clades, which we have termed α-, β-, γ-, κ-, π-, and σ-fimbriae. Members of each clade share related operon structures and encode fimbrial subunits with similar protein domains. The proposed classification system offers a simple and convenient method for assigning newly discovered chaperone/usher systems to one of the six major phylogenetic groups.

INTRODUCTION

Proteinaceous nonflagellar filaments were found on the surface of Escherichia coli cells by electron microscopy in 1950 (139), and the term fimbriae was introduced in 1955 in a study demonstrating that these surface structures mediate the adherence of E. coli to eukaryotic cells (i.e., erythrocytes) (82). The term pili is an alternate designation for nonflagellar bacterial filaments, which was introduced in 1959 and is used by many investigators today (41). However, as the oldest valid published name for filamentous surface appendages functioning in adherence, the term fimbriae takes priority (248) and is therefore the one that will be used in this review article.

Since their first description, fimbriae have been identified in a large number of bacterial species. Over the years, the abundance of fimbriae among bacteria has motivated numerous efforts to develop classification schemes. Early attempts to group fimbriae based on morphology and functional characteristics (42, 81, 248) did not reflect the phylogenetic relatedness revealed subsequently by sequence analysis of the corresponding biosynthetic gene clusters and are thus mainly of historical interest. However, some fragments of these early nomenclatures have survived, including the term “type 1 fimbriae” for rigid filaments mediating mannose-sensitive hemagglutination (81) and the term “type IV fimbriae” (originally described as group 4) for a class of polar appendages mediating twitching motility (248). A morphological classification of fimbriae based on their diameters has been proposed more recently (252), but the large range of diameters reported for one and the same fimbrial type raises the question of whether the use of this dimension is a reliable parameter for classification. Furthermore, the sole use of arbitrary criteria (i.e., morphology and functional characteristics) for classifying fimbriae makes it uninformative to know their placement within the classification scheme.

Fimbriae can be distinguished serologically by bacterial agglutination (107), a method used extensively with E. coli to assign fimbrial (F) antigen numbers (246) or coli surface (CS) antigen numbers (102) to fimbrial structures. Although serology is well suited to generate lists of distinct appendages, the method does not lend itself to developing hierarchical classification schemes since it does not readily identify features shared by larger groups of phylogenetically related surface structures. Furthermore, morphological, functional, and serological classification schemes can be applied only to fimbriae expressed in vitro, an important shortcoming given the large amount of fimbrial gene sequences identified since the dawn of the genomic era. This problem is confounded by the observations for Salmonella enterica and E. coli that many of the fimbrial operons identified by whole-genome sequencing are poorly expressed under standard laboratory growth conditions (141, 203).

Classification systems that are based on the genealogy of fimbrial structures are superior in that a phylogenetic relationship has more predictive potential than does an arbitrary classification. Studies on the biochemistry and genetics of fimbrial biosynthesis have given rise to a nomenclature that distinguishes fimbriae based on their assembly mechanism. The major assembly classes present in gram-negative bacteria include those for conjugative fertility (F) fimbriae (F pili) (87, 192, 333), type IV fimbriae (e.g., toxin-coregulated pilus) (6, 134, 216, 240, 319), fimbriae assembled by the extracellular nucleation/precipitation pathway (curli) (126), and fimbriae assembled by the chaperone/usher-dependent pathway (162, 261, 290, 327, 344). This nomenclature has gained popularity because it defines classes of fimbriae in terms of ancestry and evolutionary descent. Members of an assembly class can be readily identified by sequence homology of their fimbrial biosynthesis genes, thus making this classification applicable to putative fimbrial gene clusters identified by whole-genome sequencing.

The chaperone/usher-dependent pathway is the fimbrial assembly class that contains the largest number of representatives by far. Names of fimbriae belonging to the chaperone/usher assembly class are a source of confusion, as they are derived from various serological, functional, and morphological schemes that mask the evolutionary descent of the respective biosynthesis gene clusters from a common ancestor. The establishment of a system to subdivide members of the chaperone/usher-dependent assembly class into clades of phylogenetically related gene clusters would promote clarity, universality, and stability in a theoretical context that is relevant to microbiology. The goal of this article is to explore the utility of such an approach.

CHARACTERISTICS OF THE CHAPERONE/USHER ASSEMBLY PATHWAY

Genes involved in the biosynthesis of fimbriae belonging to the chaperone/usher assembly class are almost invariably clustered into operons. These operons encode, at minimum, three different proteins: a major structural fimbrial subunit, a chaperone, and an usher protein. Fimbrial operons often contain additional genes encoding structural proteins (i.e., minor fimbrial subunits), assembly proteins (i.e., chaperones), or regulatory proteins. This article focuses on an analysis of genes encoding assembly and structural proteins to define groups on the basis of common ancestry.

All assembly and structural proteins contain typical N-terminal signal sequences that allow their transport across the cytoplasmic membrane via the Sec general secretory pathway (210, 267). Upon the transport of fimbrial subunits into the periplasm and cleavage of the signal peptide, the formation of disulfide bonds may be catalyzed by periplasmic disulfide isomerases (e.g., DsbA and SrgA) (24, 36, 155). In the periplasm, fimbrial subunits interact with their cognate chaperones to ensure proper folding. Fimbrial subunits have immunoglobulin (Ig)-like folds, which lack the final antiparallel β-strand (G), creating a deep hydrophobic cleft. Chaperones bind fimbrial subunits by inserting their G₁ donor β-strands into this hydrophobic cleft in a parallel orientation, thus creating a nonclassical Ig-like fold (Fig. 1) (59, 137, 264, 289). In the absence of the chaperone, fimbrial subunits form periplasmic aggregates that are rapidly degraded by the protease DegP (161, 251, 306).

FIG. 1.

Schematic drawing of fimbrial assembly by the chaperone/usher pathway. The N-terminal signal peptide of fimbrial subunits is cleaved during their transport across the cytoplasmic membrane (CM) by the general secretory pathway (Sec). In the periplasm (PP), the chaperone (C) completes the Ig-like fold of fimbrial subunits with its donor β-strand (black arrow). Interaction of the chaperone/fimbrial subunit complexes with the usher protein (U) facilitates the replacement of the chaperone donor β-strand with the N-terminal β-strand of a second subunit, thereby joining subunits into a filament that is transported across the outer membrane (OM). Filaments may curl into helical structures with two subunits per turn (thin, flexible fibrillae), as shown in this schematic drawing. Alternatively, subunits may assemble helices with 3 to 3.4 subunits per turn (thick, rigid fimbriae) or coil up into nonfimbrial, capsule-like structures on the cell surface. In some cases, fimbrial filaments carry a tip adhesin at their distal end, which mediates attachment.

Interaction of the periplasmic chaperone/subunit complexes with the usher, an integral outer membrane protein, facilitates the release of fimbrial subunits and their secretion through the usher channel, with the most distal proteins being secreted first (26, 50, 113, 185, 195, 233, 234, 292, 328). The N terminus of subunits harbors a conserved β-strand motif similar to the G₁ donor β-strand motif of chaperones. The assembly of subunits is thought to proceed through a donor strand exchange reaction in which the G₁ donor β-strand of the chaperone, inserted with a parallel directionality into the subunit's hydrophobic cleft, is replaced by the N-terminal β-strand of a second subunit in the more energetically favorable antiparallel orientation, thereby completing a classical Ig-like fold and joining subunits into a fiber (Fig. 1) (59, 264, 289). For some operons belonging to the chaperone/usher assembly class, these fibers coil up into nonfimbrial, capsule-like structures on the cell surface. However, in most cases, chaperone/usher-type transport systems assemble their major subunits into fimbriae that can be visualized by electron microscopy. Depending on their orientation in the fimbrial filament, major subunits may assemble thin flexible fibrillae (with two subunits per helical turn) or thick rigid fimbriae (with 3 to 3.4 subunits per helical turn). Minor subunits may be incorporated into the fimbrial filament in the form of a tip adhesin, a tip fibrillum, or an adaptor protein joining these structures (12, 46, 154, 185). Fimbrial chaperone/usher transport systems are specific with regard to ushers interfacing with chaperone/subunit complexes from their own systems (278, 292). Even in examples where an usher can be complemented by a very close homologue from another system, the presence of both intact systems results in their operating in parental pairs with respect to chaperone/subunit recognition (176).

Tip adhesins are minor fimbrial subunits that fold their C-terminal and N-terminal amino acids into two separate domains. The C-terminal domain has the Ig-like fold of fimbrial subunits, which contains a cleft due to the absence of the final β-strand. Insertion of the N-terminal β-strand of a fimbrial subunit into this cleft completes the Ig-like fold and connects the tip adhesin to the fimbrial filament (Fig. 1). The N-terminal domain contains the receptor-binding site and is connected to the C-terminal domain with a short amino acid linker, thereby placing the receptor binding surfaces on the tip of the fimbrial filament (47, 75, 127, 142).

