Pili Assembled by the Chaperone/Usher Pathway in Escherichia coli and Salmonella

Abstract

Gram-negative bacteria assemble a variety of surface structures, including the hair-like organelles known as pili or fimbriae. Pili typically function in adhesion and mediate interactions with various surfaces, with other bacteria, and with other types of cells such as host cells. The chaperone/usher (CU) pathway assembles a widespread class of adhesive and virulence-associated pili. Pilus biogenesis by the CU pathway requires a dedicated periplasmic chaperone and integral outer membrane protein termed the usher, which forms a multifunctional assembly and secretion platform. This review addresses the molecular and biochemical aspects of the CU pathway in detail, focusing on the type 1 and P pili expressed by uropathogenic Escherichia coli as model systems. We provide an overview of representative CU pili expressed by E. coli and Salmonella, and conclude with a discussion of potential approaches to develop antivirulence therapeutics that interfere with pilus assembly or function.

OVERVIEW AND CLASSIFICATION OF PILI

Bacteria assemble a wide variety of proteinaceous structures on their surfaces, with functions including motility, adhesion, and the secretion of proteins and other molecules (1–3). Bacterial surface structures distinct from the motility organelles, flagella (Latin for whip), were first described in the late 1940s and early 1950s, with the advent of the electron microscope (4). The term fimbriae (Latin for thread or fiber) was coined by Duguid and coworkers to describe the surface structures necessary to bind and agglutinate erythrocytes (5). Soon thereafter, Brinton used the term pili (Latin for hair) to describe the nonflagellar surface structures expressed by Escherichia coli (6). In 1975, Ottow suggested that “pili” be reserved for those structures involved in bacterial mating and “fimbriae” for structures involved in adhesion (4). However, his recommendation did not stick and the terms remain used interchangeably. Here, we will generally use the term pili.

Pili are hair-like organelles that decorate the bacterial surface. Pili are typically involved in adhesion and function in a range of interactions between bacteria, bacteria and other cells, and bacteria and their surrounding environment. These functions include the formation of microcolonies and biofilms, colonization of surfaces, and receptor-mediated adhesion to host cells (1). Some types of pili also function in motility and the uptake of DNA or phage (7). By acting outside a bacterium’s capsule or other protective surface structure, pili may increase the functional reach of bacteria and confer adhesive functions while preserving the barrier properties of the cellular envelope. The ability of pili to act distantly from the cell surface also may facilitate bacterial evasion of immune surveillance and detection or uptake by host cells.

Pilus Classification Schemes

Pilus classification schemes have changed over the years. In 1965, Brinton distinguished six types of pili in E. coli (8). The following year, Duguid and colleagues proposed a classification scheme based on pilus morphology and hemagglutination potential (9). This scheme comprised seven pilus types (types 1 through 6 and F). In subsequent schemes, pili were classified based on their abilities to agglutinate human red blood cells of different blood groups in the presence or absence of mannosides. For example, P pili of uropathogenic E. coli (UPEC) bind the disaccharide Galα (1–4)Gal linkage on erythrocytes of the P blood group system and are mannose-resistant (10, 11), whereas Dr pili (also mannose-resistant) bind CD55 on DR blood group erythrocytes (12, 13). This classification scheme led to the term type 1 pili, which remains in current use, to refer to mannose-sensitive bacterial surface fibers. However, genetic analyses revealed that hemagglutination-based classification schema are arbitrary, because they may assign pili encoded by homologous genes into different groups, and pili encoded by distinct systems into the same group (14–17). Additional classification systems based on serology have emerged (18). Such schemes have been particularly useful to classify the variety of pilus antigens expressed by E. coli. However, because of the high specificity of serological classification, this method is not useful in assigning individual pili to larger groups of related surface structures. In his 1975 landmark review, Ottow proposed an updated classification scheme that distinguished six groups of pili (4). For example, group 4 included polarly inserted pili involved in bacterial locomotion. Only parts of the systems prescribed by Duguid and Ottow have endured over the years. For instance, type 1 pili, the classification designated by Duguid, remains the common descriptor for pili that mediate mannose-sensitive hemagglutination. Ottow’s “group 4” designation has evolved to “type IV” pili, the designation given to pili that mediate twitching motility and are capable of extending and retracting (7). Additional classification has been based on the pilus architecture (see “Architecture of Chaperone/Usher Pili” below) (19–21). For example, pili may be classified as either homopolymeric or heteropolymeric. Heteropolymeric pili (e.g., type 1 and P pili) are composed of a number of different pilus subunits of different stoichiometry. The architecture of heteropolymeric pili generally consists of a relatively rigid pilus rod, with a flexible distal tip. The adhesin subunit in heteropolymeric pili is present in a single copy at the most distal location of the tip. Thus, heteropolymeric pili are “monoadhesive.” Homopolymeric pili (e.g., Dr and AFA-III pili) are composed of a multiply incorporated single subunit and are generally less structured than heteropolymeric pili. Homopolymeric pili are thin and resemble the distal tip of the heteropolymeric pilus. The homopolymeric pilus subunit also functions as an adhesin, and thus these pili are “polyadhesive.”

Modern classification schemes based on the biochemistry and genetics of pilus assembly systems have proven more useful and durable for assigning pili into distinct classes. These schemes are based on the pilus assembly mechanism and group pili according to the sequence homology of the genes encoding the machinery required for pilus biogenesis. For Gram-negative bacteria, such as E. coli and Salmonella, this results in the classification of pili into at least four different groups: chaperone/usher (CU) pili, curli, type IV pili, and conjugative F pili (22–26). Pili assembled by the CU pathway are the focus of this review.

Pili Assembled by a Chaperone- and Usher-Dependent Mechanism

The CU pathway serves to assemble and secrete a superfamily of adhesive and virulence-associated surface structures in Gram-negative bacteria (27, 28). Pili are polymeric fibers assembled from multiple subunit proteins. The assembly of pili by the CU pathway involves the binding of nascent pilus subunits by a dedicated chaperone in the bacterial periplasm, and the subsequent polymerization of subunits into the pilus fiber at the outer membrane (OM) by an integral OM channel protein termed the usher. Genetic loci coding for CU pili are located both chromosomally and on plasmids, and a given bacterial genome may contain multiple different CU loci. A systematic effort by Nuccio and Bäumler (29) categorized all CU pathways into phylogenetic clades on the basis of usher gene sequence, yielding six clades: α, β, γ (subdivided into γ₁, γ₂, γ₃, and γ₄), κ, π, and σ. For the α, κ, π, and σ clades, the clade designations were assigned to reflect a particular quality of the clade or a prominent member as follows: α-pili, alternate CU family; κ-pili, K88 (F4) pili; π-pili, pyelonephritis-associated pili (P pili); and σ-pili, spore coat protein U from Myxococcus xanthus.

In addition to the six phylogenic clades based on usher gene sequence (29), the CU superfamily has been subdivided by using other criteria. The α-phylogenetic clade corresponds to a subfamily of CU pili previously termed the “alternate” CU family. The “alternate” designation was initially made based on the fact that the gene sequences for these pili possess little homology to constituents of “classical” CU pili such as the type 1 and P pili of UPEC. However, subsequent studies revealed that the alternate and classical CU pili are in fact related and share similar assembly mechanisms (29–31). The alternate CU family comprises pili expressed by enterotoxigenic E. coli (ETEC) and Salmonella enterica serotype Typhi, including colonization factor antigen I (CFA/I) and Typhi colonization factor (Tcf) pili (31–36). The alternate CU family is sometimes referred to as “class 5” pili, from a classification scheme based on sequence analysis of the pilus subunit proteins (37). An additional division of the CU superfamily into two subfamilies has been made based on conserved sequence differences in a region of the pilus chaperones. These differences relate to the length of the loop connecting the chaperone’s F1 and G1 β-strands (see “Mechanism of Pilus Assembly by the Chaperone/Usher Pathway” below). Those systems whose chaperones contain a short loop belong to the F1-G1-short (FGS) subfamily; those whose chaperones contain large loops are categorized in the F1-G1-long (FGL) subfamily (38–40). The FGS subfamily assembles a range of pilus structures, including rigid, helical rods with distinct tip fibers such as the type 1 and P pili. The FGL subfamily assembles thin, fibrillar, or nonfibrillar surface structures (38, 39). The FGL CU subfamily forms the γ₃-phylogenetic clade in the classification described by Nuccio and Bäumler, and has also been termed “class 2” pili based on sequence analysis of the subunit proteins (29, 37).

CU pili are prevalent among the Enterobacteriaceae, and most of the known types of pili expressed by E. coli and Salmonella belong to the CU family. The CU pathway assembles a variety of surface structures, ranging from rigid, rod-like pili to amorphous (afimbrial) and capsular-like structures. This architectural diversity is reflected by a broad functional diversity among CU pili. The pili assembled by E. coli and Salmonella are representative of the structural and functional diversity encoded by CU pathways (Fig. 1, Table 1). Two pili in particular have served as model systems for the study of pilus assembly by the CU pathway: the adhesive type 1 and P pili expressed by UPEC (41, 42). UPEC is the primary causative agent of urinary tract infections (UTI), one of the most common infections worldwide, affecting more than 50% of women during their lifetime (43–45). Type 1 and P pili confer UPEC with the ability to adhere to the bladder and kidney, respectively, withstand turbulent urine flow, and colonize the urinary tract. The type 1 and P pili expressed by UPEC are two of the best-characterized pilus assembly systems, and this review will focus on these pili as models.

Figure 1.

Model for pilus biogenesis by the CU pathway. Pilus subunits translocate from the cytoplasm to the periplasm as unfolded polypeptides via the Sec system. Subunit folding occurs upon interaction with the pilus chaperone (yellow) in the periplasm. Chaperone-subunit complexes then interact with the OM usher for exchange of chaperone-subunit for subunit-subunit interactions, ordered assembly of the pilus fiber, and secretion through the usher channel to the cell surface. The usher is depicted with its β-barrel channel domain in the OM and its plug, N, C1, and C2 domains labeled. The N domain forms the initial binding site for chaperone-subunit complexes, and the C domains provide a second binding site for the assembling pilus fiber. Chaperone-adhesin complexes have the highest affinity for the usher and initiate pilus assembly by binding to the usher N domain, with subsequent handoff to the usher C domains. Repeated rounds of chaperone-subunit targeting to the usher N domain and subunit-subunit interaction then lead to assembly and secretion of the pilus fiber in a top-down manner. Models of fully assembled type 1 (Fim), P (Pap), and Afa pilus fibers are shown.

Table 1.

