ABSTRACT
The enterotoxigenic Escherichia coli (ETEC) pathotype, characterized by the prototypical strain H10407, is a leading cause of morbidity and mortality in the developing world. A major virulence factor of ETEC is the type II secretion system (T2SS) responsible for secretion of the diarrheagenic heat-labile enterotoxin (LT). In this study, we have characterized the two type II secretion systems, designated alpha (T2SSα) and beta (T2SSβ), encoded in the H10407 genome and describe the prevalence of both systems in other E. coli pathotypes. Under laboratory conditions, the T2SSβ is assembled and functional in the secretion of LT into culture supernatant, whereas the T2SSα is not. Insertional inactivation of the three genes located upstream of gspC β (yghJ, pppA, and yghG) in the atypical T2SSβ operon revealed that YghJ is not required for assembly of the GspDβ secretin or secretion of LT, that PppA is likely the prepilin peptidase required for the function of T2SSβ, and that YghG is required for assembly of the GspDβ secretin and thus function of the T2SSβ. Mutational and physiological analysis further demonstrated that YghG (redesignated GspSβ) is a novel outer membrane pilotin protein that is integral for assembly of the T2SSβ by localizing GspDβ to the outer membrane, whereupon GspDβ forms the macromolecular secretin multimer through which T2SSβ substrates are translocated.
INTRODUCTION
The majority of Escherichia coli strains serve as integral members of the gut microbiota and therefore are important in maintenance of healthy gastrointestinal tract function (89). However, some strains have evolved or acquired genetic factors that enable E. coli to colonize, dominate, and disseminate into a variety of niches in the host, in the process causing disease. Infections caused by the enterotoxigenic E. coli (ETEC) pathotype occur in the epithelial layer of the small intestine and lead to development of severe diarrhea. ETEC infections are the most common cause of traveler's diarrhea (18), and in areas of the world where ETEC is endemic, it is estimated to cause 400,000 deaths annually of children under five years of age (72). A major characteristic of the ETEC pathotype, exemplified by strain H10407 that was isolated from a patient with severe cholera-like disease in Bangladesh (25), is the secretion of the heat-labile enterotoxin (LT) via the type II secretion system (T2SS) (39, 88).
LT is a member of the AB5 family of enterotoxins and is composed of five 11.6-kDa B subunits (LTb) and one 28-kDa A subunit (LTa) that are assembled into an 84-kDa protein (84). The holotoxin is secreted by the T2SS across the outer membrane (88) and binds to the GM1 ganglioside receptor on the surface of the intestinal epithelium in the small intestine of the infected individual (2, 58). Once internalized, LTa activates adenylate cyclase by ribosylating the stimulatory G protein (32, 57). This activation leads to an increased concentration of cellular cyclic AMP (cAMP) that ultimately results in the loss of water by the cell, thereby causing diarrhea (60).
The T2SS is a large complex composed of 12 to 16 proteins that span both the inner and outer membranes of Gram-negative bacteria. In most species, the genes that encode the T2SS proteins are arranged in a major operon composed of genes gspC, -D, -E, -F, -G, h-H, -I, -J, -K, -L, -M, -N, and -O (gspC-O) and, in some cases, a minor operon composed of gspA and gspB or an independently encoded gspS (77). Proteins destined to be secreted by the T2SS are translocated across the inner membrane via the Sec (general secretion) (67) or Tat (twin-arginine translocation) (93) pathways into the periplasm. From the periplasm, folded proteins are transported through a pore formed by the megadalton-sized multimeric GspD complex located in the outer membrane (6). This pore is presumably gated by the activity of a pseudopilus composed of proteins GspG-GspK that is hypothesized to extend and retract from an inner membrane platform complex composed of proteins GspE, -F, -L, and -M (10, 11, 24, 50, 69, 71, 85; reviewed in reference 49). Assembly of the pseudopilus requires the function of a prepilin peptidase that cleaves the N-terminal amino acids from the prepilin protein as it assembles into the inner membrane (96). The involvement of the T2SS in bacterial pathogenesis is well documented; in addition to the secretion of LT by ETEC, examples include the secretion of aerolysin in Aeromonas hydrophila (46), hemolysin in Burkholderia vietnamiensis (26), and cholera toxin in Vibrio cholerae (78).