TOWARDS A PHYLOGENETIC NOMENCLATURE

Use of Sequence Comparison To Develop Classification Schemes

The chaperone/usher class contains one group, designated the alternate chaperone/usher family, whose members share significant sequence homology to each other but not to other members of this assembly class (286). Operons that do not belong to the alternate chaperone/usher family are sometimes classified as members of a classical chaperone/usher family. For each family, the assembly mechanism involves the export of fimbrial subunits through the action of a periplasmic chaperone and an outer membrane usher protein. Although sequence similarities between the classical and alternate chaperone/usher families are limited, their usher proteins share conserved domains (PFAM00577 and/or COG3188) identified through the Conserved Domain Database component of the National Center for Biotechnology Information (NCBI) Entrez query retrieval system (211, 212). PFAM and COG designations refer to conserved domain entries in the Protein Family (PFAM) and Clusters of Orthologous Groups (COG) databases, which represent motifs and recurrent patterns in protein sequences (94, 324, 325).

Phylogenetic groups of fimbriae can be defined based on sequence comparisons of structural proteins or assembly proteins. A classification scheme based on sequence comparisons of 21 major fimbrial subunits of E. coli and S. enterica distinguishes the alternate chaperone/usher family (defined as class 5) from the classical chaperone/usher family (classes 1, 2, and 3) and other assembly classes, including type IV fimbriae (class 4) and curli (class 6) (204). Phylogenetic trees constructed by sequence comparison of major subunits belonging to the alternate chaperone/usher family demonstrate their common ancestry (8, 102). One group within the classical chaperone/usher family, defined as class 2 (204), contains subunits that assemble into nonfimbrial surface structures, and their N-terminal donor β-strands share common features (369). Sequence comparison of fimbrial subunits is superior to classification schemes based on serology, morphology, or function, as it defines groups by their descent from a common ancestor. However, a shortcoming of this approach is that an operon may contain several subunits that group to distant branches of a phylogenetic tree, which poses a problem for using this method as the sole criterion for classifying putative fimbrial operons since the identity of a major subunit may not be apparent from the nucleotide sequence.

Comparison of assembly proteins belonging to the classical chaperone/usher family reinforces the idea that fimbriae can be clustered into groups with common phylogenies. Computer modeling of the three-dimensional structure of 16 chaperones distinguishes two subfamilies based on the length of an amino acid sequence that corresponds to a loop that connects the F₁ and G₁ β-strands in the E. coli PapD crystal structure (371). A subsequent study on 26 chaperones defined proteins containing a relatively long F₁-G₁ loop as the FGL chaperone subfamily (e.g., MyfB, PsaB, Caf1M, CS3-1, AggD, AfaB, SefB, and CssC) and proteins with a short F₁-G₁ loop as the FGS chaperone subfamily (143). Interestingly, the FGL chaperone subfamily assembles thin fibrillae or nonfimbrial surface structures, while the FGS chaperone subfamily assembles fimbrial filaments (143, 371). Sequence comparison of 31 chaperones by the neighbor-joining method suggests that the classical chaperone/usher family can be divided into several clades, each including members that apparently share a common ancestor that is not shared by another protein outside of the clade (35). This phylogenetic tree suggests that members of the FGL chaperone subfamily (i.e., MyfB, PsaB, Caf1M, CS3-1, AggD, AfaB, NfaE, SefB, and CssC) share a common ancestor. However, this analysis also shows that the FGS chaperone subfamily cannot be defined by a single node or branch on the phylogenetic tree, suggesting that further subdivision is needed to explicitly categorize the respective fimbriae into clades on the basis of common ancestry. Sequence comparison of chaperones may not be ideally suited for developing such a subdivision because some fimbrial operons encode more than one chaperone, thus raising the question as to which protein should be used to assign the respective operon to a phylogenetic group.

The Fimbrial Usher Protein Family

Comparison of usher sequences has been used to derive phylogenetic trees of members of the chaperone/usher assembly class (8, 367). This approach has the advantage that the resulting definition of phylogenetic groups is unambiguous, because all fimbrial operons belonging to the chaperone/usher assembly class contain only a single usher gene, while multiple genes encoding subunits or chaperones may be present.

The fimbrial usher protein (FUP) family is a large and rapidly growing group of proteins that are distributed among genera of the phyla Proteobacteria, Cyanobacteria, and Deinococcus-Thermus (367). To make sense of the diversity within this large protein family, it is helpful to split this group into clades, thus creating a nonoverlapping hierarchical arrangement.

An initial comparison of 58 members of the FUP family distinguished 10 clusters on the basis of common ancestry (367). A revised phylogenetic tree of the FUP family constructed by comparing 189 proteins containing the usher domains PFAM00577 and/or COG3188 is shown in Fig. 2 (a list of the 189 usher proteins, NCBI Entrez accession numbers for protein sequences used in the usher analysis, and GenBank or RefSeq DNA accession numbers from which operon pictures in Fig. 3 to 11 were derived are provided in Table S1 in the supplemental material). Ushers were gathered for analysis by performing a multitude of BLAST searches with various usher amino acid sequences. The 189 ushers analyzed comprise previously reported fimbriae with sufficiently sequenced DNA to visualize the operons to which they belong and those ushers originating from conceptual translations from completed genome sequences. Subunits and chaperones not found in the local DNA surrounding an usher were not included in this analysis.

FIG. 2.

Phylogenetic tree of the FUP family. The graph shows an unrooted phenogram generated using usher amino acid sequences listed in Table S1 in the supplemental material. Ushers are grouped into six fimbrial clades (highlighted in gray) termed α, β, γ, κ, π, and σ. Bootstrap values of nodes defining these clades (indicated by open circles at the base of each fimbrial clade) are shown. The phylogenetic tree and bootstrap values were generated as follows. Usher amino acid sequences were aligned using ClustalW (MacVector 7.2.3) with a BLOSUM 30 matrix for pairwise alignment and a BLOSUM series matrix for multiple alignment, both with default settings. The phylogeny inference package software (PHYLIP 3.65) developed by Felsenstein (93) was used to perform the remaining analyses. Bootstrapping of the aligned usher sequences was performed in triplicate using the bootstrap algorithm of BOOTSTRAP to determine the number of instances, out of 1,000 sets analyzed (using random seeds 123, 345, and 567), where the members of each clade grouped completely behind the displayed node. Protein distance matrices were generated with PROTDIST using a Jones-Taylor-Thornton matrix with default settings for the unrooted tree or set to analyze 1,000 data sets for bootstrapping. Neighbor joining was performed with NEIGHBOR with default settings or set to analyze 1,000 data sets using the random seeds 345, 567, and 789, respectively, for the above-mentioned bootstrapping runs. Consensus trees (not shown) to interpret the bootstrap data were generated using the majority-rule-extended algorithm of CONSENSE. The unrooted phenogram displayed was generated using DRAWTREE without tree improvement iteration.

FIG. 3.

Phylogenetic relationship of operons belonging to the α-fimbriae. The branch of the FUP tree representing α-fimbriae is shown on the left. Numbers at the end of each branch of the phylogenetic tree correspond to the numbers given for each operon in the center. The gene order for each operon is shown on the right (arrows). The predicted functions for each gene product and the sequence homologies to protein families are indicated in the legend at the bottom. IS, homology to insertion sequence (IS) element.

FIG. 11.

Phylogenetic tree of fimbrial chaperones. An unrooted phenogram illustrating the phylogenetic relatedness of the chaperones associated with the ushers listed in Table S1 in the supplemental material is shown. Chaperone amino acid sequences were aligned and analyzed in the same fashion as that described in the legend of Fig. 2. Although PapJ has been shown to act in some capacity as a periplasmic chaperone (326), it and its homologues (PixJ, SfpJ, and PrfJ) were not included in this analysis, as they do not possess a fimbrial chaperone domain, nor do they share sequence similarity with other fimbrial proteins. The six fimbrial clades (α, β, γ, κ, π, and σ) defined by sequence comparison of usher proteins (Fig. 2) are indicated in the phenogram using gray highlighting. Arrows and circled branches indicate operons containing multiple chaperone genes and are inserted to the right and below the phenogram. The positions of chaperone genes in fimbrial operons (open arrows) in the phenogram are indicated by lowercase letters or by black arrows.

The phylogenetic analysis of 189 usher proteins suggests a classification into clades, which is similar but not identical to that proposed based on analysis of 58 usher proteins (367). To avoid confusion with this previous division into clusters 1 to 10 (367) we propose a nomenclature using Greek letters to refer to individual clades. This approach also distinguishes the FUP classification system from previous nomenclatures using Roman or Arabic numerals to define fimbriae of groups 1 to 6 (248), classes 1 to 6 (204), or types 1 to 6 (81).

Using a node-based definition, the FUP family can be divided into six clades, designated α-, β-, γ-, κ-, π-, and σ-fimbriae, each stemming from a common ancestor represented by a node in the phylogenetic tree (Fig. 2). The γ-fimbrial clade is further subdivided into four clades, termed γ₁-, γ₂-, γ₃-, and γ₄-fimbriae. The α-, κ-, π-, and σ-fimbrial clade names were assigned arbitrarily to recall a particular characteristic of the clade or a prominent member as follows: α-fimbriae, alternate chaperone/usher family; κ-fimbriae, K88 (F4) fimbriae; π-fimbriae, pyelonephritis-associated fimbriae (P fimbriae); and σ-fimbriae, spore coat protein U from Myxococcus xanthus. The β- and γ-fimbriae were assigned names alphabetically.