CU pili present in E. coli and Salmonella

Pilus	Adhesin	Chaperone/usher	Gene cluster	Associated disease or function	Receptor	Refs
Escherichia coli
P	PapG	PapD/PapC	papAHCDJKEFG	Pyelonephritis or cystitis	Galα1,4 Galβ, GbO3, GbO4, GbO5, kidney epithelium, erythrocytes	41
Type 1	FimH	FimC/FimD	fimAICDFGH	Cystitis, biofilm formation	Uroplakin UP1a, β1α3 integrins, laminin, CD48, collagen (type I and IV), bladder and kidney epithelium, buccal cells, erythrocytes, mast cells, neutrophils, macrophages	42
Afa/Dr	AfaE	Afa(Dra)B/Afa(Dra)C	afa(dra)BCDPE	Pyelonephritis	Dr blood group, α5β1 integrin, CD55/DAF, CEACAMS, uroepithelium, erythrocytes	90, 147, 343
AAF-I	AggA	AggD/AggC	aggDCBA	Diarrhea	Hep-2 cells	344
AAF-II	AafA	AafD/AafC	aafBCDA	Diarrhea	Intestinal mucosa	223
AAF-III	Agg-3A	Agg-3D/Agg-3C	agg-3DCBA	Diarrhea	Intestinal mucosa	215
AFA-III	Afa3-E	Afa-3B/Afa-3C	afa-3BCDE	Diarrhea, cystitis	Dr blood group, α5β1integrin, CD55/DAF, CEACAMs, uroepithelium, erythrocytes	345
Afa-8	Afa-8E	Afa-8B/Afa-8C	afa-8BCDE	Diarrhea/septicemia	DAF	346
AF/R1	AfrE	AfrC/AfrB	afrABCDE	Diarrhea in rabbits	Rabbit sucrose-isomaltase protein complex	347
Auf	AufG	AufB,F/AufC	aufABCDEFG	N.D.,a auf identified in a UPEC strain	N.D.	348
Csh	N.D.	CshC/CshB	cshABCDEFG	N.D.	N.D.	349
CS1, CFA/I	CooD	CooB/CooC	cooBACD	Diarrhea	Intestinal epithelium, bovine erythrocytes	283
CS2, CFA/II	CotD	CotB/CotC	cotBACD	Diarrhea	Intestinal epithelium, bovine erythrocytes	129
CS3	CS3-3	CS3-1/CS3-2	CS3-1,-2,-3	Diarrhea	Adhesion to intestinal mucosa	350
CS4	CsaE	CsaA/CsaC	csaABCE	Diarrhea	Intestinal epithelium	32
CS5	N.D.	CsfB,F/CsfC	csfABCEFD	Diarrhea	Intestinal epithelium	351
CS6	(probably CssB)	CssC/CssD	cssABCD	Diarrhea	sulfatide (SO3-3Galβ1Cer)	352
CS12	CswG	CswB,C,E/CswD	cswABCDEFG	Diarrhea	Human intestinal epithelium	353
CS14	CsuD	CsuB/CsuC	csuBA1A2CD	Diarrhea	Intestinal epithelium	354
CS17	CsbD	CsbB/CsbC	csbBACD	Diarrhea	GlcNAc β1,2 Man, Epithelial intestinal cells	354
CS18	FotG	FotB/FotD	fotABCDEFG	Diarrhea	Intestinal epithelium	355
CS19	CsdD	CsdB/CsdC	csdBACD	Diarrhea	Intestinal epithelium	356
CS20	N.D.	N.D.	N.D.	N.D.; associated with an Indian ETEC strain	Human intestinal epithelium	357
CS31	ClpG	ClpE/ClpD	clpBCDEFGHI	Diarrhea	Intestinal epithelium	358
DAF	DafaE	DafaB/DafaC	dafaABCDE	Diarrhea	N.D.	359
ECP (Yag)	N.D.	YagV/YagX	yagZYXWV	N.D.	HEp-2 and HeLa epithelial cells	360
Fst	FstE	N.D.	N.D.	N.D.	Fibronectin, rat basolateral membranes of kidney tubules	361, 362
F1C	FocH	FocC/FocD	focAICDFGH	Cystitis, pyelonephritis	GalNAc β 1,4 Galβ (bladder endothelium; distal tubules, collecting ducts, glomeruli, endothelium of kidney)	363
F7 (1)	FsoE	N.D.	N.D.	N.D.	Fibronectin, rat basolateral membranes of kidney tubules	361, 362
F9 (Yde)	YdeQ	ND/YdeT	ydeTSRQ	N.D., biofilm formation	Bovine epithelial cells, mouse model of chronic cystitis	364, 365
F17	F17G	F17D/F17C	F17aADCG	Diarrhea	N-Acetylglucosamine, intestinal villi	366
F107 (F18ab)	FedF	FedC/FedB	fedABCEF	Edema and diarrhea in pigs	Blood group H type 1 determinant; pig intestinal epithelium	367
F1651	N.D.	N.D.	N.D.	N.D.	N.D. (found in a porcine/bovine pathogenic E. coli strain 4787	368
F1845	Probably DaaE	DaaB/DaaC	daaFABCDPE	Diarrhea	Decay-accelerating factor (DAF)	231
Hda	HdaA	HdaD/HdaC	hdaDCBA	N.D.	Aggregative adhesion to Hep-2 cells	216
K88 (F4)	FanH	FanE/FanD	fanCDEFGH	Diarrhea in neonatal piglets	NeuGc α2,3 Gal β1,4 GlcNAc	369
K99 (F5)	FaeH	FaeE/FaeD	faeCDEFGHIJ	Diarrhea in neonatal calves, lambs, piglets	NeuGc α2,3 Gal β1,4 GlcNAc	369
Lda	LdaH	LdaE/LdaD	ldaCDEFGHI	Diarrhea	Diffuse adhesion on Hep-2 cells	370
Lpf	LpfD	LpfB/LpfC	lpfABCD	Associated with diarrheagenic strains	Intestinal epithelium	174
Nfa	NfaA	NfaE/NfaC	nfaEDCBA	Urinary tract infections, newborn meningitis	Neutrophils	90, 371
PCF071	CosD	CosB/CosC	cosBACD	Diarrhea	Intestinal epithelium	372
Pix	PixG	PixD,J/PixC	pixAHCDJFG	N.D., associated with pyelonephritis isolate	N.D.	47
PRF	PrfG	PapD,J/PrfC	prfAHCDJKEFG	N.D.	N.D.	373
REPEC	RalG	RalE/RalD	ralCDEFGHI	Diarrhea in rabbits	Rabbit intestinal epithelium	374
S	SfaH	SfaE/SfaF	sfaADEFGSH	Urinary tract infections, newborn meningitis	NeuAc α 2,3 Galβ sialic acid residues, plasminogen, bladder and kidney epithelial cells, erythrocytes, endothelial cells	263
Sfm	SfmH	SfmC/SfmD	sfmACDHF	N.D.	Bladder epithelium	48
Sfp	SfpG	SfpD,J/ SfpC	sfpAHCDJFG	Diarrhea	N.D.	167
Stg	StgD	StgB/StgC	stgABCD	Respiratory tract infection in birds	Avian respiratory tissues	375
Yad	YadC	EcpD/HtrE	yadNecpDhtrEyadMLKC	UTI in mouse model	Abiotic surfaces, bladder epithelium	376
Ybg	N.D.	YbgP/YbgQ	ybgDQPO	N.D.	N.D.	48
Ycb	YcbT	YcbR,F/YcbS	ycbQRSTUVF	N.D.	Abiotic surfaces	(48)
Yeh	YehA	YehC/YehB	yehDCBA	UTI in mouse model	Abiotic surfaces, embryonic kidney cells	48, 376
Yfc	N.D.	YfcS/YfcT,U	yfcVUTSRQPO	N.D.	N.D.	48
Ygi	YgiL	YgiH/YgiG	ygiLyqiGHI	UTI	Human kidney epithelium, abiotic surfaces	48
Yhc	N.D.	YhcA/YhcD	gltFyhcADEF	N.D.	N.D.	47
Yqi	N.D.	YqiH/YqiG	yqiLGHI	N.D.	N.D.	29
Yra	YraK	YraI/YraJ	yraHIJK	N.D.	Bladder epithelium	48
987P	FasG	FasB,C,E/FasD	fasABCDEFG	Diarrhea in neonatal pigs	Brush border vesicle-derived Histone H1 proteins of neonatal piglet intestinal epithelium	377, 378
Salmonella spp.
Bcf	BcfD	BcfB/BcfC	bcfABCDEFGH	Intestinal persistence in mice	N.D.	49, 292
Fae	FaeG	FaeE/FaeD	faeBCDEFGHIJKA	N.D.	N.D.	379
Lpf	LpfD	LpfB/LpfC	lpfABCDE	Intestinal persistence in mice, typhoid in mouse model, gastroenteritis, biofilm formation	Peyer’s patches in murine model	290, 292
Mrk	MrkD	MrkB/MrkC	mrkABCDFJI	N.D.	N.D.	379
Pef	PefD	PefC/PefA	pefBACDEFI	Gastroenteritis, biofilm formation	Lewis X blood group antigen	299
Peg	PegD	PegB/PegC	pegABCDE	Intestinal and oviduct colonization in chickens; egg contamination; systemic infection in chickens and mice	N.D.	380, 381
Peh	PehD	PehB/PehC	pehABCDE	N.D.	N.D.	379
Saf	SafD	SafB/SafC	safABCD	Gastroenteritis, enteric typhoid fever	N.D.	382
Sba	SbaH	SbaC/SbaD	sbaABCDEFGH	N.D.	N.D.	379, 383
Sbc	SbcF	SbcD/SbcC	sbcABCDEFGH	N.D.	N.D.	379
Sbb	SbbD	SbbC/SbbB	sbbABCD	N.D.	N.D.	379, 384
Sdk	SdkD	SdkC/SdkB	sdkABCD	N.D.	N.D.	379
Stj	StjA	StjC/StjB	stdABCD	Intestinal persistence in mice	N.D.	385
Sdc	SdcE	SdcC/SdcD	sdcABCDE	N.D.	N.D.	379
Sdd	SddD	SddB/SddC	sddABCDEF	N.D.	N.D.	379
Sde	SdeD	SdeB/SdeC	sdeABCDE	N.D.	N.D.	379
Sdf	SdfD	SdfB/SdfC	sdfABCDEF	N.D.	N.D.	379
Sdg	SdgG	SdgB/SdgC	sdgABCDEFG	N.D.	N.D.	379
Sdh	SdhG	SdhB/SdhC	sdhABCDEFG	N.D.	N.D.	379
Sdi	SdiD	SdiB/SdiC	stiABCD	N.D.	N.D.	379, 384
Sdj	SdjD	SdjB/SdjC	sdjABCD_yhoN	N.D.	N.D.	379
Sef14/18	SefD	SefB/SefC	sefABCDR	Enteric typhoid fever	N.D.	386
Sta	StaG	StaB/StaC	staABCDEFG	N.D.	N.D.	49
Stb	StbD	StbB/StbC	stbABCDEF	Intestinal persistence in mice	N.D.	49, 292
Stc	Stc	StcB/StcC	stcABCD_yhoN	Intestinal persistence in mice	N.D.	49, 292, 300
Std	Std	StdA/StdB	stdABCD	Intestinal persistence in mice	α(1,2)-fucose	49, 292
Ste	SteG	SteB/SteC	steABCDEFGHIJ	N.D.	N.D.	49
Stf	StfH	StfC/StfB	stfABCDEFGH	Systemic and fatal infection in inbred mice	N.D.	289, 387
Stg	StgD	StgB/StgC	stgABCD	N.D.	Association, invasion, and permeabilization of HEp-2 epithelial cells and macrophage-like cells	49, 388
Sth	SthE	SthB/SthC	sthABCDE	Intestinal persistence in mice	N.D.	49, 292
Sti	StiH	StiB/StiC	stiABCH	N.D.	N.D.	17, 379
Stk	StkG	StkB/StkC	stkABCDEFG	N.D.	N.D.	51
Tcf	TcfD	TcfA/TcfC	tcfABCD	N.D.	N.D.	(49)
Type 1	FimH	FimD/FimC	fimAICDHFZYW	Intestinal persistence in mice and pigs, biofilm formation	Mannose, adherence to laminin via mannose, horse and chicken erythrocytes, enterocytes, glycoprotein-2	289, 389

^a N.D., not determined.