The GspD protein of the T2SS is a member of a family of outer membrane transporters termed the secretins. Other members of this family include those of the type III secretion system (T3SS), toxin-coregulated pili, type IV pili, type IV bundle-forming pili, and filamentous phage (90). In each system, the secretin functions as the outer membrane pore through which proteins or macromolecular complexes are translocated. In some systems, localization of the secretin in the outer membrane requires the function of a small lipoprotein that serves as a pilotin to direct the secretin to the outer membrane and protect the multimer from degradation. To date, two highly similar pilotins involved in localization and protection of a T2SS secretin have been elucidated; these include PulS of Klebsiella oxytoca (36, 61) and OutS of Dickeya dadantii (83). Other pilotins involved in secretin assembly of other systems have been described and include the pilotin MxiM for assembly of the MxiD secretin of the Shigella flexneri type III secretion system (T3SS) (81) and the pilotin YscW for assembly of the YscC secretin of the Yersinia enterocolitica T3SS (8). The requirement for a pilotin in assembly of the T2SS secretin may not be entirely conserved since several species that encode a T2SS, including A. hydrophila and Aeromonas Salmonicida, absolutely require the peptidoglycan-binding and ATPase functions of the inner membrane complex GspAB for secretin assembly and do not encode a recognized pilotin protein in the T2SS operon (3, 40, 45, 53). V. cholerae, Vibrio vulnificus, and Vibrio parahaemolyticus also contain the GspAB complex; however, gspA mutations significantly decrease but do not eliminate the assembly of the secretin, indicating that other factors, possibly including an unidentified pilotin, are involved in its assembly (86).
In this study, we describe the presence of two T2SSs, designated alpha and beta, encoded within the genome of ETEC strain H10407 and determine the prevalence of these systems among other E. coli pathotypes. Similar to the situation in E. coli K-12 (28), we provide evidence that the T2SSα is not assembled in ETEC under standard laboratory conditions since only upon replacement of the T2SSα endogenous promoters with inducible ones is the GspDα secretin multimer assembled. Under laboratory conditions the GspDβ secretin of the T2SSβ is readily detectable, and, as previously observed, expression of a functional T2SSβ is required for secretion of LT into culture supernatant (88). Characterization of T2SSβ assembly and function in mutants of three genes of the T2SSβ operon that are located atypically upstream of the gspCβ-Mβ genes revealed that YghJ is not required for assembly or function of T2SSβ and could be a substrate of the T2SSβ, that pppA likely encodes the T2SSβ prepilin peptidase that is required for assembly of the pseudopilus, and that the hypothetical open reading frame (ORF) yghG encodes a novel outer membrane lipoprotein that functions as a pilotin. The pilotin function of YghG is integral for T2SSβ assembly since in the absence of YghG, most GspDβ is degraded or localized to the inner membrane in monomeric form. In the presence of YghG, GspDβ is localized to the outer membrane and assembles in the multimeric secretin form required for assembly of a functional T2SSβ. As a result, YghG serves as an essential member of the GspDβ outer membrane macromolecular assembly system that is required for construction of a functional T2SSβ involved in secretion of the main ETEC virulence factor LT.
MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions.The strains and plasmids used in this study are shown in Table 1. E. coli H10407 was routinely grown in CAYE medium (2% Casamino Acids, 0.15% yeast extract, 43 mM NaCl, 50 mM K2HPO4, 0.25% glucose, 0.1% trace salts [0.415 M MgSO4, 40 mM MnCl2, 30 mM FeCl3]) until mid-exponential phase of growth at 37°C (76). E. coli DH5α was cultured in Luria-Bertani (LB) medium (5). Antibiotics were used at the following concentrations unless otherwise specified: kanamycin (Kan) at 30 μg/ml, chloramphenicol (Cam) at 20 μg/ml, and ampicillin (Amp) at 30 μg/ml.