This subdivision of the FUP family largely confirms the subdivisions proposed initially (367), but former FUP clusters 4 and 5 now form a single clade (γ₃-fimbriae). The subdivision of the chaperone/usher class into six FUP clades confirms that the alternate chaperone/usher family (α-fimbriae) (8) contains operons that stem from a common ancestor (Fig. 2). The FGL chaperone subfamily forms a monophyletic group (γ₃-fimbriae) within the classical chaperone/usher family (143). However, the FGS chaperone subfamily is composed of several clades (β-, γ₁-, γ₂-, γ₄-, κ-, and π-fimbriae) that are not more closely related to each other than to the FGL chaperone subfamily (γ₃-fimbriae). The analysis also reveals the existence of a major FUP clade (σ-fimbriae) that was represented only by two usher proteins in a previous analysis (367) and whose members share limited or no sequence homology to members of the alternate chaperone/usher family (α-fimbriae) or the classical chaperone/usher family (β-, γ-, κ-, and π-fimbriae). An analysis of each FUP clade presented below shows that this classification scheme defines groups whose members share common characteristics, thus illustrating the utility of the proposed phylogenetic nomenclature.

THE ALTERNATE CHAPERONE/USHER FAMILY

The α-Fimbriae: Class 5 Fimbriae

The α-fimbrial clade includes the alternate chaperone/usher family (286), a group also known as class 5 fimbriae (204); the colonization factor antigen I (CFA/I)-like group (102); or FUP cluster 8 (367). Phylogenetic trees constructed by comparing fimbrial subunits, chaperones, and ushers demonstrate the common ancestry of members of this family (8, 102). With only 15 representatives from genera belonging to the Gammaproteobacteria and one representative from the Betaproteobacteria, the α-fimbriae are a relatively small but highly divergent group within the FUP family (Fig. 3). Previous studies (8, 102) listed ushers of Burkholderia cepacia (CblC) (282, 332), E. coli (CfaC, CooC, CosC, CotC, CsaC, CsbC, CsdC, and CsuC) (7, 8, 97, 99, 100, 125, 164), S. enterica serotype Typhi (TcfC) (72, 95), and Yersinia pestis (YPO3798) (254) as being members of the alternate chaperone/usher family. The α-fimbrial clade defined in this study includes additional usher proteins, one of which, CsfC (CS5 fimbriae), had not previously been classified as a member of the alternate chaperone/usher family. The α-fimbrial clade thus includes, but is not identical to, the group previously defined as class 5 fimbriae or CFA/I-like fimbriae (8, 102, 204). New members of the α-fimbriae include AHA_0060 and AHA_1021 from Aeromonas hydrophila (300), CsfC and YagX from E. coli (33, 85, 266), and SO_4439 from Shewanella oneidensis (66, 130). Many of the fimbriae belonging to operons of the α-fimbrial clade have been characterized morphologically.

A well-studied member of the α-fimbriae is the coo operon, encoding CS1 fimbriae of human enterotoxigenic E. coli (286). CS1 fimbriae are rigid, 6- to 7-nm fimbrial filaments that do not contain a visible tip fibrillum (102, 194). The fimbrial shaft is formed by the major subunit CooA and carries the CooD tip adhesin at its distal end, both of which are stabilized in the periplasm prior to surface assembly by the chaperone CooB (259, 285, 347). Other fimbriae of human enterotoxigenic E. coli that belong to the α-fimbrial clade include CFA/I (CfaC) (125), CS2 (CotC) (100), CS4 (CsaC) (7), CS5 (CsfC) (60, 85), CS14 (CsuC) (8), CS17 (CsbC) (8), CS19 (CsdC) (8), and PFC071 (CosC) (8). Similar to CS1 fimbriae, most of these organelles comprise a rigid fimbrial shaft of polymerized major subunits (6 to 7 nm in diameter) and an apparently tip-localized minor subunit, which mediates adherence (90, 102, 120, 220, 286). A notable exception is the csf operon, which encodes CS5 fimbriae composed of 2-nm flexible fibrillae (83-85). The only representative of the α-fimbriae that is present in a member of the Betaproteobacteria, the cable pilus (cable fimbriae) of B. cepacia, exhibits a morphological appearance that has been described as giant (2 to 4 μm in length) intertwined fibers (281). Expression of the cbl operon in E. coli results in an elaboration of fimbriae that are morphologically similar to other fimbrial structures belonging to the α-fimbrial clade (282).

The CFA/I, CS1, CS2, CS4, CS5, CS14, CS17, CS19, and PCF071 antigens of human enterotoxigenic E. coli isolates are carried by plasmids (8, 60, 89, 92, 97, 98, 120, 219, 270, 308, 358). The presence of these fimbrial antigens in human isolates is thought to reflect their selective binding to human tissue (102). Adhesion to small intestinal enterocytes or colonic epithelial carcinoma cells implicates CFA/I, CS1, CS2, CS4, and CS5 in mediating the colonization of the human intestine (67, 169, 170, 179, 180, 346). Volunteer studies provide direct evidence of a contribution of CFA/I fimbriae to the colonization of the human intestine with enterotoxigenic E. coli (91). The fact that all fimbrial operons of the α-fimbriae that are present in E. coli encode human-specific colonization factors is somewhat surprising, since it is not apparent why variants with tip adhesins that would bind receptors present in other vertebrate host species were not selected during diversification of this group. Interestingly, another member of the α-fimbrial clade, TcfC, is present in strictly human-adapted pathogens, including S. enterica serotype Typhi and S. enterica serotype Paratyphi A (95, 217, 336). However, fimbriae encoded by the chromosomally located tcf operon have not been functionally characterized.

Operons encoding CFA/I (cfa), CS1 (coo), CS2 (cot), CS4 (csf), CS17 (csb), CS19 (csd), PCF071 (cos), cable fimbriae (cbl), and serotype Typhi colonization factor (tcf) share the same organization, in which a gene encoding a chaperone is followed by genes encoding a major subunit, an usher, and a tip adhesin (7, 8, 100, 125, 336) (Fig. 3). The operon encoding CS14 (csu) has the same operon structure but contains two major subunit genes arranged in tandem (8, 101). Of 189 usher proteins analyzed, only CsuC (CS14 fimbriae) exhibited an E value for the PFAM00577 and COG3188 protein families that was above the cutoff of 0.01 used in this study (E values of 8.6 for PFAM00577 and 0.013 for COG3188) (212). The major fimbrial subunits and tip adhesins of most α-fimbriae contain the conserved protein domains PFAM04449 and PFAM07434, respectively. These protein domains are not found in fimbrial subunits of any other fimbrial clade. The more divergent members of the α-fimbrial clade are located in operons with different gene orders that encode fimbrial subunits with no significant homology to the PFAM04449 or PFAM07434 domains (Fig. 3), which may explain why these operons were not previously classified as being members of the alternate chaperone/usher family. Some chaperones of the α-fimbriae exhibited homology to the protein family COG3121, while only one (TcfA) had homology to the protein family PFAM00345, and no homology to PFAM02753 was detected. CooB and CsfB did not exhibit homology to any of the chaperone protein families, but experimental data show that these proteins function as chaperones for the biosynthesis of CS1 and CS5 fimbriae, respectively (84, 347). Similarly, no COG3121, PFAM00345, or PFAM02753 domains were detected in AHA_1019, CblB, CosB, and YPO3800, and the putative functions of these proteins were deduced from their sequence homologies to known chaperone proteins.

THE CLASSICAL CHAPERONE/USHER FAMILY

The β-Fimbriae

The β-fimbrial clade was previously defined as FUP cluster 7 (367), which contained a single representative, YhcD of E. coli K-12 (33). Additional members of the β-fimbriae identified since that initial report include Bamb_1679 of B. cepacia, BPSS0120 of Burkholderia pseudomallei, StjB of S. enterica serotype Typhimurium, and YPO0302 of Y. pestis (62, 136, 218, 254) (Fig. 4). Unfortunately, all putative fimbrial operons encoding usher proteins of the β-fimbriae were identified by whole-genome sequencing and remain uncharacterized.

FIG. 4.

Phylogenetic relationship of operons belonging to the β-fimbriae. (A) The branch of the FUP tree representing β-fimbriae is shown on the left. Numbers at the end of each branch of the phylogenetic tree correspond to the numbers given for each operon in the center. The gene order for each operon is shown on the right (arrows). The predicted functions for each gene product and the sequence homologies to protein families are indicated in the legend at the bottom. IS, homology to IS element. (B) Comparison of the gltB-yhcG intergenic regions in E. coli strain K-12 and S. enterica serotype Typhimurium strain LT2. Genes are indicated by arrows.