CHAPERONE/USHER GENETIC LOCI AND REGULATION

Genes encoding CU pili are found on both the bacterial chromosome and plasmids. The pilus genes are typically clustered together with a similar organization: an upstream regulatory region followed by a single downstream operon that encodes the pilus structural proteins, adhesive component(s), and assembly machinery (Fig. 2). CU pilus gene clusters are often located in proximity to other genes encoding virulence factors, in regions termed pathogenicity islands that have signatures suggesting acquisition by horizontal gene transfer (46). A genomic analysis found that E. coli strains encode as many as 17 CU gene clusters and that UPEC strains encode 9 to 12 intact CU pilus operons (47). Many of these CU loci are cryptic and do not express pili under laboratory growth conditions, and thus the functions of their gene products remain elusive (48). However, evidence suggests that many, if not most, of these cryptic operons encode products that are functional for adhesion and may contribute to ability of E. coli to colonize a wide variety of environmental niches (48). Salmonella strains similarly encode multiple pilus gene clusters, many of which are also not expressed under laboratory growth conditions (17, 49, 50).

Figure 2.

CU pilus gene clusters and regulatory regions. (A) Gene clusters, including upstream regulatory regions, are shown for P (pap), type 1 (fim), and Dr/Afa pili, with the functions of the genes indicated. (B) Regulatory region of the pap gene cluster, shown in phase-OFF and phase-ON states. Lrp binding to the GATC_proximal site turns off expression from the papBA promoter. Binding of Lrp, together with PapI, to the GATC_distal site allows Dam methylation of GATC_proximal, resulting in phase-ON expression. Production of PapB during phase-ON expression initiates a positive feedback loop through upregulation of PapI. (C) Regulatory region of the fim gene cluster, showing the phase-OFF and phase-ON orientations of the fimS switch region. The left and right inverted repeat sequences (IRL and IRR) that flank the fimS switch are indicated. Binding of H-NS maintains fimS in the phase-OFF position, whereas binding of IHF and Lrp favors phase-ON expression.

Pilus gene sequences are among the most polymorphic regions in the genomes of E. coli and Salmonella. Comparisons of pilus repertoires of S. enterica serotypes Typhimurium (LT2), Typhi (CT18 and Ty2), and Paratyphi A (SARB64) illustrate that pilus gene sequences were lost or acquired frequently during the divergence of their lineages. The strictly human adapted serotypes Typhi and Paratyphi A carry stop codons or frameshift mutations in a number of their pilus gene clusters, which may reflect their host restriction or the decreased importance of intestinal persistence during systemic infections, such as typhoid fever (49, 51, 52). Alternatively, the loss of these pili may reflect a negative impact of expression of these structures during infection of the host. In E. coli, the tip adhesin subunits of pili are under strong evolutionary pressure, selecting for functionally adaptive amino acid replacements (53–55). This is consistent with the importance of adhesin-receptor interactions in determining bacterial tropism and the ability to colonize different niches.

The expression of CU pili is highly regulated, subject to ON/OFF phase variation and responsive to environmental cues (56–59). Regulatory cross talk occurs among the different CU gene clusters present within a given bacterium (60–63). This cross talk among pilus gene clusters likely serves as a way to limit unnecessary expression of surface structures and ensure expression of only one adhesin at time. Such regulation may be particularly important in the context of infection, to prevent recognition of bacterial surface structures by the host immune system and to control bacterial tropism to specific colonization sites. Furthermore, expression of adhesive pili is inversely correlated with the expression of flagella for motility (64). This allows bacteria to stick (colonize a specific site) or swim (turn off adhesion and move to new site). Here, we provide descriptions of the regulation of type 1 and P pilus expression by UPEC, as examples of CU pilus regulatory mechanisms.

Type 1 Pilus Regulation

Type 1 pili are encoded by the fim chromosomal gene cluster (Fig. 2A). Phase variable expression of type 1 pili is dictated by the orientation of fimS, an invertible DNA segment containing the promoter for the fim operon (Fig. 2A). The FimB and FimE recombinases, part of the λ-integrase (tyrosine-dependent site-specific recombinase) family, directly influence fimS orientation. FimB and FimE act on two 9-base pair inverted repeat sequences (IRR and IRL) that flank the invertible 314-base pair fimS DNA central switch, influencing the orientation of the switch and thus phase variation of the fim operon (65, 66). The orientation in which fim transcription occurs is referred to as phase-ON, and the nontranscriptional orientation is phase-OFF. The FimB and FimE recombinases have opposing influences on the fimS switch. FimE primarily flips the switch from the phase-ON to phase-OFF orientation, whereas FimB promotes fimS switching to either orientation (67, 68). This process of DNA inversion presumably involves the formation of a FimB or FimE complex with the fimS sequence, generation of a Holliday junction, and the breaking, rejoining, and resolution of the DNA strands. Phase variation is reversible, and the rate of switching has been calculated at 10⁻³ to 10⁻⁴ per cell per generation (69, 70). As a result of phase variation, cultures of UPEC exhibit a mixture of type 1 piliated and nonpiliated cells. This may ensure that a subpopulation of bacteria is always present to adapt to the local environmental niche.

The fimB and fimE genes, located upstream of the fim operon, are controlled by independent promoters (Fig. 2A). Because of their direct influence on fim phase variation, the regulation of both recombinases is key for type 1 pilus expression. For example, biofilm formation is influenced by the iron-sulfur cluster regulator (IscR) through modulation of type 1 pilus expression (71). Under iron-limiting conditions, IscR binds to the fimE promoter region, resulting in increased fimE expression and switching of fimS to the phase-OFF position. Conversely, when iron is available, fimE expression is downregulated, thereby positively influencing fim gene expression and pilus production. This model is consistent with the observation of decreased E. coli biofilm and pilus production under iron-limiting conditions (71). In contrast to IscR working through fimE, the global regulators ppGpp and DksA influence type 1 pilus production through transcriptional control of fimB (67, 72). As examples of the cross talk among CU pilus gene clusters, PapB, which regulates P pilus expression, also affects orientation of the fimS switch (60, 63, 73), and SfaB and SfaX, which are S pili regulatory proteins, negatively influence fim phase-ON switching (73, 74).

The fimS switch orientation is also influenced by other factors, including the bacterial nucleoproteins leucine-responsive regulatory protein (Lrp), integration host factor (IHF), and histone-like nucleoid-structuring (H-NS) protein (75). IHF and Lrp facilitate phase-ON switching, whereas H-NS maintains fimS in the phase-OFF position (Fig. 2C). The degree of DNA supercoiling, controlled by DNA gyrase activity, is postulated to be related to Lrp, IHF, and H-NS binding (75, 76). Changes in DNA gyrase activity and Lrp and IHF levels during stationary growth phase likely facilitate phase-ON switching of the fimS region, which is consistent with the early observation of robust pilus production by cultures maintained under static growth conditions (77). Type 1 pilus expression is also correlated with the osmolality and pH of the growth conditions. Increased osmolality and an acidic pH repress fim transcription (78). Such observations lead to a model consistent with infection of the human bladder and kidney; the bladder has a pH relatively higher and an osmolality relatively lower than the human kidney. Thus, type 1 pili, which facilitate colonization of the bladder, are likely upregulated in response to cues within the bladder niche and downregulated in bacteria that have ascended to the kidneys. In addition, a link between oxygen tension and type 1 pilus expression has been demonstrated in the context of bacterial biofilms grown in culture, with the fim promoter switched to phase-ON at the air-exposed surface of the biofilm, but switched OFF in deeper layers of the biofilm (79, 80).

P Pilus Regulation

P pili are encoded by the pap chromosomal gene cluster (Fig. 2A). P pili are regulated by a distinct mechanism compared to type 1 pili. Expression of the pap operon from the papBA promoter is regulated by the PapI and PapB proteins (81). Rather than an invertible switch element, P pilus expression is dictated by the methylation pattern of two GATC sites located upstream of the papBA promoter. Methylation at these sites, in conjunction with the action of PapI, influences binding of the global regulator Lrp (82). At high Lrp and low PapI concentrations, Lrp binds efficiently to the GATC site proximal to the papBA promoter (GATC_proximal) (Fig. 2B). Lrp binding to GATC_proximal results in inhibition of pap transcription (83, 84). This binding can only occur in the absence of GATC_proximal methylation, and therefore Lrp competes with the DNA adenine methylase (Dam) for binding (Fig. 2B). PapI binds to Lrp and is required for phase-ON expression (85). P pilus phase variation is dictated by Lrp concentration, which in turn is dependent on cell growth stage and conditions, and Lrp accessibility to the promoter. When cells reach early stationary phase, Lrp concentration increases, thereby promoting the transition from phase-ON to phase-OFF. At low Lrp concentrations, for example, soon after cell replication and DNA division, and in the presence of PapI, Lrp binds to the hemimethylated distal GATC site (GATC_distal), thereby preventing Dam from accessing this site for additional methylation, and thus encouraging Dam methylation of GATC_proximal, which in turn contributes to phase-ON variation (Fig. 2B). In addition, phase-ON expression results in PapB expression, which initiates a positive feedback loop through upregulation of PapI (86).

P pili, as with other CU pili, are subject to multiple layers of regulation. Consistent with the pathophysiology of the human urinary tract, the phase-OFF to phase-ON transition in P pili occurs in response to growth in urine with a high-amino-acid content (87). This regulation may allow pap genes to remain phase-OFF within the fecal flora, but switch to phase-ON upon exposure to the urinary environment. The CpxAR signal transduction system also modulates P pilus phase variation. CpxR competes for binding to both the proximal and distal papBA promoter sites by binding at positions adjacent to the GATC regions, inhibiting expression of the pap operon (88). This additional mode of control has been observed in response to an alkali environment and is not altered by methylation (88). UPEC also use small regulatory RNAs (sRNAs), which have short half-lives and thus allow rapid responses, to control P pilus expression. The sRNA repertoire of UTI89, a pathogenic UPEC strain, differs significantly for bacteria grown in laboratory culture versus the mouse bladder infection model (89). A novel sRNA, termed papR (P-fimbriae regulator), was shown to influence P pilus phase variation (89). The papR sRNA mediates posttranscriptional repression of PapI, and a papR deletion strain has increased bladder and kidney adhesive properties.

ARCHITECTURE OF CHAPERONE/USHER PILI

Structures assembled by the CU pathway range in shape from rigid, helical rods, which are typically thought of as classical pili, to thin, flexible fibers that, in some cases, form amorphous, capsular-like, “afimbrial” structures. The type 1 and P pili of UPEC are prototypical examples of classical pilus structures. Afa/Dr pili, expressed by various pathogenic E. coli strains (90, 91), are examples of thin, flexible pilus structures. The overall architectures of mature type 1 and P pili are shown in Fig. 1. Both pilus fibers consist of a rigid, helical rod with a diameter of ∼7 to 8 nm, forming the bulk of the 1- to 2-μm pilus length. The pilus rod is anchored in the bacterial OM and is composed of thousands of copies of the major subunit protein, FimA (type 1 pili) or PapA (P pili). The major subunit protein polymerizes into a linear fiber, which then further folds into a one-start, right-handed helix. The helical rod is terminated by the PapH subunit in P pili (Fig. 1), and the terminating subunit also has a role in anchoring the pilus fiber in the OM (92). The FimI protein might perform the equivalent function as a rod terminator for type 1 pili, but this remains to be confirmed (93, 94). The end of the pilus rod distal from the bacterial surface contains the adhesive tip portion of the pilus (Fig. 1). The pilus tip is an open linear fiber, with the most distal of the subunits being the adhesive subunit or adhesin. Type 1 pilus tip fibers contain the FimH adhesin, followed by the FimG and F adaptor subunits. P pilus tip fibers contain the PapG adhesin, followed by the PapF, E, and K subunits. The P pilus tip fiber is longer and more flexible than the type 1 pilus tip, because of the incorporation of 5 to 10 PapE subunits per P pilus tip. In contrast to the type 1 and P pili, Afa/Dr are thin (∼2 nm in diameter) flexible, linear fibers composed of a single-subunit protein, AfaE (DraE), which also functions as the adhesive subunit (95–99) (Fig. 1). A second pilus-associated protein, AfaD (DraD), is present in a single copy at the distal end of the AfaE pilus fiber and is thought to function as an invasin for uptake into host cells (99, 100). The DraD protein may exist in a secreted form separate from the pilus but located on the bacterial surface (100–102). Pili such as the type 1 and P pili that contain a single distal adhesin have been termed monoadhesive pili (21). In contrast to monoadhesive pili, polyadhesive pili such as Afa/Dr possess a major subunit protein that provides adhesive properties and receptor-binding sites along the exposed surface of the pilus. Thus, for polyadhesive pili, the entire pilus fiber may act in adhesion (20, 99, 103, 104).