No structural or functional information is available for this small but distinct group of putative fimbrial operons, and none contain genes with homology to fimbrial subunits present in other fimbrial clades. However, in each operon, the genes encoding a chaperone and an usher are preceded by at least one open reading frame that encodes a protein with a domain of unknown function (PFAM06551) (Fig. 4A). The E. coli gltF yhcADEF operon encodes two proteins with a PFAM06551 domain, GltF and YhcF. GltF was initially proposed to be a regulator of glutamate synthase (GltBD) in E. coli (52, 53), but this hypothesis was refuted by subsequent studies (115, 118). Comparisons of the corresponding DNA regions of E. coli K-12 and S. enterica serotype Typhimurium (Fig. 4B) illustrate that the putative gltF yhcADEF fimbrial operon is in fact an insertion downstream of the glutamate synthase genes (gltBD). Given the presence of genes encoding PFAM06551-containing proteins in all operons belonging to the β-fimbrial clade, it is likely that the PFAM06551 protein domain defines fimbrial subunits characteristic of β-fimbriae. Operons of the β-fimbriae do not contain genes resembling typical tip adhesins. The absence of tip adhesins is characteristic of operons that assemble thin fibrillae or nonfimbrial surface structures (see sections below on γ₃-fimbriae and κ-fimbriae). The apparent lack of genes resembling tip adhesins thus suggests that operons of the β-fimbrial clade encode nonfimbrial or fibrillar structures.

The γ-Fimbriae

The γ₁-fimbriae: type 1 fimbria-like operons.

The γ₁-fimbrial clade corresponds to FUP cluster 1 (367) and consists of ushers present in several bacterial species, all belonging to a single family within the Gammaproteobacteria (the Enterobacteriaceae). Representatives are present in Citrobacter freundii (FimD) (133), E. coli (AufC, ECs0594, ECs2110, FimD, FmlC, FocD, SfaF, SfmD, YcbS, and YdeT) (17, 33, 45, 56, 73, 128, 177, 202, 203, 271, 274, 297, 340, 352), Proteus mirabilis (AtfC) (213, 214), and S. enterica serotype Typhimurium (BcfC and FimD) (218, 337). A number of usher proteins are closely related to the γ₁-fimbrial clade but are not connected by a node with a good bootstrapping value (Fig. 5). These usher proteins will be discussed together with the γ₁-fimbriae in the following paragraphs and include ECs4022, ECs4667, LpfC[789], LpfC[EH41], StgC, and YraJ of E. coli (33, 77, 128, 146, 203, 208, 334, 335); PhfD and plu2157 of Photorhabdus luminescens (79); LpfC, SthB, and StiC of S. enterica serotype Typhimurium (28, 218); and StgC of S. enterica serotype Typhi (72, 336). Many of the fimbriae encoded by operons belonging to, or closely related to, the γ₁-fimbrial clade are well studied, with type 1 fimbriae being perhaps the best-known representatives.

FIG. 5.

Phylogenetic relationship of operons belonging to the γ₁-, γ₂-, and γ₃-fimbriae. The branch of the FUP tree representing γ₁-, γ₂-, and γ₃-fimbriae is shown on the left. The bootstrap value for the node defining each subclade is displayed at the top and was generated in the analysis performed for Fig. 2. Numbers at the end of each branch of the phylogenetic tree correspond to the numbers given for each operon in the center. The gene order for each operon is shown on the right (arrows). The predicted functions for each gene product and the sequence homologies to protein families are indicated in the legend at the bottom. IS, homology to IS element; in, homology to invasin domain PFAM05775; rt, homology to reverse transcriptase; ds, homology to DsbA.

Type 1 fimbriae are defined as rigid fimbriae that are 7 to 10 nm in diameter and mediate mannose-sensitive hemagglutination (80). This designation can be confusing, since naturally occurring mutations resulting in a single-amino-acid substitution in the S. enterica FimH tip adhesin can result in a loss of mannose-sensitive hemagglutination (175). Furthermore, gene clusters encoding type 1 fimbriae of S. enterica and E. coli are both referred to as fim operons, although these are not orthologous gene clusters. The S. enterica serotype Typhimurium fimAICDHF operon has an orthologue, termed sfmACDHF, which is present at the corresponding location in the E. coli K-12 chromosome. Comparison of usher sequences illustrates the close relationship between S. enterica FimD and E. coli SfmD (Fig. 5). However, the sfmACDHF operon is cryptic, and type 1 fimbriae of E. coli are encoded by a paralogous operon, fimBEAICDFGH, which is located at a separate region of the chromosome (33, 218). Nonetheless, the E. coli sfmACDHF and fimBEAICDFGH operons share enough sequence similarities for both to be included in the γ₁-fimbrial clade.

E. coli type 1 fimbriae bind to mannosylated glycoproteins in the bladder and are a virulence factor that is important for cystitis caused by uropathogenic E. coli (10, 229, 309, 310). They are composed of a rigid helical fimbrial shaft that ends in a short tip fibrillum (163). The helical shaft has a diameter of 7 nm and consists of 3.4 copies of the FimA major subunit per turn (42, 122). The short tip fibrillum is composed of the FimF and FimG minor fimbrial subunits and carries the FimH tip adhesin at its distal end (122, 163, 279, 291).

The γ₁-fimbriae also contain the ushers of S fimbriae (SfaF) (181) and type 1C fimbriae (FocD) (271, 340) from uropathogenic E. coli. The presence of S fimbriae and type 1C fimbriae in E. coli is associated with urinary tract infection (160, 168, 257, 258, 302). Fimbriae assembled by SfaF and FocD are morphologically similar to type 1 fimbriae but exhibit different binding activities. S fimbriae attach to sialic acid-containing receptors and mediate neuraminidase-sensitive hemagglutination (181, 255), while type 1C fimbriae bind GalNAc(β1-4)β-galactosidase residues and do not mediate hemagglutination (172, 178, 340). Uropathogenic P. mirabilis isolates also contain a member of the γ₁-fimbrial clade, the ambient-temperature fimbria (atf) operon; however, the encoded adhesin is not required for cystitis in a mouse model (213, 214, 373).

Gene clusters that are closely related to the γ₁-fimbriae include the lpf operons of S. enterica (28), enterohemorrhagic E. coli (77, 334, 335), enteropathogenic E. coli (232), and extraintestinal pathogenic E. coli (146). In S. enterica serotype Typhimurium, the lpf operon has been implicated in intestinal colonization in a mouse model (29, 351).

The core structure of operons belonging to the γ₁-fimbriae is composed of a major subunit gene followed by genes encoding a chaperone and an usher and a minor subunit encoding a tip adhesin. Additional minor subunits may be present at different positions in the operon (Fig. 5). Subunits of γ₁-fimbriae contain the PFAM00419 and COG3539 fimbrial conserved domains, which are also found in subunits of γ₂-fimbriae, γ₄-fimbriae, π-fimbriae, and some of the κ-fimbriae.

The γ₂-fimbriae.

The γ₂-fimbrial clade represents FUP cluster 2 (367) and contains usher proteins assembling the CS12 fimbriae (CswD) (322), the CS18 fimbriae (FotD) (138), and the F6 fimbrial antigen (987P fimbriae) (FasD) (152, 295) of E. coli. Uncharacterized members include ECP_3814 of E. coli (135) and plu0268, plu0784, and plu0791 of P. luminescens (79) (Fig. 5).

The best-studied member of the γ₂-fimbriae is the fas operon, which encodes the F6 fimbrial antigen (987P fimbriae) (295). The fas operon is commonly carried by plasmids in porcine isolates of enterotoxigenic E. coli (296). F6 fimbriae are rigid filaments that are 7 nm in diameter (153). Attachment is mediated by a tip adhesin, FasG, which is linked by a minor subunit, FasF, to the rigid fimbrial shaft composed of the major subunit FasA (49, 173). Assembly of F6 fimbriae involves three chaperones, FasB, FasC, and FasE. The FasB chaperone interacts with the major fimbrial subunit FasA but not with FasG. Conversely, FasC interacts only with FasG but not FasA, suggesting that chaperones are subunit specific (86). F6 fimbriae bind to receptors found in the brush border of piglet intestinal cells (57, 58, 71, 171), which have recently been identified as histone H1 proteins (372).

CS12 and CS18 are fimbriae that are present in human isolates of enterotoxigenic E. coli (132, 138). CS18 fimbriae are rigid filaments that are 6 to 8 nm in diameter (345). The morphologies of F6 fimbriae and CS18 fimbriae suggest that a type 1 fimbria-like appearance may be a shared feature of adhesins belonging to the γ₂-fimbrial clade.

A distinguishing characteristic of operons belonging to the γ₂-fimbriae is the presence of three chaperone genes (Fig. 5). While all three chaperones of the plu0261-268, plu0779-0785, and plu0786-0792 operons each contain three chaperone domains (COG3121, PFAM00345, and PFAM02753), the chaperones of the csw, ECP_3811 to ECP_3817, fas, and fot operons are more varied. The first chaperone in the operon (CswB, ECP_3812, FasB, and FotB) contains three chaperone domains (COG3121, PFAM00345, and PFAM02753), while the third chaperone (CswE, ECP_3815, FasE, and FotE) contains only two (COG3121 and PFAM00345). The second chaperone contains a COG3121 (ECP_3813 [FasC]) or PFAM00345 domain (FasC) or no domain at all (CswC and FotC), in which case the putative function of these proteins was assigned based on sequence homology to the known chaperone FasC (86). As with γ₁-fimbriae, most of the fimbrial subunits of γ₂-fimbriae contain the conserved domains PFAM00419 and COG3539. However, unlike γ₁-fimbriae, the PFAM00419 and COG3539 domains are not present in the C-terminal half of tip adhesins.

The γ₃-fimbriae: the FGL chaperone subfamily.

Representatives of the γ₃-fimbrial clade include AfaC-3, AfaC-7, AfaC-8, AggC, Agg3C, CS3-2, CssD, DraC, and HdaC of E. coli (32, 39, 76, 105, 158, 187, 189, 190, 238, 239, 293, 361, 363, 366); PSPPH_A0064 of Pseudomonas syringae (159); SefC of S. enterica serotype Enteritidis (61); SefC of S. enterica serotype Paratyphi A (217); SafC of S. enterica serotype Typhimurium (95, 218); MyfC of Yersinia enterocolitica (150); and Caf1A and PsaC of Y. pestis (199, 254, 370) (Fig. 5). The operons assigned to the γ₃-fimbriae based on sequence homology of their usher sequences are identical to those belonging to the FGL chaperone subfamily, defined by structural similarities of their chaperones (143). The γ₃-fimbriae include two groups previously defined based on sequence comparisons of usher proteins, FUP clusters 4 and 5 (367). Furthermore, one group of fimbriae previously defined by sequence similarities of fimbrial major subunits (i.e., class 2) (204) consists of representatives that all belong to the γ₃-fimbrial clade.

The fact that the γ₃-fimbrial clade is readily distinguishable from other subfamilies on the basis of shared features in both their assembly proteins (i.e., chaperones and ushers) (143, 367, 371) and their structural proteins (i.e., major subunits) (204, 369) may reflect the distinct morphology of the assembled surface structures. Some members of the γ₃-fimbriae assemble their major subunits into nonfimbrial structures on the cell surface (e.g., afa, nfa, css, and caf operons) (4, 65, 124, 131, 183, 205, 361, 369). These nonfimbrial structures are thought to be composed of fibrillae that are too thin to be resolved by electron microscopy. Other members of the γ₃-fimbrial clade assemble very thin (2 to 3 nm in diameter), flexible, fibrillar organelles that can be visualized by electron microscopy (e.g., dra, agg, psa, myf, cs3, and sef operons) (61, 150, 194, 200, 230, 236, 293, 353).

Generally, organelles belonging to the γ₃-fimbriae are composed of only one or two structural subunits and do not contain specialized subunits with homology to tip adhesins. The dra, daa, nfa, and afa operons encode adhesins of the Dr family, which mediate agglutination of erythrocytes carrying the Dr blood group antigen by binding to decay-accelerating factor (also known as CD55) and carcinoembryonic antigen-related proteins (9, 31, 140, 182, 188, 237-239, 260, 262, 342). The afa-3 operon encodes a nonfimbrial structure in which the AfaE major subunit assembles into fibers that are capped by the AfaD minor subunit (65). Studies of the close homologue DraE suggest that major subunit assembly can also proceed in the absence of the minor subunit cap (368). The AfaD minor subunit is an invasin that mediates bacterial entry into HeLa cervical carcinoma cells (104, 116, 165) via an interaction with an α₅β₁ integrin (121). However, binding to the decay-accelerating factor or carcinoembryonic antigen-related receptor of Dr family adhesins is mediated by the AfaE major subunit (31, 103, 182, 342, 343). The distinction between major subunits involved in adhesion and minor subunits involved in invasion may not be so rigid for all Dr family adhesins, as the major subunit DraE has been implicated in facilitating cell invasion (68). Members of the Dr family are commonly present in diffusely adhering E. coli isolates associated with diarrhea or urinary tract infections (299).

The γ₃-fimbrial clade contains two adhesins that are associated with human isolates of enterotoxigenic E. coli, CS3 and CS6. The CS3 antigen mediates attachment to human colonic epithelial cancer cells (231) and forms thin, flexible fibrillae with a diameter of 2 nm on the bacterial surface (194). The css operon encodes the CS6 antigen of human enterotoxigenic E. coli, which is present on the cell surface but has a morphology that is beyond the limits of resolution by electron microscopy (205, 361). The css operon contains two major subunit genes, and the encoded proteins (CssA and CssB) are present at a 3:1 ratio on the bacterial surface (363, 366). The assembly of this surface structure results in the binding of E. coli to human enterocytes (132).

In Y. pestis, the caf operon encodes the fraction 1 (F1) capsule-like antigen, which confers antiphagocytic properties (78). The major subunit Caf1 assembles into fibers that form a large gel-like capsule on the bacterial surface (18, 369).

Operons of the γ₃-fimbriae share an arrangement of genes encoding a major subunit, a chaperone, an usher, and a minor subunit. Some operons lack a subunit gene at the beginning of the operon, and the major subunit is encoded by a gene located downstream of the usher gene. Unlike other members of the γ-fimbriae, fimbrial subunits of the γ₃-fimbriae do not contain the conserved domains PFAM00419 and COG3539. The S. enterica sef operon shown in Fig. 5 is from the genome of serotype Paratyphi A and does not contain a minor subunit. However, sef operons present in other S. enterica serotypes (e.g., serotype Enteritidis) carry a minor subunit gene (sefD) (61) encoding a protein containing a PFAM05775 domain. The PFAM05775 domain is also found in all invasins (i.e., minor subunits) of the Dr family of adhesins.

The γ₄-fimbriae: Acu, Hif, and Mrk.

The γ₄-fimbrial clade corresponds to FUP family 10 (367) and contains usher proteins of bacterial genera representing two classes, the Betaproteobacteria and the Gammaproteobacteria. The majority of operons belonging to the γ₄-fimbriae were identified by whole-genome sequencing, and only a few members, the acu, hif, and mrk operons, are well characterized (Fig. 6). Representatives of the γ₄-fimbrial clade are present in Acinetobacter spp. (ACIAD0121, ACIAD0389, and AcuC) (23, 110), Bordetella avium (FimC) (311), Bordetella pertussis (FhaA or FimC) (201, 253), B. cepacia (Bamb_2831 and Bamb_4024) (62), B. pseudomallei (BPSS0093) (136), Burkholderia xenovorans (Bxe_B0728, Bxe_B2970, and Bxe_C1151), (54), Chromobacterium violaceum (CV_1296) (70), E. coli (ECs0022, ECs0143, ECs1278, ECs2915, HtrE, UTI89_C4905, and YehB) (33, 56, 128, 203, 268), Haemophilus spp. (GhfC, HafC, HifC[86-0295], and HifC[AM30]) (43, 117, 269, 341, 349), Klebsiella pneumoniae (MrkC) (5), P. luminescens (plu0505) (79), Pseudomonas aeruginosa (CupA3 and CupB3) (278, 317, 339), Pseudomonas fluorescens (PFL_1464) (256), Ralstonia eutropha (H16_B2090) (263), Ralstonia metallidurans (Rmet_0574, Rmet_1665, Rmet_4252, and Rmet_4959), S. enterica serotype Typhi (StaC) (72, 336), S. enterica serotype Typhimurium (StbC and StcC) (218), Xylella fastidiosa (XF0081) (305), and Y. pestis (YPO1709, YPO1920, and YPO3879) (254).

FIG. 6.

Phylogenetic relationship of operons belonging to the γ₄-fimbriae. The branch of the FUP tree representing γ₄-fimbriae is shown on the left. The bootstrap value for the node defining the γ₄-fimbriae is displayed at the top and was generated in the analysis performed for Fig. 2. Numbers at the end of each branch of the phylogenetic tree correspond to the numbers given for each operon in the center. The gene order for each operon is shown on the right (arrows). The predicted functions for each gene product and the sequence homologies to protein families are indicated in the legend at the bottom.

The node connecting usher proteins BPSL1628 of B. pseudomallei, CupC3 of P. aeruginosa, EtfC of Edwardsiella ictaluri, EtfC of Edwardsiella tarda, PFL_3924 of P. fluorescens, and PSPPH_1201 and PSPPH_2011 of P. syringae (136, 159, 256, 278, 283, 284, 317, 339), as a sister group to the γ₄-fimbrial clade, has a low bootstrapping value (37.3%), suggesting that there is insufficient information to decide where to place them among the γ-fimbriae (Fig. 2 and 6). One member, CupC, assembles structures exhibiting a fimbrial morphology (278), but the remaining members of this group remain uncharacterized.

The mrk operon of K. pneumoniae encodes thin (2- to 4-nm-wide) fimbriae (5), termed type 3 fimbriae based on their ability to mediate mannose-resistant agglutination of tannic acid-treated erythrocytes (80). The fimbrial shaft is composed of the major subunit MrkA, while the mrkD gene encodes a minor subunit that is located at the tip of the fimbrial filaments and mediates adhesion (106, 157, 191, 323). The hif operon of H. influenzae (341, 349) encodes fimbriae that agglutinate erythrocytes carrying the blood group AnWj antigen (64) and promotes colonization of the upper respiratory tract (108, 109, 184, 350). The 7-nm-wide fimbrial filaments are composed of a helical shaft containing three copies of the HifA major subunit per turn (228). The filaments carry a short tip fibrillum composed of the minor subunit HifD (312), which contains the minor subunit HifE located at its tip (221). AcuC, the usher protein assembling thin fimbriae (pili) of Acinetobacter spp., is closely related to MrkC and HifC (Fig. 6). The acuADCG operon encodes thin (2- to 3-nm) fimbriae that are packed into right-handed bundles (110). The fimABCD operon of B. pertussis (201, 355-357) encodes serotype 2 fimbriae, which have diameters of 5 to 6 nm and are composed of a helical shaft containing 2.5 copies of the major subunit per turn (313).