Structures of Pilus Subunits and Adhesins

All pilus subunits belonging to the CU superfamily possess a common structural domain termed the pilin domain, which consists of an incomplete immunoglobulin (Ig)-like fold (Fig. 3). Canonical Ig folds contain seven β-strands arranged into two sheets to form a β-sandwich (105). However, the pilin domain lacks the seventh, C-terminal β-strand (the G strand) (106–109). Tip-located adhesin subunits contain a second, N-terminal adhesin or lectin domain in addition to the pilin domain (Fig. 3). The adhesin domain, which has the receptor-binding site, typically has a complete Ig-like fold termed a β-barrel jelly-roll fold. The adhesin domain likely evolved through duplication of the pilin domain, allowing specialization of the adhesin domain for receptor binding. Although sharing similar overall structures, adhesin domains from different CU pili vary significantly in sequence and display marked differences in their receptor-binding sites and receptor-binding mechanisms, reflective of their diverse functions and adaptation to specific niches (110).

Figure 3.

Structures of pilus chaperone and subunit proteins. (A) Structure of a FimC-H chaperone-adhesin complex (PDB ID: 1QUN) from the type 1 pilus system. The FimC chaperone is in yellow and the FimH adhesin in green, with the FimH lectin and pilin domains indicated. The chaperone is engaged in donor strand complementation (DSC) with the subunit. The chaperone donates its G1 β-strand (in blue) to complete the Ig fold of the FimH pilin domain. (B) Structure of a FimG-FimH subunit-subunit complex (PDB ID: 4J3O) from the type 1 pilus system. FimH is depicted as in (A) and FimG is in orange. The N-terminal extension (Nte) of FimG is engaged in donor strand exchange (DSE) with FimH. The Nte of FimG completes the Ig fold of the FimH pilin domain. (C) Topology diagrams depicting the Ig folds of the FimG (orange) and FimF (red) pilin domains. FimF is depicted in DSC with the donated G1-strand of the FimC chaperone, which is inserted parallel to the FimF F-strand. FimG is depicted in DSE with the Nte of FimF, which is inserted antiparallel to the FimG F-strand. (D) and (E) Structures of the lectin domains of the FimH (type 1 pili; PDB ID: 1KLF) and PapG (P pili; PDB ID: 1J8R) adhesins, with bound mannose and globoside molecules, respectively. The sugars are depicted in dark gray stick representation.

The type 1 and P pilus adhesins are illustrative of the variations in adhesin structure. Type 1 pili bind to mannosylated receptors via the tip-located FimH adhesin. The FimH adhesin domain has a typical β-barrel jelly-roll fold (Fig. 3D). The mannose-binding site is located at the tip of the adhesin domain, formed by a deep, negatively charged pocket surrounded by a hydrophobic ridge (106, 111). Because the type 1 pilus tip fiber is relatively short and inflexible, this positions the FimH-binding site at the most distal part of the mature pilus, in a location where it can readily bind host cells. P pili bind to Gal(α1-4)Gal receptors via the tip-located PapG adhesin. The PapG adhesin domain also has a typical β-barrel jelly-roll fold. However, in contrast to FimH, the PapG receptor-binding site lies in a shallow pocket on one side of the adhesin domain, formed by three β-strands and a connecting loop region (Fig. 3E) (112, 113). The P pilus tip fiber is longer and more flexible than the FimH tip fiber (Fig. 1). The side-on orientation of the PapG-binding site, in combination with the length and flexibility of the P pilus tip, facilitates interaction of PapG with the globoside moieties that are oriented parallel to the membrane surface (112). For polyadhesive pili such as Afa/Dr, the major subunit protein also functions as the adhesin, and consists of a single pilin domain. The high stability of the DraE subunit, as well as other pilus subunits, has been attributed to a disulfide bond in a noncanonical position in the Ig fold (114). The AfaE and DraE subunits possess distinct receptor-binding sites located on opposite sides of the pilin domain, which recognize CD55/decay accelerating factor (DAF) and members of the carcinoembryonic antigen family (CEACAM) (21, 99, 103). These binding sites are repetitively presented along the length of the assembled pilus fiber.

Mechanics of Pilus-Mediated Adhesion

Pili are exquisitely adapted to facilitate binding to specific receptors, under variable environments and conditions, with both the adhesin subunit and overall pilus architecture contributing to pilus function. The mechanics of the type 1 pilus adhesin FimH have been well studied. To mediate colonization in the urinary tract, type 1 pili must be able to withstand the shear forces generated by the flow of urine. Application of shear stress shifts FimH-mediated binding from a “loose” to a “firm” mode, with increased receptor affinity in the latter state (115). This FimH binding behavior is due to a catch-bond mechanism (116). Catch bonds are similar in concept to a Chinese finger trap toy, which grips tighter as it is stretched. The determining factor of “tight” versus “loose” binding is the connective region between the pilin and adhesin domains of FimH. FimH in its relaxed state (with an interdomain angle of 142°) has an open mannose binding pocket, which is poised for receptor binding (117, 118). Application of shear force causes FimH to become stretched (180° interdomain angle). This leads to a twisting of the β-sandwich fold of the lectin domain that imparts tighter FimH binding by causing the mannose pocket to clamp shut.

The catch-bond behavior allows type 1-piliated bacteria to “stick and roll” on the mannose-coated bladder epithelium, providing functions of both migration (rolling) and attachment (sticking) during times of both urinary stasis and turbulence, respectively (116, 119). Recent evidence has further elucidated roles for the low- and high-affinity states of FimH. In the context of high flow (for example, during micturition), the low-affinity state of FimH facilitates initial binding because of its capacity to form bonds rapidly. The shear stress then switches FimH into the high-affinity state, which is slower to release from its ligand. This nonequilibrium cycle that makes use of the rapid binding and slow unbinding of two kinetic conformational variants suggests a model wherein FimH possesses a higher effective affinity than it would if the active and inactive variants existed at equilibrium (120). The reduced affinity of FimH under low shear may also provide a mechanism for bacteria to sample receptors in the urinary tract, because shear stress will only be generated on FimH if the bacteria bind to an immobilized receptor such as on the bladder wall, allowing the bacteria to avoid latching onto soluble mannose moieties in the urine such as Tamm-Horsfall protein (121).

The kidney is a less turbulent environment compared with the bladder lumen, and P pili are adapted to function in this niche. Binding of the P pilus adhesin PapG to the globoseries of glycolipids on human kidney epithelial cells, via the Gal(α1-4)Gal linkage, facilitates attachment of UPEC during an ascending UTI and is one of the initiating steps of pyelonephritis. Rather than the catch-bond mechanism used by FimH, PapG uses what is termed a slip bond. The slip bond between PapG and the galabiose moiety is weaker than the FimH-mannose interaction. The PapG interaction has a shorter bond lifetime that decreases as force increases (122, 123). This binding mechanism is adapted for bacterial rolling and rapid spreading in the lower shear environment of the kidney.

In addition to the binding behavior of the pilus adhesin, the helical pilus rod is important for resistance to shear forces such as encountered during urine flow, through the rod’s structural properties of compliance and flexibility (124). Studies of pilus rod stretching and flexibility have been performed for both the type 1 and P pilus systems (125–127). The entire stretching process is reversible (126, 127). Presumably, this reversible nature allows bacteria to regain proximity to host cells after exposure to shear force, as in micturition. The P pilus rod exhibits remarkable reversible stretching properties (127). The PapG-receptor bond is weak enough for rolling but strong enough to initiate unstacking of the PapA helical rod, and thus the redistribution of force throughout the pilus (122). Recently, the molecular basis for this property of the PapA rod was determined via a high-resolution cryo-EM structure (128). The structure revealed that each PapA subunit in the helical rod makes contact with ten additional PapA subunits, five below and five above. These extensive subunit-subunit interfaces are mostly composed of polar interactions, which are weaker than the hydrophobic interactions that mediate subunit-subunit polymerization. Thus, the quaternary structure of the pilus rod may be linearized as these polar interactions are disrupted, but the strong subunit-subunit interactions remain intact, and the pili do not break. Upon reduction of shear stress, the pilus rod can then recoil to its helical conformation. Comparing the unwinding forces of different pili illustrates the adaptation of bacteria to binding in various niches. Type 1 and P pili of UPEC exhibit unwinding forces of 21 to 30 pN, and Klebsiella pneumoniae type 3 pili exhibit an unwinding force of >65 pN (129). In comparison, the CS, CFA/I, and CFA/II pili, which facilitate gut colonization by E. coli, exhibit lower unwinding forces of <15 pN (129). These values reflect the relative turbidities of the urinary tract, respiratory tract, and gastrointestinal tract.

MECHANISM OF PILUS ASSEMBLY BY THE CHAPERONE/USHER PATHWAY

All pili assembled by the CU pathway share a common mechanism of biogenesis, which depends on a dedicated periplasmic chaperone and an OM-localized assembly and secretion channel termed the usher. Pilus biogenesis at the bacterial OM occurs in the absence of input from an external energy source such as ATP, which is not available in the periplasm (130, 131). Instead, the CU pathway harnesses protein-protein interactions to drive fiber assembly and secretion in a defined order. The CU pathway is perhaps the best understood bacterial secretion system, and the UPEC type 1 and P pilus systems have served as central model systems for these molecular and structural studies.

Donor Strand Complementation

Individual pilus subunits are synthesized with an N-terminal signal sequence that directs them to the Sec general secretory pathway for translocation from the cytoplasm to the periplasm (Fig. 1) (132). The signal sequence is cleaved in the periplasm, and the subunits undergo disulfide bond formation in a process catalyzed by the periplasmic oxidoreductase DsbA (130, 133). Following disulfide bond formation, subunits are bound by a dedicated periplasmic chaperone (FimC for type 1 pili, PapD for P pili). Pilus chaperones are ∼25 kDa, boomerang-shaped proteins composed of two Ig-like domains (Fig. 3A). The chaperone specifically recognizes and binds unfolded, but disulfide-bridged (oxidized), pilus subunits, ensuring that the subunits have productively interacted with DsbA prior to chaperone engagement (134). Engagement with the pilus chaperone is required for subunit folding. In the absence of the chaperone, subunits misfold, aggregate, and are degraded by the DegP periplasmic protease (135–139). The chaperone forms stable, binary complexes with the subunits in the periplasm, preventing premature subunit-subunit interactions and maintaining the subunits in an assembly-competent state.

Atomic-resolution structures have been solved for many different chaperone-subunit complexes, permitting a detailed understanding of this key step in the pilus assembly pathway. For type 1 and P pili, structures have been solved for all chaperone-subunit complexes except FimC-G; the chaperone-subunit interface is conserved for each of these complexes (106–108, 134, 140–144). As discussed in “Architecture of Chaperone/Usher Pili” above, each pilus subunit possesses an incomplete Ig-like fold termed the pilin domain, which is characterized by a missing C-terminal β-strand (the seventh or G β-strand). The absence of this strand leads to an exposed hydrophobic groove on the surface of the subunit. The chaperone completes this hydrophobic groove of the pilin domain with its own G1 β-strand, simulating a complete Ig-like fold in the subunit (Fig. 3A and C). This in trans β-strand donation process is termed donor-strand complementation (DSC). Through the chaperone’s contribution of the missing structural information to the subunit, the chaperone provides stabilization and thus catalysis of subunit folding in the periplasm, increasing the folding rate by at least 4 orders of magnitude relative to chaperone-independent folding (134).