The core operon structure present in all members of the γ₄-fimbrial clade consists of genes encoding a major subunit, a chaperone, and an usher (Fig. 6). In the majority of operons, a gene encoding a minor fimbrial subunit resembling the MrkD and HifE tip adhesins is present downstream of the usher gene. In some cases, additional minor subunits may be present, which assemble into a tip fibrillum (e.g., HifD) (314). A subgroup of operons carries a second chaperone gene downstream of the gene encoding the tip adhesin. Like γ₁-fimbriae and γ₂-fimbriae, operons belonging to the γ₄-fimbriae encode subunits containing the conserved domains PFAM00419 and COG3539.

The κ-Fimbriae: Flexible Fibrillae

All usher proteins belonging to the κ-fimbrial clade are present in bacteria belonging to the family Enterobacteriaceae (Fig. 7). The κ-fimbriae (FUP cluster 6) (367) comprise usher proteins required for the assembly of the F4 fimbrial antigen (K88 fimbriae) (FaeD) (20, 22, 224, 245, 315, 316, 338), the F5 fimbrial antigen (K99 fimbriae) (FanD) (1, 20, 149, 151, 247, 275-277, 303), and the F18 fimbrial antigen (F107 fimbriae) (FedB) (147, 148, 307) found in porcine and bovine isolates of enterotoxigenic E. coli (51, 362). Furthermore, this group includes adherence factor/rabbit 1 (AF/R1) fimbriae (AfrB) (48), an uncharacterized operon of E. coli (CshB), the locus for diffuse adherence (LdaD) of atypical enteropathogenic E. coli (294), plasmid-encoded fimbriae (PefC) of S. enterica serotype Typhimurium (96, 218), and an intestinal colonization factor of rabbit enteropathogenic E. coli (RalD) (2).

FIG. 7.

Phylogenetic relationship of operons belonging to the κ-fimbriae. The branch of the FUP tree representing κ-fimbriae is shown on the left. Numbers at the end of each branch of the phylogenetic tree correspond to the numbers given for each operon in the center. The gene order for each operon is shown on the right (arrows). The predicted functions for each gene product and the sequence homologies to protein families are indicated in the legend at the bottom.

In addition to sequence homology of their usher proteins, members of the κ-fimbrial clade share morphological characteristics; that is, each operon encodes thin (2- to 5-nm-diameter) flexible fibrillae (2, 96, 123, 272) that can be readily distinguished morphologically from the thicker, rigid fimbriae assembled by operons of the γ- or π-fimbrial clades. The only exception is the ldaCDEFGHI operon, which encodes subunits that are assembled into a nonfimbrial structure on the bacterial surface (294). In the case of F18 fibrillae, the major subunit FedA is assembled into a helix containing two subunits per turn (123). Minor fimbrial subunits have been implicated in the adherence of F18 fimbriae and AF/R1 fimbriae (48, 148, 307). However, in the case of F4 and F5 fimbriae, minor subunits are dispensable for adherence (22, 303, 304), and the major fimbrial subunits (FaeG and FanC) have been implicated in determining the binding activity of their respective adhesins (21, 156, 250).

Many members of the κ-fimbriae that are present in E. coli or S. enterica are carried by conjugative plasmids (2, 3, 40, 96, 242, 301, 359, 360). Conjugative plasmids with similarity to the well-characterized fertility (F) plasmid are found in a considerable fraction of E. coli isolates (38). An F-like OriT is also present on the virulence plasmids of S. enterica serotype Typhimurium, serotype Enteritidis, and serotype Gallinarum (249). Comparative sequencing of genes (traD, traY, and finO) involved in the DNA transfer of E. coli F-like plasmids reveals close homology to the corresponding genes on the virulence plasmid of S. enterica serotype Typhimurium, which carries the pef operon (37). The average pairwise differences between F-plasmid genes (traD, traY, and finO) present in E. coli and S. enterica isolates are similar to those observed between different isolates within each species. These data suggest that the genera Escherichia and Salmonella share an F-like plasmid pool (37). The plasmids contained within the shared F-like plasmid pool are mosaic in structure, with different regions acquired from different sources (38). For example, the DNA regions upstream and downstream of the pef operon are conserved between the F-like virulence plasmid of S. enterica serotype Typhimurium and the S. enterica serotype Gallinarum virulence plasmid. However, the virulence plasmid of S. enterica serotype Gallinarum carries a different fimbrial operon at the corresponding position. Interestingly, sequence analysis shows that the fimbrial proteins encoded by the S. enterica serotype Gallinarum virulence plasmid share high sequence identity to proteins of the E. coli F4 fimbrial antigen (76% identity to FaeH and 70% identity to FaeI) (280). Collectively, these data suggest that E. coli and S. enterica have exchanged fimbrial operons belonging to the κ-fimbrial clade by interspecies conjugative transfer of F-like plasmids.

Members of the κ-fimbriae share similarities in their operon structures and lack a gene encoding a tip adhesin (Fig. 7). Operons belonging to the κ-fimbrial clade also share a common core structure composed of genes encoding a major subunit, an usher, and a chaperone (Fig. 7). Some operons encode subunits that contain the conserved domains PFAM00419 and COG3539 present in subunits of π-fimbriae and γ-fimbriae. However, fimbrial subunits encoded by the csh, fae, lda, and ral operons contain the PFAM02432 domain, which is present only in members of the κ-fimbriae (Fig. 7).

The π-Fimbriae: P-Fimbria-Like Operons

Usher proteins of the π-fimbrial clade (FUP cluster 3) (367) are found in bacteria belonging to the Betaproteobacteria and the Gammaproteobacteria. Members of the π-fimbriae are present in A. hydrophila (AHA_0521) (300), B. cepacia (Bamb_1524) (62), B. pseudomallei (BPSL1800) (136), Erwinia carotovora (ECA0942) (30), E. coli (PapC, PrfC, PixC, SfpC, YbgQ, YfcUT, YqiG, ECs0743, and ECs3221) (33, 44, 56, 74, 128, 203, 206, 226, 235, 321), Haemophilus ducreyi (HD_0283), P. luminescens (MrfC[K122], MrfC[TTO1], and plu0416) (79, 223), P. mirabilis (MrpC and PmfC) (13, 16, 215, 243), S. enterica serotype Paratyphi A (SteB) (217), S. enterica serotype Typhimurium (StdB and StfC) (88, 218, 225), Serratia entomophila (SefC) (144, 145), Xenorhabdus nematophila (MrxC) (129), and Y. pestis (YPO0698) (254) (Fig. 8). While most members of the π-fimbriae remain poorly characterized, the pap and mrp operons are well studied. The pap operon in particular has served as a paradigm to elucidate some of the mechanisms by which the chaperone/usher pathway assembles fimbrial organelles.

FIG. 8.

Phylogenetic relationship of operons belonging to the π-fimbriae. The branch of the FUP tree representing π-fimbriae is shown on the left. Numbers at the end of each branch of the phylogenetic tree correspond to the numbers given for each operon in the center. The gene order for each operon is shown on the right (arrows). The predicted functions for each gene product and the sequence homologies to protein families are indicated in the legend at the bottom. NG, not grouped but related to π-fimbriae; IS, homology to IS element.

The pap operon encodes pyelonephritis-associated (P) fimbriae that allow uropathogenic E. coli to bind Galα(1-4)Gal moieties present in glycolipids expressed by kidney epithelial cells (34, 310, 312, 318, 320, 321). P fimbriae are composed of a rigid helical fimbrial shaft that ends in a flexible tip fibrillum (46, 185). The helical shaft has a diameter of 8 nm and is composed of 3.3 copies of the PapA major subunit per turn (46, 227). It is thought to be connected to the outer membrane by the minor subunit PapH (12). The tip fibrillum composed of the minor subunit PapE is 2 nm wide and carries the tip adhesin PapG in a single copy at its distal end (185, 207, 273). The tip adhesin PapG is joined to the tip fibrillum by the PapF minor subunit, which serves as an adaptor protein (154, 185). Another minor subunit encoded by the pap operon, PapK, serves as an adaptor protein to link the tip fibrillum to the helical fimbrial shaft (154). While PapD is the periplasmic chaperone for the P-fimbrial subunits (25, 186, 306, 365), PapJ plays an auxiliary role by ensuring the proper integration of PapA into the fimbrial shaft (326). E. coli fimbrial operons that are closely related to pap include prf (P-related fimbriae) (226), pix (Pix fimbriae) (206), and sfp (Sfp fimbriae) (44). While P-related fimbriae and Pix fimbriae are morphologically similar to P fimbriae, Sfp fimbriae are only 3 to 5 nm in diameter (44).

Mannose-resistant/Proteus-like (MR/P) fimbriae encoded by the mrp operon of P. mirabilis contribute to the colonization of the urinary tract and the development of pyelonephritis in a mouse model (14, 16). MR/P fimbriae decorate the bacterial surface as rigid fimbrial filaments with diameters of 6 to 7 nm, which are composed of the major subunit MrpA (15, 243). The tip adhesin, MrpH, mediates the binding of MR/P fimbriae to erythrocytes (196) and is linked to the fimbrial shaft by the minor subunit MrpG (198). The minor subunit MrpB terminates fimbrial assembly and is thought to be a functional homologue of PapH (197).

Genes encoding a major fimbrial subunit, an usher, and a chaperone form the core structure of operons belonging to the π-fimbrial clade (Fig. 8). This gene order is shared with the κ-fimbriae but is distinct from core structures found in all other FUP clades in which the chaperone normally precedes the usher. Each operon contains a gene located downstream of the chaperone gene and encodes either a protein with homology to tip adhesins (e.g., PapG and MrpH) or a similarly sized protein of unknown function. A large subgroup within the π-fimbriae contains an additional three minor subunit genes that are located between the genes encoding the chaperone and the tip adhesin. Major subunits, minor subunits, and some of the tip adhesins contain the conserved domains PFAM00419 and COG3539, which are also present in subunits of the κ-fimbriae and γ-fimbriae. Distinct protein domains have been described for the N-terminal (PFAM03627) and C-terminal (PFAM03628) portions of the tip adhesins of P fimbriae and P-related fimbriae (Fig. 8).

An usher protein from Alcanivorax borkumensis (ABO_0125) (298), a member of the order Oceanospirillales, forms a sister group of the π-fimbriae (Fig. 8). However, the corresponding node is supported by a low bootstrapping value (35.9%), suggesting that not enough information is available at this time to decide whether the operon is related more closely to the π- or the κ-fimbrial clade based upon the usher sequence alone. Nevertheless, the presence of a gene resembling a tip adhesin (ABO_0123) at the end of the operon suggests that it is most closely related to the π-fimbrial clade.

ARCHAIC CHAPERONE/USHER SYSTEMS

The σ-Fimbriae

The σ-fimbrial clade contains usher proteins that contain a PFAM00577 domain but that share limited sequence similarity to usher proteins of the alternate chaperone/usher family (α-fimbriae) or the classical chaperone/usher family (clades β, γ, κ, and π). The group was initially proposed as FUP cluster 9, which contained only two usher proteins, Orf1 (PA4652) of Pseudomonas aeruginosa and CsuD of Vibrio parahaemolyticus, both members of the Gammaproteobacteria (367). Our analysis expands the σ-fimbriae considerably to include representatives of the phyla Proteobacteria, Cyanobacteria, and Deinococcus-Thermus, making it the most widely distributed FUP cluster (Fig. 9). This wide phylogenetic distribution suggests that the σ-fimbrial clade represents an archaic group within the chaperone/usher assembly class.

FIG. 9.

Phylogenetic relationship of operons belonging to the σ-fimbriae. The branch of the FUP tree representing σ-fimbriae is shown on the left. Numbers at the end of each branch of the phylogenetic tree correspond to the numbers given for each operon in the center. The gene order for each operon is shown on the right (arrows). The predicted functions for each gene product and the sequence homologies to protein families are indicated in the legend at the bottom. fk, homology to flagellum hook-associated protein; SCPU, spore coat protein U.

Within the Gammaproteobacteria, representatives of the σ-fimbriae are present in A. borkumensis (ABO_0701), Acinetobacter spp. (ACIAD3334), Acinetobacter baumannii (CsuD), E. carotovora (ECA3076), Methylococcus capsulatus (MCA0304), Alkalilimnicola ehrlichei (Mlg_0778), Pseudoalteromonas haloplanktis (PSHAa2122), P. aeruginosa (PA4652), P. fluorescens (PFL_3950), Psychrobacter cryohalolentis (Pcryo_1831), Vibrio vulnificus (VV1_2741 and VV1520), Vibrio parahaemolyticus (CsuD and VPA1504), Xanthomonas axonopodis (XAC1425), Xanthomonas campestris (XCC1379 and XCV1482), Xanthomonas oryzae (XOO1871[MAFF311018] and XOO1980[KACC10331]), and Y. pestis (YPO1696) (19, 23, 30, 55, 69, 174, 193, 209, 222, 244, 254, 256, 298, 317, 329, 331, 348). Members of the σ-fimbrial clade are present in several Betaproteobacteria including B. cepacia (Bamb_2475), B. pseudomallei (BPSL1007 and BPSL2027), B. xenovorans (Bxe_B0342), Polaromonas spp. (Bpro_0610), and Ralstonia solanacearum (RSp1498 and RSc2693) (54, 62, 63, 136, 287). Within the Alphaproteobacteria, σ-fimbriae are present in Agrobacterium tumefaciens (Atu4731), Granulibacter bethesdensis (GbCGDNIH1_1329), Hyphomonas neptunium (HNE_0088 and HNE_2495), Mesorhizobium loti (mll7724), Novosphingobium aromaticivorans (Saro_0764), and Rickettsia bellii (RBE_0977) (11, 114, 119, 166, 241, 364). Members of the σ-fimbrial clade are present in three Deltaproteobacteria, including Anaeromyxobacter dehalogenans (Adeh_3599), Pelobacter carbinolicus (Pcar_2057), and Myxococcus xanthus (MXAN_3883) (111, 112, 288). Finally, usher proteins of the σ-fimbriae are found one each in the phyla Deinococcus-Thermus (DR_B0063 of Deinococcus radiodurans) (354) and Cyanobacteria (Slr0019 of Synechocystis spp.) (167). Although the phylum Deinococcus-Thermus is classified as being gram positive, D. radiodurans has a complex cell wall profile that includes an outer membrane-like structure (330), which may explain the requirement for a specialized secretion system of the chaperone/usher type to transport proteins to the cell surface.

Given the wide phylogenetic distribution of operons belonging to the σ-fimbriae, surprisingly little is known about the morphology or function of the encoded surface structures. The little that we know about this clade is derived from studies of A. baumannii and M. xanthus, which implicate σ-fimbriae in biofilm formation and spore coat formation, respectively.

Analysis of biofilm-deficient transposon mutants of A. baumannii demonstrates that the csu fimbrial operon is required for biofilm formation and for the elaboration of fimbriae on the bacterial surface (331). These data suggest that at least some of the members of the σ-fimbriae assemble their subunits into fimbrial filaments. The open reading frames MXAN_3885 to MXAN_3882 in the M. xanthus genome encode a member of the σ-fimbrial clade (111). The first gene in this operon encodes protein U, a spore coat protein produced at the late stage of development of M. xanthus (112). These data suggest that some members of the σ-fimbrial clade may assemble their subunits into nonfimbrial structures on the bacterial surface. However, much work remains to be done to generate basic insights into the properties and characteristics of the archaic σ-fimbrial clade.

The core operon structure of σ-fimbriae consists of genes encoding a major subunit, a chaperone, an usher, and a tip adhesin (Fig. 9). Most subunits of σ-fimbriae have significant homology to COG5430, a conserved protein domain that is characteristic of archaic chaperone usher systems. Chaperones of the σ-fimbriae contain either a COG3121 domain only or a COG3121 domain and a PFAM00345 domain.

EVOLUTION OF FIMBRIAL OPERONS

Comparative analysis shows that fimbrial operons represent hypervariable DNA regions in S. enterica genomes, indicative of their frequent horizontal transfer between different organisms (lateral gene transfer) (27, 265, 336). This variability is still greater between species, as illustrated by the fact that of the 12 chaperone/usher-type fimbrial operons present in S. enterica serotype Typhimurium, only fim and stc are well conserved in the closely related organism E. coli (33, 218). Their limited phylogenetic distributions and frequent lateral transfers between species make fimbrial operons not suited for inferring relationships between bacteria. Instead, the utility of delineating the genealogy of fimbrial operons lies in its value for predicting their evolutionary relationship and the morphology of the encoded surface structures. For instance, operons of the γ₃- and κ-fimbrial clades encode exclusively fibrillae and nonfimbrial structures, while operons belonging to the α-, γ₁-, γ₂-, γ₄-, and π-fimbrial clades exclusively assemble fimbrial structures.

It is informative to know the placement of members of the chaperone/usher assembly class within the usher-based classification scheme to make sense of the diversity in their operon structures. The proposed classification scheme divides the classical chaperone/usher family into clades of fimbrial operons with related gene orders (Fig. 3 to 9). In most cases, the divergence of operon structures within each clade can be explained by proposing a single duplication, acquisition, or deletion event, which illustrates that the suggested evolutionary scenario is highly parsimonious. Thus, the gene order within fimbrial operons concurs in supporting the division of the chaperone/usher assembly class into clades based on the sequence homology of usher proteins.