The subunit groove caused by the missing β-strand contains five binding pockets, termed P1 to P5. The donor G1 β-strand of the chaperone contains an alternating sequence of four bulky, hydrophobic amino acids, termed residues P1 to P4. In DSC, the chaperone inserts its P1 to P4 residues into the P1 to P4 pockets of the subunit’s hydrophobic groove (the P5 pocket of the subunit is left unoccupied). The donated G1 β-strand from the chaperone is inserted in a parallel orientation to the F strand of the pilin domain, thus completing the subunit’s Ig fold in a noncanonical manner (Fig. 3C). This parallel orientation, coupled with the bulky size of the P1 to P4 residues inserted by the chaperone, maintains the subunit in a stable, high-energy intermediate state, which keeps the subunit structurally competent for assembly into the pilus fiber (108, 109, 145).

The type 1 and P pilus chaperones (FimC and PapD) belong to the FGS subfamily, with a short loop connecting their F1 and G1 β-strands. In comparison with these chaperones, the longer F1-G1 loop of FGL chaperones allows the G1 donor strand to fill the P5 binding pocket of the bound subunit, but this interaction is weaker than at the other pockets (145, 146). Compared with the FGS chaperones, FGL chaperones assemble simpler, homopolymeric pilus fibers, comprising only one or two different subunits (such as the Afa/Dr pili). FGL chaperones characteristically possess two conserved cysteines (one in the F1 β-strand and one in the G1 β-strand) which form a disulfide bond that is indispensable for chaperone-subunit complex formation (139, 147). Correlative studies have compared the size of the F1-G1 loop with the number of subunits the chaperone is capable of binding; the longer the F1-G1 sequence, the more specifically the chaperone binds (20, 109). The FGS chaperones may have evolved to generate less stringent specificity, because pili such as the type 1 or P pili are composed of up to 7 different subunits, each of which must be bound by the chaperone in the periplasm.

Donor Strand Exchange

Following formation in the periplasm, chaperone-subunit complexes next target the OM-spanning usher (FimD for type 1 pili, PapC for P pili), where chaperone-subunit interactions are exchanged for subunit-subunit interactions in a mechanism termed donor strand exchange (DSE) (108, 109). The usher catalyzes the formation of subunit-subunit interactions, and the mature pilus grows through a channel provided by the usher, starting with the adhesin and followed by the rest of the tip fiber and finally the pilus rod (Fig. 1).

Each pilus subunit (except the adhesin) contains an N-terminal extension (Nte; Fig. 3), which, much like the chaperone G1 β-strand, contains a series of alternating hydrophobic residues. The subunit Nte ranges from 10 to 20 amino acids in length and is disordered when the subunit is in complex with its chaperone. In the DSE reaction, the Nte of an incoming subunit displaces the chaperone donor strand and occupies the P1- or P2-to-P5 pockets of the preceding subunit’s hydrophobic groove (Fig. 3). The exchange of chaperone-subunit for subunit-subunit interactions is initiated by the Nte’s P5 residue binding to the P5 pocket (the only pocket left unoccupied by the chaperone) of the preceding subunit and proceeding toward P1 (146). This mechanism of exchange has thus been termed “zip-in, zip-out.” The P4 residue of the Nte is a strictly conserved glycine, the only amino acid small enough to fit in the P4 pocket (a bulky aromatic residue juts into this groove). This strict glycine conservation ensures that the donated Nte is properly oriented and registered when binding its partner. In P pili, the PapH subunit acts to terminate assembly of the pilus rod and facilitates anchoring of the pilus to the OM (92, 142). The mechanism by which PapH acts as an efficient terminator is well understood. The hydrophobic groove of the PapH pilin domain lacks a P5 pocket. The P1-P4 pockets are intact, and thus PapH can undergo DSC with the chaperone, but not DSE with a subsequent subunit. Therefore, PapH remains bound to the chaperone on the periplasmic face of the usher, anchoring the pilus fiber (Fig. 1).

In contrast to DSC, in which the chaperone’s G1 β-strand is inserted parallel to the F strand of the subunit’s pilin domain, the DSE reaction results in the donated Nte being inserted in a canonical, antiparallel orientation (Fig. 3C). In addition, the Nte donates smaller residues than the bulky residues inserted by the chaperone’s G1 β-strand. Together, these factors allow DSE to produce a complex residing at a lower energy state compared with DSC, resulting in a highly stable subunit-subunit interaction (136). The subunit-subunit interaction has a binding affinity among the strongest noncovalent interactions in nature, with a K_d of 1.5 × 10⁻²⁰ M and a dissociation rate of 3 × 10⁻⁹ years (136, 148). Thus, the assembled pilus fiber, which is an array of Ig folds linked to each other noncovalently through DSE, is an extremely stable organelle that is well suited for bacterial adhesion in varied and stressful environments.

The Pilus Usher

Assembly of the pilus fiber takes place on the periplasmic face of the OM usher protein, with the usher also providing the channel for secretion of the fiber to the cell surface (Fig. 1). The usher catalyzes pilus assembly by a rate of over 1000-fold compared with spontaneous pilus assembly in the absence of the usher (149). The catalytic activity of the usher is thought to be due in large part to its ability to orient two periplasmic chaperone-subunit complexes relative to each other in such a way that they undergo DSE efficiently. The usher is a multidomain, membrane-spanning protein that is present in the OM in oligomeric form. Several structural snapshots of pilus assembly at the usher are now available: the FimD and PapC usher channel domains in their apo states; the isolated FimD N domain bound to FimC-H or FimC-F chaperone-subunit complexes; and the active FimD usher bound to FimC-H (FimD-C-H) or to an assembling pilus tip fiber (FimD-C-F-G-H) (Figs. 4 and 5) (117, 140, 150–153). From these snapshots, as well as accumulated biochemical, genetic, and computational evidence, the stages and mechanism of pilus assembly at the usher are becoming well defined.

Figure 4.

Structures of apo and activated FimD usher from the type 1 pilus system. (A) and (C) Structure of apo FimD (PDB ID: 3OHN), shown from side (A) and top (C) views. The transmembrane β-barrel channel domain is pictured in light blue and the plug domain in pink. The plug domain is positioned laterally within the β-barrel domain, closing the usher channel. The N, C1, and C2 domains are not present in this structure. (B and D) Structure of activated FimD (PDB ID: 3RFZ), shown from side (B) and top (D) views. The channel and plug domains are depicted as in (A), the N domain is in dark blue, the C1 domain is in cyan, and the C2 domain is in purple. In the activated usher, the plug is expelled from the channel and resides adjacent to the N domain in the periplasm.

Figure 5.

Type 1 pilus assembly cycle at the FimD usher and corresponding crystal structures. (A) Cartoon depictions of domain organization and color coding for the Fim proteins shown in panels (B through F). (B) Structure of apo FimD (PDB ID: 3OHN) with the transmembrane β-barrel channel closed by the plug domain. (C) Structure of the FimD N domain bound to a FimC-FimH pilin domain complex (PDB ID: 1ZE3). (D) Structure of the activated FimD usher with bound FimC-H chaperone-adhesin complex (PDB ID: 3RFZ). The FimH lectin domain is inserted inside the usher channel and the FimH pilin domain and bound FimC chaperone are located at the usher C domains. The FimD plug domain resides adjacent to the N domain in the periplasm. Structure of the FimD-C-F-G-H type 1 pilus assembly intermediate (PDB ID: 4J3O). The FimF-G-H pilus tip fiber is traversing the usher channel, with FimH exposed to the cell surface, and FimF bound by FimC located at the usher C domains. (F) Model for type 1 pilus assembly at the FimD usher. In its resting (apo) state, the FimD plug domain resides laterally within the usher channel (structure shown in B). The plug closes the usher channel and also functions to mask the C domains. Pilus assembly initiates with the binding of a FimC-H chaperone-adhesin complexes to the FimD N domain (step 1; structure shown in C). FimC-H binding to the N domain activates the usher by triggering opening of the plug domain and unmasking of the C domains. FimC-H then undergoes handoff from the N to the C domains, concomitant with insertion of the FimH lectin domain into the usher channel (step 2; structure shown in D). The usher N domain functions to recruit the next chaperone-subunit complex, FimC-G, from the periplasm (step 3). The FimC-G complex bound at usher N domain is perfectly positioned to undergo DSE with FimC-H bound at the C domains, forming the first link in the pilus fiber and displacing FimC from FimH. FimC-G is then handed off from the N to the C domains, concomitant with movement of the nascent pilus fiber through the usher channel toward the cell surface (step 4). Repeated cycles of chaperone-subunit recruitment and DSE (step 5) then result in assembly and secretion of the pilus tip (structure shown in E) and finally the pilus rod.

Each usher protomer comprises a 24-stranded β-barrel channel domain, a plug domain that serves as a channel gate, an N-terminal (N) periplasmic domain, and two C-terminal (C1 and C2) periplasmic domains (Fig. 4). In the resting (apo) usher, the plug domain closes the channel pore, preventing nonspecific influx or efflux of proteins (Fig. 4A and C). The usher’s N and C periplasmic domains serve as chaperone-subunit binding sites during pilus assembly. Chaperone-subunit complexes first bind to the N domain of the usher. A quality control mechanism ensures that “empty” chaperones do not bind the usher. Research on the Yersinia F1 capsule CU system revealed a conserved two-proline “lock,” in which the lock occludes the chaperone’s usher-binding interface (154). Only upon chaperone binding to a subunit is the lock rotated away from the usher-binding interface, allowing docking at the usher N domain.

The usher must undergo a global conformational change to initiate pilus assembly. In its resting state, the usher channel is occluded by the plug domain. In the type 1 pilus system, the usher can only be activated by binding of a FimC-H chaperone-adhesin complex to the usher N domain. Activation results in expulsion of the plug from the channel lumen to a periplasmic location, adjacent to the N domain (Fig. 4B and D) (151). During activation, the geometry of the usher channel changes from reniform to round, and the channel area increases by about 100 Å (Fig. 4) (151). Together, these changes allow the usher to accommodate folded pilus subunits within its channel. In the closed, apo state of the usher, the plug domain also functions to mask the usher C domains (Fig. 5F) (155). Therefore, plug expulsion not only exposes the channel lumen, but also frees the C domains and allows them to participate in chaperone-subunit binding.

Following activation, the chaperone-adhesin complex bound to the usher N domain must be “handed off” to the usher C domains (Fig. 5F). The usher N domain has lower affinity for the chaperone-adhesin complex compared with the C domains, and this handoff is likely driven by differential affinity (155). Additionally, the usher C2 domain may facilitate handoff by destabilizing chaperone-subunit binding to the N domain (156). Following handoff, the N domain is cleared of the chaperone-adhesin complex and available to recruit the next chaperone-subunit complex (FimC-G for type 1 pili) from the periplasm (Fig. 5D and F). Together with the plug domain (in the activated usher, the plug domain resides adjacent to the N domain), the N domain serves as a platform for the recruitment of additional chaperone-subunit complexes (151, 156). The newly recruited complex bound at the N domain is thought to be oriented perfectly for its Nte to undergo DSE with the previously recruited chaperone-subunit complex (FimC-H) bound at the C domains (Fig. 5F) (151). DSE displaces the chaperone from the subunit bound at the usher C domains, resulting in formation of the first link of the pilus fiber (FimG-H). The process continues with handoff of the chaperone-subunit complex bound at the N domain to the C domains, concomitant with translation of the nascent pilus fiber through the usher channel toward the cell surface (Fig. 5F). The usher N and C domains bind to the same surface of the chaperone; therefore, handoff requires rotation of the chaperone-subunit complex and pilus fiber (140, 150, 151). Computation analysis indicated that this rotation is facilitated by a helical low-energy pathway through the usher channel (117). Iterative rounds of chaperone-subunit recruitment to the usher N domain, DSE between subunits, and handoff from the usher N to the C domains drive pilus elongation and secretion, resulting in assembly of the tip fiber (Fig. 5E) and finally the rod. The usher channel is only large enough to accommodate a linear fiber of folded pilus subunits. Therefore, the pilus rod adopts its final helical quaternary conformation on the cell surface, upon exiting the usher channel.

The model for the subunit incorporation cycle at the usher as described above and in Fig. 5 shows that a monomeric usher is sufficient for pilus biogenesis, having sites for both recruitment and assembly of the pilus fiber (the N and C domains). However, multiple lines of evidence indicate that the usher is present in the OM as dimeric or higher-order oligomeric complexes (151, 152, 155, 157–159). How the usher oligomer contributes to pilus biogenesis remains to be determined. However, recent evidence indicates that individual usher protomers are capable of functioning in trans to assemble adhesive pili in vivo (155). The usher oligomer might function in an asymmetric manner during pilus assembly, with the individual protomers serving to recruit chaperone-subunit complexes to the OM, but only one usher providing the active translocation channel. Such a mechanism could facilitate pilus biogenesis by increasing the local concentration of chaperone-subunit complexes at the OM. This could be particularly important during assembly of the pilus rod, where thousands of copies of the major pilus subunit (FimA or PapA) must be incorporated.

Ordered Assembly of the Pilus Fiber

The order of assembly of pilus subunits is of great importance for pilus function to ensure proper presentation of the pilus adhesin. Regulation of the sequence of pilus assembly is due to (i) selective activation of the usher; (ii) differential affinities of chaperone-subunit complexes for the usher; (iii) the rate of DSE for each subunit-subunit pairing; and (iv) periplasmic concentrations of chaperone-subunit complexes. In its apo state, the usher C domains are masked and the usher is thought to sample periplasmic chaperone-subunit complexes via its N domain (155). Pilus assembly begins with the adhesin, and chaperone-adhesin complexes have highest affinity for the usher compared with the other chaperone-subunit complexes (156, 160–162). In the type 1 pilus system, activation of the usher (release of the plug domain and unmasking of the C domains) only occurs upon binding of a FimC-H chaperone-adhesin complex to the usher N domain. This provides a mechanism to ensure the assembly of functional pili, with the adhesin at the tip (155). Following activation, the recruitment platform formed by the usher N and plug domains likely facilitates ordered pilus assembly, because the N domain has differing affinities for each chaperone-subunit complex (156, 160–162). However, rates of DSE between subunit-subunit pairs is likely the determining factor in ordering of the pilus fiber. Not surprisingly, DSE is faster (by 2- to 50-fold) for cognate subunits (those that are adjacent in the mature pilus) than for noncognates. Additionally, the relative number of correctly versus incorrectly paired subunits is similar in catalyzed (with usher) and uncatalyzed (no usher) reactions (149, 163). This suggests that the basis for DSE rate differences lies in the specificity of a subunit’s Nte for the pilin domain of the neighboring subunit. Finally, a mathematical model incorporating both experimentally determined DSE rates and usher-binding affinities indicated that order accuracy was also highly dependent on the periplasmic concentration of subunits (163). For example, using the experimentally measured usher-binding affinities and DSE rates, a tenfold excess of FimC-G over FimC-F was needed to obtain a correctly sequenced type 1 pilus tip fiber. Notably, the native periplasmic concentrations of chaperone-subunit complexes have not been thoroughly investigated, and therefore, the extent to which this affects pilus biogenesis remains to be determined.

CHAPERONE/USHER PILUS REPERTOIRES IN E. COLI AND SALMONELLA

In this section, we present a survey of CU pili expressed by E. coli, followed by CU pili of Salmonella spp. Our overview highlights those pili that are assembled by the CU pathway, are fairly well characterized, and have been shown to be involved in human pathology. Detailed descriptions are provided for the type 1 and P pili expressed by UPEC, because these CU pili have been extensively studied. Table 1 provides a list of CU pili present in E. coli and Salmonella.

E. coli CU Pili

CU pili and their component proteins facilitate E. coli pathogenesis through a variety of mechanisms. The prototypical type 1 and P pili are found in UPEC and mediate adhesion to the urinary tract. Type 1 pili are widely present among other E. coli strains as well, and contribute to biofilm formation and colonization of different sites within the host. E. coli strains such as UPEC and ETEC express a variety of additional CU pili to promote adherence, biofilm formation, and other functions. Unlike ETEC and UPEC, which express multiple pili when cultured in vitro, isolates of enterohemorrhagic E. coli (EHEC) express only a few surface structures detectable by electron microscopy upon laboratory culture (164–168). However, whole-genome sequencing of two E. coli O157:H7 isolates revealed the presence of at least 17 pilus gene clusters (47, 169, 170). The construction of transcriptional fusions with 13 of these gene clusters showed that most of these genes are not expressed in vitro (171). Sequence analysis also indicates that some O157:H7 isolates are unable to produce functional type 1 pili (111, 172). Nevertheless, analyses of cloned operons and phenotypes resulting from mutational inactivation demonstrate that many of the EHEC CU loci encode functional pili (173–177). These studies reinforce the concept that E. coli strains express multiple CU pili, contributing to the diverse capabilities of these bacteria to recognize different receptors and cause different pathologies.

Type 1 pili

The chromosomal fimBEAICDFGH gene cluster coding for type 1 pili (Fig. 2) is widely distributed among E. coli strains, including nonpathogenic and laboratory strains. As discussed in “Architecture of Chaperone/Usher Pili” above, FimA is the major pilin subunit that forms the helical pilus rod proximal to the cell surface, FimF and FimG are adapter subunits present in the distal tip fibrillum, and FimH is the tip-associated adhesin (Fig. 1). FimC is the pilus chaperone and FimD the usher protein for type 1 pili. FimI’s function is unknown, although it may be the rod-terminating subunit, analogous to PapH of P pili (93, 94). FimB and FimE are regulatory proteins involved in phase variation of type 1 pilus expression, as described in “Chaperone/Usher Genetic Loci and Regulation.”

Type 1 pili mediate binding to a variety of surfaces and host tissues in a mannose-sensitive manner. Type 1 pili are a major virulence factor of UPEC and facilitate adhesion to the bladder via specific binding to uroplakins and other mannosylated proteins that coat the bladder lumen (178, 179). Antibodies against the type 1 pilus adhesin, FimH, confer protection against UTI in both nonhuman primates and mice (180, 181). However, a definitive requirement for type 1 pili in human UTI remains elusive, likely because of the many adhesins and other virulence factors with redundant or overlapping functions in uropathogenic strains (182). Type 1 pili bind to a variety of receptors and surfaces in addition to urothelial cells, including surface glycoproteins of immune cells, extracellular matrix proteins, and abiotic surfaces (121, 183–185).

Type 1 pili perform a variety of roles in addition to initiating host cell contact and facilitating colonization. The steps by which E. coli infects the urinary tract are illustrative of these pilus-mediated functions. Introduction of UPEC into the bladder typically occurs via ascension of the urethra by fecal bacteria that contaminate the periurethral area. UPEC bind to the bladder surface via interaction of the type 1 pilus adhesin FimH with mannosylated receptors such as the uroplakins. This allows the bacteria to gain a foothold in the urinary tract and resist eradication by urine flow (178, 186). In the mouse UTI model, the bladder epithelium rapidly responds to type 1 pilus-mediated binding of E. coli through the activation of NF-κB. This upregulation occurs specifically in the terminally differentiated superficial umbrella cells of the bladder and is dependent on a dual ligand/receptor interaction: FimH and the uroplakin Ia receptor; and lipopolysaccharide and Toll-like receptor 4 (TLR4) (187).

Bacterial binding via type 1 pili also activates host cell pathways that lead to actin cytoskeletal rearrangements in the urothelium, and subsequent bacterial invasion into the host cells via a zipper-like mechanism (186, 188). Bacterial uptake by bladder epithelial cells is initiated by FimH binding to host β1- and α3-integrins (179, 188). Type 1 pili also promote adhesion to phagocytes and survival inside macrophages (185, 189–191). In addition to type 1 pilus functions mediated by the FimH adhesin, the FimA major rod subunit acts to potently inhibit Bax-mediated cytochrome C release in infected cells, thereby suppressing host cell apoptosis (192). FimA does this by binding to VDAC1 (voltage-dependent anion channel-1) and stabilizing the mitochondrial VDAC1-hexokinase complex, preventing its dissociation in response to apoptotic stimuli (192). This effect may be mediated by a soluble version of the FimA monomer, resulting from intramolecular self-DSC (193).

Following invasion, UPEC replicate within the superficial umbrella cells of the bladder and form intracellular biofilm-like communities (IBC). Type 1 pili continue to be expressed by the intracellular bacteria and are important in IBC formation, illustrating a function for pili that is distinct from initial binding and invasion (194). Bacteria expressing mutants of FimH that are functional for mannose binding, but defective for IBC formation, are attenuated for pathogenesis and rapidly cleared in the mouse UTI model, underscoring the distinct functions of type 1 pili (195). Type 1 pili also contribute to extracellular biofilm production by E. coli in a variety of environments (183).

Bladder cells respond to UPEC binding and invasion by sloughing off into the bladder lumen, a mechanism used by the host to eradicate the bacteria (186). This sloughing off is counteracted by UPEC fluxing out of the host cells and undergoing further rounds of binding to the bladder epithelium and invasion, in a process likely mediated by type 1 pili as in the initial round of infection (196, 197). Through additional iterations of this process, UPEC may gain access to underlying layers of the bladder epithelium, wherein the bacteria survive in quiescent reservoirs. The bacteria may then reseed the bladder lumen from these reservoirs and initiate a subsequent infection. Such reseeding of the bladder from these quiescent reservoirs likely accounts for a subset of recurrent UTI cases, which will also depend on type 1 pili for host cell adhesion and invasion (196, 198).

P pili

P pili are encoded by the chromosomal pap (pyelonephritis-associated pili) gene cluster papIBAHCDJKEFG (Fig. 2). The pap gene cluster resides on pathogenicity islands of UPEC strains (the cluster is absent in laboratory and many nonpathogenic strains), and is also found in E. coli strains that cause neonatal meningitis (199). More than one pap gene cluster may be present in an individual E. coli strain. The papI and papB genes form the upstream regulatory region, as described in “Chaperone/Usher Genetic Loci and Regulation.” PapA is the major pilus subunit that forms the helical pilus rod; PapK, E, and F are adapter subunits that form the distal P pilus tip fiber; and PapG is the tip-associated adhesin (Fig. 1). PapD is the chaperone and PapC is the usher for P pili. PapH is the rod-terminator subunit that determines pilus length and anchors the pilus fiber in the OM (92, 142). PapJ’s function is unknown, although it may be involved in ensuring the integrity of the pilus fiber during assembly (200).