When the most common operon structures within each clade are placed at the end of each of the corresponding branches on the FUP tree, an evolutionary scenario explaining the divergence of operon structures from a common precursor can be derived (Fig. 10). The exact relationship between major clusters in the FUP tree (Fig. 2) is currently not clear, since the nodes connecting the α-, β-, γ-, κ-, π-, and σ-fimbrial clades are supported by low bootstrapping values. However, the operon structures of κ- and π-fimbriae are related to each other, as both share a core structure composed of genes encoding a major subunit, an usher, and a chaperone. Furthermore, a close relationship of the γ-fimbriae to the κ- and π-fimbriae is indicated by the presence of a PFAM00419 domain exclusively in subunits of operons belonging to these three clades (Fig. 10). These data suggest that members of the γ-, κ-, and π-fimbrial clades form a monophyletic group, which will be referred to as the γκπ cluster from here on. The resulting subdivision of the chaperone/usher assembly class into four major groups is supported by the presence of distinct protein domains in their respective fimbrial subunits, including PFAM00419 in the γκπ cluster, PFAM06551 in the β-fimbriae, PFAM04449 in the α-fimbriae, and COG5430 in the σ-fimbriae. The γκπ cluster and the β-fimbriae together comprise the previously defined classical chaperone/usher family. The low bootstrapping values of nodes connecting members of the classical chaperone/usher family to the α-fimbriae or the σ-fimbriae made it impossible to infer whether the classical chaperone/usher family represents a monophyletic group (Fig. 2). However, a monophyletic origin of this group is supported by the finding that the chaperones encoded by operons of the classical chaperone/usher family all share a protein domain (PFAM02753) that is absent from chaperones of the α-fimbriae and the σ-fimbriae (Fig. 10).

FIG. 10.

Evolution of fimbrial operons of the chaperone/usher assembly class from a hypothetical ancestor. The predicted functions for each gene product and the sequence homologies to protein families are indicated in the legend at the bottom. For a detailed discussion, see the text.

The α-fimbrial operons are composed of genes encoding a chaperone, a major subunit, an usher, and a tip adhesin. The α-fimbriae are the only group in which the first two genes in the operon comprise a chaperone followed by a subunit, and this gene order is not found in the most divergent members of this group (e.g., the csf operon, encoding CS5 fimbriae of E. coli) (Fig. 3). Based on these observations, it is likely that the operon structure characteristic of most α-fimbriae evolved after this group had already diverged from the other lineages. In contrast, a common core structure can be found in at least some members of the remaining three groups (i.e., β-fimbriae, the γκπ cluster, and σ-fimbriae). The γκπ cluster and σ-fimbriae contain many operons in which a major subunit gene is followed by genes encoding a chaperone, an usher, and a tip adhesin. This operon structure likely gave rise to that found in β-fimbriae, in which a gene encoding a tip adhesin is absent (Fig. 10). These considerations suggest that a major subunit gene followed by genes encoding a chaperone, an usher, and a tip adhesin may represent the ancestral operon structure of the chaperone/usher assembly class. A divergence from an ancestral gene cluster was not apparent from previous comparative studies on fimbrial operons.

Only two representatives of the chaperone/usher assembly class (both members of the σ-fimbriae) are present in bacteria outside the phylum Proteobacteria, one in the phylum Cyanobacteria (Synechocystis sp.) and one in the phylum Deinococcus-Thermus (D. radiodurans). Thus, the chaperone/usher assembly class is largely restricted to the phylum Proteobacteria (367). Within this phylum, fimbrial operons belonging to the σ-fimbrial clade are distributed most widely, with representatives being present in the Alpha-, Beta-, Gamma-, and Deltaproteobacteria. In contrast, representatives of the α-, β-, γ-, κ-, and π-fimbriae are restricted to two classes, the Beta- and Gammaproteobacteria. This distribution suggests that σ-fimbriae may be phylogenetically older than the other fimbrial clades, thus making COG5430 a likely candidate for a protein domain present in fimbrial subunits of a hypothetical ancestor of the chaperone/usher assembly class (Fig. 10).

Fimbrial operons contain conserved subunit protein domains characteristic of either the α-fimbriae (PFAM04449), the β-fimbriae (PFAM06551), the γκπ cluster (PFAM00419), or the σ-fimbriae (COG5430) but never a mixture thereof, suggesting that the diversity in operon structures found within each clade never involved lateral exchange of subunit genes among these four major groups. Major fimbrial subunits and the C-terminal domains of tip adhesins both have an incomplete Ig-like fold containing a cleft in which a donor β-strand is inserted to incorporate the subunit into a fimbrial filament (Fig. 1). It is thus not surprising that the conserved protein domain, which is present in the major fimbrial subunit, is also commonly present in the C-terminal domain of the corresponding tip adhesin (Fig. 10). The N-terminal domain of tip adhesins contains the receptor-binding site and also has an Ig-like fold (47). With respect to the α-, β-, γ-, κ-, and π-fimbriae, the N-terminal domain of tip adhesins never contains the conserved protein domain present in its C terminus. Interestingly, some operons belonging to the σ-fimbriae, the most archaic group within the chaperone/usher assembly class, encode tip adhesins that contain COG5430 domains in both the C-terminal and the N-terminal domains of their tip adhesin (Fig. 9). This observation suggests that tip adhesins arose from a gene fusion event involving two fimbrial subunit genes.

A phylogenetic tree constructed by comparing chaperone proteins is shown in Fig. 11. The phylogenetic analysis of chaperone proteins suggests a division into six groups that are in most cases identical to the α-, β-, γ-, κ-, π-, and σ-fimbriae. The only exception is the classification of CsfB and CsfF (the chaperones of CS5 fimbriae) as a sister group to the α- and σ-fimbriae (Fig. 11). In the FUP classification, CsfC (the usher of CS5 fimbriae) is a member of the α-fimbriae (Fig. 3). Overall, an analysis of chaperone proteins provides compelling support for the proposed FUP nomenclature. Although sequence comparisons of chaperone proteins yielded similar results, an analysis of the FUP family is better suited for deriving a classification scheme, because operons encoding more than one chaperone complicate the analysis. A closer examination of fimbrial operons that contain more than one chaperone shows that in most cases, chaperones encoded by the same operon form sister groups on the chaperone tree (Fig. 11). These data suggest that the evolution of the respective fimbrial operons involved a duplication of their chaperone gene. One notable exception is a subgroup of γ₂-fimbrial operons, including csw, ECP_3811 to ECP_3817, fas, and fot (Fig. 3). These operons encode three chaperones each, only one of which groups with other chaperones of the γ₂-fimbriae. The remaining two chaperones form a distinct branch on the chaperone tree that does not belong to any group defined in the FUP tree (Fig. 11). A possible interpretation of this result is that the respective γ₂ operons evolved by acquiring (and subsequently duplicating) a chaperone gene from a donor that does not resemble any of the currently sequenced members of the chaperone/usher assembly class. This represents the only example detected in our analysis in which an operon structure may have been altered by horizontal gene transfer.

CONCLUSIONS

Assembly proteins (i.e., chaperones and ushers) show better sequence conservation than structural subunits and thus are superior for generating trees that illustrate the phylogenetic relationships between divergent members of the chaperone/usher assembly class. This and a previous study (367) defined fimbrial clades by nodes on an FUP family tree (Fig. 2) because all chaperone/usher systems contain only a single usher gene, while multiple genes encoding chaperones may be present. An FUP-based definition of the α-, β-, γ-, κ-, π-, and σ-fimbrial clades is supported by (i) similarities in operon structures within each group (Fig. 3 to 9), (ii) the presence of distinct conserved protein domains in major subunits of different clades (Fig. 10), (iii) the definition of almost identical fimbrial clades using a chaperone tree (Fig. 11), and (iv) the formation of clusters of operons encoding similar surface structures.

The construction of FUP trees similar to the one shown in Fig. 2 is not a rapid and convenient method for classifying new fimbrial operons identified by whole-genome sequencing. However, our analysis suggests that results from a simple BLAST search may be sufficient for the placement of a putative fimbrial operon in our classification scheme (Table 1). This method may not always succeed in distinguishing between members of clades within the γ-fimbriae, because some groups (particularly γ₁- and γ₄-fimbriae) share similar characteristics. However, Table 1 illustrates that information on gene order and conserved protein domains found in fimbrial subunits will be sufficient for assigning a new operon to either the α-, β-, γ-, κ-, π-, or σ-fimbrial clades. In conclusion, the classification scheme proposed in this article (Table 1) delineates for the first time a simple and convenient method for assigning newly discovered chaperone/usher systems to one of six major phylogenetic clades (Fig. 2).

TABLE 1.

FUP-based classification scheme for fimbrial operons

Major division	Clade	Subclade	Fig.	Alternate designation	Domain(s) in subunit	Domain(s) in chaperone	Core operon structurea	Other feature(s)
Alternate chaperone/usher family	α		3	Class 5 fimbriae, FUP cluster 8	PFAM04449, PFAM07434	COG3121 or no domain	CMUT	None
Classical chaperone/	β		4a	FUP cluster 7	PFAM06551	COG3121, PFAM00345,	MCU	No tip adhesin
usher family	γ	γ3	5	FGL chaperone subfamily, Class 2 fimbriae, FUP clusters 4 and 5	PFAM04619, PFAM06443, or PFAM05775	and PFAM02753	(M)CU(M)	No tip adhesin
		γ1	5	FUP cluster 1	PFAM00419, COG3539		MCUT	None
		γ2	5	FUP cluster 2			MCCU	Three chaperones
		γ4	6	FUP cluster 10			MCUT	None
	π		8	FUP cluster 3			MUCT	None
	κ		7	FUP cluster 6	PFAM00419, COG3539, or PFAM02432		MUC	No tip adhesin
Archaic chaperone/usher systems	σ		9	FUP cluster 9	COG5430	PFAM00345 and/or COG3121	MCUT	None

^aM, major subunit; C, chaperone; U, usher; T, tip adhesin.