P pili are associated with the ability of UPEC to colonize the kidney and cause pyelonephritis (112, 201, 202). P pili bind to Gal(α1-4)Gal moieties present on the globoseries of glycolipids of kidney epithelial cells. Binding of PapG to kidney epithelial cells facilitates attachment of UPEC during an ascending UTI, and is one of the initiating steps of pyelonephritis. The glycolipid receptor is also present on the P blood group antigen, allowing P pilus-mediated agglutination of human erythrocytes (and the basis of the useful laboratory hemagglutination assay to quantitate adhesive pilus assembly) in a mannose-resistant manner (10). There are three variants of the PapG adhesin: PapGI, GII, and GIII (203, 204). Each of these recognizes distinct isoreceptors, differing in carbohydrate residues distal to the Gal(α1-4)Gal core. PapGI, GII, and PapGIII bind to globosides GbO3, GbO4, and GbO5, respectively. PapGII binds preferentially to the common kidney glycolipid isoreceptor, globoside (GbO4; tetrasaccharide GalNAcβ1-3Galα1-4Galβ1-4Glc linked to ceramide), and is correlated with acute pyelonephritis in humans. In contrast, PapGIII is correlated with human bladder infection.

In support of the role of P pili in UPEC pathogenesis, vaccination of both mice and primates with P pili results in protection against pyelonephritis (205, 206). However, as found for type 1 pili, mutagenesis studies have had mixed results in establishing P pili as essential for bacterial infection (207). P-pilus-mediated adhesion of E. coli to the urinary tract stimulates cytokine production and release, and thus induces inflammatory responses that likely exacerbate kidney damage in acute pyelonephritis (208–210). P pilus binding to its globoside receptor also induces release of the second messenger ceramide. Ceramide is the membrane-anchoring portion of the receptor and a TLR4 agonist. In turn, TLR4 activation leads to the production of proinflammatory cytokines (IL-6) and chemokines (CXCL-8), and neutrophil recruitment (211, 212). Additionally, through TLR4 interaction, P pilus binding modulates the local secretory antibody immune response by downregulating polymeric immunoglobulin receptor (PIGR) expression, thereby impairing immunoglobulin A transport to the kidney lumen (213). Thus, UPEC evades this host-protective mechanism, facilitating the establishment of an ascending infection (43, 213). These findings demonstrate a link between bacterial adhesion and the induction and modulation of host innate immune pathways. UPEC binding via PapG also activates signaling pathways within the bacteria that lead to upregulation of iron acquisition processes and machinery, and may prepare UPEC for urinary tract colonization (214).

Aggregative adherence fimbriae

Aggregative adherence fimbriae (AAF) are a family of pili expressed by enteroaggregative E. coli (EAEC). Genes clusters coding for AAF are present on large plasmids in EAEC strains and comprise aafD (chaperone), aafC (usher), aafB (minor pilin subunit), and aafA (major pilin subunit) (90, 215–218). AAF belong to the FGL subfamily of CU pathways, and assemble as thin (2- to 3-nm diameter) fibers with variable morphology. The AAF/I and AAF/II pilus fibers are somewhat flexible and form bundles, wrapping around neighboring bacterial cells, whereas AAF/III and AAF/IV pilus fibers are longer and more flexible (217, 219–221).

EAEC use AAF to promote binding to the human intestinal wall (215, 222, 223). AAF bind to several different epithelial cell receptors, including decay-accelerating factor (DAF) and extracellular matrix proteins (90, 224). Once bound to the intestine, EAEC elaborate toxins, which in turn cause induction of proinflammatory cytokines from the infected epithelium and a watery diarrheal illness (225, 226).

Afa/Dr pili

Both uropathogenic and diarrhea-associated E. coli strains use afa gene clusters to encode afimbrial adhesins (95–98). Based on the commonality of binding to the decay-accelerating factor (DAF, CD55) as a pilus receptor, the Afa/Dr family includes Afa, Dr, and F1845 pili, among others (90, 91). The genes coding for the structural and assembly components of Afa/Dr pili are organizationally well conserved and ordered as follows: afaABCDE. The afaA gene encodes a regulatory protein, afaB the pilus chaperone, afaC the usher, afaD the tip-located invasin subunit, and afaE the major subunit protein. AfaE, which also functions as the pilus polyadhesin, is the least conserved in sequence among family members, accounting for receptor-binding variability. UPEC code for Dr adhesins using the draABCDE gene cluster (Fig. 2) (227–229). F1845 pili facilitate diffuse cell adherence of E. coli diarrheal isolates, and have a genetic organization similar to both afa and dra. While the genes encoding F1845 pili were first cloned from the bacterial chromosome, hybridization studies revealed that similar coding regions may be plasmid-associated in other strains (230, 231).

Afa/Dr pili belong to the FGL subfamily of CU pathways. The pili assemble as flexible, polyadhesive fibers (Fig. 1), which confer mannose-resistant hemagglutination (95, 232–235). Afa/Dr pili enable binding to the Dr^a blood-group antigen present on CD55/DAF, a signaling molecule involved in complement regulation (233). CD55/DAF prevents amplification of the complement cascade, via direct binding of C3b or C4b (236, 237). The Afa/Dr adhesins also bind carcinoembryonic antigen (CEA)-related cellular adhesion molecules (CEACAM). Pathogen binding to this receptor is followed by a CEACAM-associated signaling cascade, which triggers a remodeling of the host cell surface that assists bacterial colonization of the urinary tract (238, 239).

The different binding tendencies of the Afa/Dr pili give rise to different pathologies. There are correlations between Afa/Dr-expressing strains of E. coli and low birth weight and preterm delivery in mice, as well as gestational pyelonephritis (240, 241). Additionally, there is a correlation between Afa/Dr-encoding strains and increased colonization and invasive capacity (242). Afa/Dr pili have also been implicated in inducing proinflammatory responses in host cells. E. coli strains expressing Afa/Dr pili promote the basolateral secretion of IL-8 through the activation of mitogen-activated protein kinases (MAPK) (243). IL-8 then induces transmigration of neutrophils across the epithelial monolayer, which in turn induces tumor necrosis factor alpha and IL-1β, upregulating DAF and thereby increasing Afa/Dr adhesion (244). Additionally, Afa/Dr diffusely adhering E. coli induce activation of CD55/DAF signaling in an adhesin-dependent manner, which in turn activates vascular endothelial growth factor (VEGF). Elevated VEGF promotes angiogenesis in inflammatory bowel diseases, including Crohn’s disease and ulcerative colitis, providing a link between pilus-mediated bacterial adhesion and angiogenesis in disease (245).

F1C pili

F1C pili are nonhemagglutinating adherence factors present in approximately 14% of UPEC isolates and 7% of E. coli fecal isolates (246–250). Eight chromosomally encoded genes, termed foc, are necessary for F1C pilus biogenesis, and have a nearly identical organization to the type 1 pilus fim operon. F1C pili are similar in architecture to type 1 pili, with dimensions of ∼1 μm in length and 7 nm in diameter (251). FocA is the major pilus subunit, FocF and FocG are the minor pilus subunits, and FocH is the tip-located adhesin. FocC is the pilus chaperone and FocD the usher (251, 252). F1C pili adhere to the distal tubules and collecting ducts of the kidney, and to vascular endothelial cells of the kidney and bladder cells (248, 250, 253, 254). F1C pili have also been implicated in biofilm formation, and in induction of the proinflammatory cytokine IL-8 on renal epithelial cell attachment (255, 256).

S pili

S pili are virulence-associated adhesive organelles present in extraintestinal pathogenic E. coli (ExPEC) strains that cause urinary tract infections and neonatal meningitis (250, 257–259). The genes coding for S pili were first isolated from a UPEC strain (260). The sfa gene cluster is chromosomal and consists of nine genes (261). Similar to type 1 and P pili, S pili contain a proximal helical rod (7 nm in diameter) and a distal linear tip fiber. The rod is composed of the major subunit SfaA. The tip, similar to P pili, is relatively long and flexible, and contains the sialic acid-binding adhesin SfaS at its distal end (248, 262). The usher in the S pilus assembly system is SfaE, and the chaperone is SfaF (21, 263, 264).

S pili bind to sialic acid-containing oligosaccharides, facilitating the recognition of and binding to the urinary tract and also to brain microvascular endothelial cells (265, 266). S pili have also been implicated in binding to erythrocytes, human fibronectin, and the extracellular matrix protein laminin (267, 268). Furthermore, S pili immobilize human plasminogen and tissue-type plasminogen activator (tPA), thereby activating plasminogen and promoting plasmin proteolytic activity on the bacterial cell surface (269).

CS1, CFA/I, and CFA/II pili

ETEC express a large group of pili, termed colonization factor antigens (CFAs) or coli surface antigen (CS), which are part of the alternate family of CU pili that belong to the σ clade (29). The coli surface antigen 1 (CS1) and CFA/I pili are representative members of this CU family. CS1 pili are encoded by the cooBACD operon, which is present on the large pCoo plasmid found in ETEC strains (270). Expression of this operon requires the Rns transcriptional activator, which belongs to the AraC DNA-binding protein family, and is encoded on a separate plasmid (271, 272). In the absence of Rns, coo operon expression is repressed by the global regulatory protein H-NS (272). CFA pili are coded for by the cfaABCE gene cluster, present on the NTP113 ETEC plasmid (273, 274). Expression of these pili is regulated by the CfaD-positive regulator, present on the same plasmid, and homologous to Rns (275). CFA/I pili are expressed at 37°C, but remain repressed at lower temperatures, in a process mediated by H-NS, similar to the CS pili (276).

CS1 and CFA pili are large and rigid: approximately 6 to 7 nm in diameter (277). The pili are composed of two structural subunits and lack a distinct tip structure (31, 278). CFA and CS pilus fibers are composed of repeating copies of the major pilus subunit, cooA for CS1 and CfaB for CFA/I pili. CooD and CfaE are the minor subunits for CS1 and CFA/1, respectively (279, 280). Both the major and minor subunits may contribute to adhesion, with each appearing to possess distinct binding properties (31, 281–284). In CS1 pili, CooB serves as the periplasmic chaperone and CooC serves as the OM usher. In CFA/1 pili, CfaA and CfaC are the chaperone and usher, respectively.

ETEC colonizes the human small intestine and secretes enterotoxins into the gut lumen. This causes a disruption in fluid homeostasis, which leads to acute gastroenteritis (285, 286). ETEC is the predominant causative agent of gastroenteritis, otherwise known as traveler’s diarrhea, and also is a major cause of diarrhea in developing nations, particularly among children (287, 288). The CS and CFA pili, which primarily mediate adhesion to the small intestine, are critical for ETEC virulence (277).

Salmonella CU Pili

The CU pili of Salmonella are less well characterized than the E. coli pili, in part because of a general lack of pilus expression by strains grown under laboratory culture conditions (50). The S. enterica serotype Typhimurium genome contains 13 gene clusters potentially coding for surface structures—fim, agf (csg), pef, lpf, saf, bcf, stb, stc, std, sth, stf, sti, and stj (17). However, among the CU pathways, evidence for in vitro expression of the various pili is limited (50, 289). Nevertheless, findings from studies on pilus expression in vivo and on the role of pilus gene clusters during colonization of the mouse intestine suggest that most of the Salmonella CU genetic loci encode functional pili (50, 289–292).

Long polar fimbriae

The long polar fimbriae (LPF) locus consists of chromosomal genes lpfABCDE and was identified during a screen for S. Typhimurium loci not present in related Enterobacteriaceae (293, 294). Based on the homology of flanking chromosomal regions to E. coli, the lpf genes likely were acquired by horizontal gene transfer. The lpf operon has a similar organization to that of the type 1 pilus fim operon, and the Lpf protein sequences are homologous to the Fim proteins (293, 295). LpfD is the adhesin, LpfA and LpfE encode subunit proteins, LpfB serves as the pilus chaperone, and LpfC is the OM usher (21, 293). LPF facilitate bacterial adhesion to M cells of Peyer’s patches in the small intestine, as determined using a mouse model of infection (290). Deletion of the lpf locus has only a minor effect on the virulence of S. Typhimurium in the mouse infection model, but a double-deletion mutant of lpf together with the invA type III secretion system (T3SS) component produces a synergistic effect (296), suggesting that LPF assist in bacterial docking to host cells to facilitate delivery of effector proteins via the T3SS injectisome. Phase-variable lpf expression has been demonstrated, with LPF phase-ON bacteria appearing more frequently in the mouse Peyer’s patches than phase-OFF bacteria (297). Such a differential was not seen in bacteria isolated from the mesenteric lymph nodes or spleens of the infected animals. Additionally, phase-ON variants were strongly selected against upon immunization of mice with the LpfA major fimbrial subunit protein (297). These findings support a model wherein LPF are expressed in vivo and are important for an early adhesive step of infection via the oral route, but not during later stages of infection when the bacteria have disseminated to systemic sites.

Plasmid-encoded fimbriae

Plasmid-encoded fimbriae (PEF) are expressed from the plasmid-based pefBACD locus. The pef gene cluster is responsible for formation of surface fibers in both E. coli and S. Typhimurium (298). PefD is the chaperone, PefC the usher, PefA the major structural pilus subunit, and PefB encodes a regulatory protein. PefA polymerizes as a thin and flexible fiber, with a diameter of 2 to 4 nm. PEF promote adhesion of S. Typhimurium to the villous small intestine, and are necessary for infection-associated pathology such as fluid accumulation in an infant mouse model (299). Additionally, PEF promote bacterial adherence to various human cell lines, including HEp-2, T-84, Int-407, and HeLa cells (300). The pef genes are only found in a subset of phylogenetically related strains, and were likely recently acquired by Salmonella (301, 302). The acquisition of pilus loci such as pef may provide an evolutionary mechanism for Salmonella serovars and other bacteria to expand their host range.

Salmonella enteritidis fimbriae

The S. enteritidis fimbriae (SEF) are encoded by the chromosomal sefABCD locus. SefA and SefD are subunit proteins, SefB is the periplasmic chaperone, and SefC the OM usher (303, 304). SefC belongs to the FGL subfamily of CU pilus chaperones, and assembles two distinct, thin, polyadhesive fibers composed of either SefA or SefD, termed SEF14 or SEF18, respectively. Thus, the SefB and SefC chaperone-usher pair assembles homopolymers of either SefA or SefD. As with the pef locus, the sefABCD genes were likely acquired recently by Salmonella, and are expressed only by S. enteritidis and closely related serovars (301, 302). SEF are expressed in vivo during infection, but the exact contributions of these pili to virulence are not well characterized (304, 305). Evidence suggests that SefD may function as an adhesin for S. enteritidis that facilitates invasion inside host macrophages, and that the pili may be important for invasive stages of infection following the initial intestinal colonization (305).

Type 1 pili

The type 1 pili of Salmonella are encoded by a chromosomal gene cluster containing six genes encoding structural and assembly components (fimAICDHF) together with three separately transcribed regulatory genes (fimZYW) (306, 307). These pili exhibit mannose-sensitive hemagglutination, similar to the type 1 pili of E. coli (308). The Fim proteins of Salmonella exhibit a corresponding functional relationship to their counterparts in E. coli type 1 pili, but the Salmonella fim genes and gene products exhibit little sequence similarity with the E. coli genes. FimA is the major subunit, FimF the adaptor subunit, and FimH the tip-located adhesin. FimC is the periplasmic chaperone and FimD the OM usher. FimI has an unknown role, as a fimI deletion mutant resulted in an eightfold higher hemagglutination titer, but morphologically normal pili (306). FimI in Salmonella does not appear to be related to the E. coli FimI protein (94). Type 1 pili play a role in the initial stages of Salmonella infections in a variety of animal models including mice, chicks, and hens (309–313). Type 1 pili have also been shown to limit bacterial dissemination and colonization of mice, as well as induce intestinal inflammation during Salmonella invasion (314).

CU PILI AS THERAPEUTIC TARGETS

Pili function as virulence factors for many bacterial pathogens (1, 315). Pilus-mediated adhesion is critical for the early stages of infection, allowing bacteria to establish a foothold within the host and dictating tropism toward specific tissues. Following bacterial attachment, pili also modulate host cell signaling pathways, promote or inhibit invasion inside host cells, and function in bacterial-bacterial interactions leading to the formation of community structures such as biofilms. Given their varied and central functions in pathogenesis, pili make attractive therapeutic targets (316, 317). Knowledge gained from basic science efforts, as described in the preceding sections, has generated a comprehensive understanding of the molecular biology of CU pilus assembly and adhesion. This detailed mechanistic information, in turn, has fueled translational efforts to develop pilus-focused therapeutics, such as vaccines and small-molecule inhibitors, with the goal of disrupting pilus assembly or function, and thereby treating or preventing disease caused by Gram-negative bacterial pathogens.

Therapeutics directed against virulence factors such as pili represent an alternative to traditional antibiotics. Traditional, broad-spectrum antibiotics interfere with essential biological processes, and indiscriminately target the beneficial host flora along with bacterial pathogens. In contrast, antivirulence therapeutics are designed to inhibit systems only required for pathogens to cause disease within the host (318–320). Therefore, antivirulence therapeutics should spare the commensal bacteria and also reduce the evolutionary pressure leading to antimicrobial resistance. Rates of antibiotic resistance have reached alarming levels, raising the possibility of a postantibiotic era, in which even common infections could become life threatening (321–324). This looming health crisis is compounded by the limited availability of new antibiotics and the rapid emergence of resistance once new antibiotics are introduced into use. Antivirulence therapeutics directed against targets such as bacterial pili represent one approach to address this urgent situation.

In this section, we describe strategies being taken for the therapeutic targeting of CU pili, using the UPEC type 1 and P pili as models. Effective strategies against UPEC type 1 and P pili will likely also be relevant to the many other pili assembled by the CU pathway, due to conservation of the CU assembly mechanism, pilus structure, and pilus function.

Vaccination Approaches for CU Pili

The location of pilus fibers on the bacterial cell surface exposes them to the immune system, making pili ideal targets for vaccination. However, efforts to develop vaccines using whole pili have generally been unsuccessful, because of factors such as phase-variable expression and antigenic variation. In addition, vaccination with whole pili may not generate antibodies that inhibit pilus function, particularly for pili such as the UPEC type 1 and P pili, where the adhesin subunit is present at only one copy per pilus and the immune response is biased toward the main structural subunits instead. One promising approach to address this issue is to vaccinate with the purified adhesin subunit, rather than using whole pili. In both murine and primate cystitis models, substantial protection against UPEC infection was conferred on systemic vaccination with the FimH (type 1 pili) or PapG (P pili) adhesins (180, 181, 325, 326). Similarly, a truncated FimH construct containing only the adhesin domain conferred protection against cystitis in mice vaccinated intramuscularly or intranasally, using CpG oligonucleotides as an adjuvant (327). The effectiveness of vaccination with FimH was largely due to the generation of antibodies that blocked FimH-mediated bladder colonization. However, in some cases, rather than blocking type 1 pilus-mediated adhesion, the host antibody response may instead enhance binding of FimH to its receptor by stabilizing the adhesin’s high-affinity binding state (328). An improved understanding of the catch-bond mechanism of FimH binding, and pilus adhesion under fluid flow conditions in the urinary tract, may allow tailoring of the antigen used for vaccination to generate a desired, antiadhesive immune response. Importantly, vaccination against FimH does not appear to alter the commensal E. coli in the gut (180).

Small-Molecule Inhibitors of CU Pili

An alternative approach to vaccination is to target the assembly or function of CU pili through the use of small-molecule inhibitors. Several approaches are being taken to develop such inhibitors for UPEC and other pathogens that express CU pili, taking advantage of the extensive knowledge of the pilus assembly pathway and pilus-mediated adhesion mechanisms.

One approach is to develop small molecules that competitively inhibit adhesin-receptor interactions. For the FimH adhesin of type 1 pili, which bind to mannosylated proteins, soluble receptor analogs termed mannosides are being pursued (329, 330). Mannosides bind to and occupy the FimH receptor-binding site, and act as antiadhesives by preventing bacterial colonization of the urinary tract. Indeed, analysis of a mouse model showed that such molecules act prophylactically to prevent bacterial invasion of the bladder following urinary tract infection (329). Studies have indicated that mannosides are orally bioavailable and are also effective against an established UTI, as well as against catheter-associated UTI (329–332). Mannosides have been shown to act in synergy with traditional antibiotic treatments in their capacity to reduce urinary tract UPEC titers in mice (332). Galabiose-based inhibitors are similarly being developed to inhibit bacterial adhesion mediated by P pili (333).

A second approach toward small-molecule anti-pilus therapeutics is to develop inhibitors that target the pilus assembly and secretion process, with the goal of preventing pilus fibers from being expressed on the bacterial surface. One such class of small-molecule CU pilus inhibitors, known as pilicides, consists of molecules with a 2-pyridine scaffold (334, 335). Pilicides bind to the periplasmic chaperone and interfere with chaperone-subunit interactions or the binding of chaperone-subunit complexes to the usher. In vitro studies have demonstrated that such molecules attenuate type 1 and P pilus biogenesis and pilus-mediated adhesion and biofilm formation (334, 336). The mechanism of action and effects of pilicides may be broader than initially envisioned. In a recent study, pilicide ec240 was found to disrupt type 1, P, and S pili, and to also affect flagellar motility (337). The ec240 pilicide inhibited pilus expression as well as interfered with pilus assembly. Ec240 caused downregulation of the type 1 pilus genes by switching fimS to the phase-OFF position (337). Treatment of bacteria with ec240 also increased levels of the S and P pili regulatory proteins SfaB and PapB, which both further promote fim phase-OFF variation (337). Thus, pilicides may also work by influencing pilus phase variation and thus pilus expression, in addition to disrupting protein-protein interactions required for pilus assembly. Modified pilicides have also been developed that have activity against curli in addition to type 1 pili (338). Treatment of UPEC with one such pilicide reduced biofilm formation and attenuated bacterial colonization in the mouse UTI model (338).

Additional targets in the CU pathway are being exploited for therapeutic intervention. Small-molecule “pilicides” have been developed with alternate mechanisms of action compared to the originally developed pilicides. Lo et al. used computational screening to identify a small molecule, AL1, that inhibits polymerization of the type 1 pilus fiber by disrupting the DSE reaction between the FimH and FimG subunits (339). The AL1 compound disrupted type 1 pilus biogenesis by UPEC, and reduced biofilm formation and bacterial adhesion to human bladder epithelial cells. Another small-molecule compound, nitazoxanide (NTZ), was shown to inhibit biofilm formation by EAEC, by preventing assembly of the AAF CU pili (340). Further analysis demonstrated that NTZ inhibits assembly of type 1 and P pili as well, and that the drug acts through a novel mechanism of action to prevent proper folding of the usher protein in the bacterial OM (341). Coilicides are a recently developed class of anti-pilus inhibitors (342). Coilicides act by preventing uncoiling and recoiling of the helical pilus rod. As discussed in “Architecture of Chaperone/Usher Pili,” the compliance of the pilus rod is critical for a redistribution of forces during shear stress such as caused by urine flow. In a proof-of-concept study, the PapD chaperone was shown to bind to an uncoiled P pilus rod, preventing recoiling of the rod and thus locking it in a noncompliant form (342). Only a few molecules of PapD per pilus rod (>1000 subunits) were necessary for coilicide activity, thus making recoil attenuation a viable opportunity for therapeutic intervention.

The examples described in this section highlight the potential for a new class of antimicrobial therapeutic agents that target the assembly and function of CU pili. Such agents have the potential to alleviate disease by disrupting critical host-pathogen interactions, while sparing the beneficial bacterial flora of the host and addressing the growing health threat of antibiotic resistance.