Strains, plasmids, and primers used in this study
Construction of deletion mutations in H10407.Nonpolar deletion mutations were constructed in the ETEC genome by use of a λRed recombination system (97) (Gene Bridges, Heidelberg, Germany). Three clones of wild-type H10407 were transformed with the temperature-sensitive pRED/ET(Amp) plasmid and were utilized in triplicate to construct the various deletion mutants. Deletion mutations were created by removal of the entire open reading frame (ORF) with exception of 30 bp at the beginning and end of the gene, with insertion of a 69-bp “scar” sequence encoding a flippase recombination target (FRT) in the middle. By this method, a 53-amino-acid protein is encoded that is composed of 10 amino acids of the N terminus of the protein, 10 amino acids of the C terminus of the protein, and 33 amino acids encoded by the scar sequence.
Linear DNA encoding chloramphenicol acetyltransferase (cat) and aminoglycoside 3′ phosphotransferase (kan) genes that confer Cam and Kan resistance, respectively, flanked by FRT sequences were constructed for replacement into the ETEC genome by λRed recombination. The cat gene carried by pBAD322C was amplified with primers US372 and US373 and inserted into the BspEI site of FRT-kan-FRT carried by plasmid TS12. The resultant plasmid TS100 was digested with BstEII to remove the FRT-cat-kan-FRT fragment (TS101) to be used as the template for amplification with primers listed in Table 2. As shown in Table 2, two primers were used to amplify each end of the FRT-cat-kan-FRT fragment, and they added 60 bp of H10407 genomic target sequence of the gene to be deleted. PCRs were performed using a 100-fold molar excess of short primers to long primers. Primers used for amplification of FRT-cat-kan-FRT and subsequently for λRed recombination and deletion of gspDα, gspDβ, yghJ, pppA, and yghG are given in Table 2. Routinely, 100 to 200 ng of amplified product was used for electroporation into competent H10407(pRED ET/Amp) cells. Following electroporation into electrocompetent cells of H10407 (59) containing pRED ET/Amp, transformation cultures were grown at 37°C for 4 h and plated onto LB medium containing Kan (30 μg/ml) and Cam (2.5 μg/ml) and grown at 37°C. The cat and kan genes were removed from the genome by recombination by the FLP recombinase encoded within plasmid pCP20 (15). Deletion mutations that include the 69-bp scar sequence were verified by PCR and sequencing analysis.
Primers used in this study
Construction of complementation plasmids.The yghG open reading frame (ORF) was amplified from the H10407 genome with primers US420 and US398 (Table 2) for cloning into vector pBAD322C to construct complementation plasmid pBAD/yghG. Primers US420 and US400 were used to amplify yghG and introduce a 10-amino-acid myc tag at the C terminus of YghG when cloned into pBAD322C (to construct plasmid pBAD/yghG-myc). Overlapping PCR was used to create substitution mutations in the yghG ORF. A C25A mutation in YghG [YghG(C25A)] was generated by overlapping PCR using templates created by primer sets US459/US462 and US460/US461 for amplification. Likewise, amplification with templates created with primers US459/US464 and primers US460/US463 was used to create the YghG(A26D S27D) substitution mutation.
Construction of H10407 strains containing alternate promoters of the gspABα and gspC-Oα operons.In H10407 strain TGS49, the native promoters of the gspABα and gspC-Oα operons have been replaced with an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible Ptac promoter and an arabinose-inducible Pbad promoter, respectively. This strain was constructed by a set of overlapping PCRs to generate an FRT-cat-kan-FRT cassette flanked with Ptac and Pbad promoters on either end with 60 bp of sequence specific to gspAα on one end and gspCα on the other. The Ptac promoter from plasmid pMAL-p4X was amplified in succession by primer sets 337/339, 326/339, and 327/329 (Table 2) to generate a 390-bp sequence that included gspA sequence on the upstream end and FRT-cat-kan-FRT sequence on the downstream end. Likewise, the Pbad promoter was introduced upstream of gspC by a set of overlapping PCRs with primer sets 330/332, 330/333, and 330/334. A final amplification with primers 327 and 334 generated a 3,105-bp fragment that was used for lambda red recombination to construct the Ptac-gspABα and Pbad-gspC-Oα strains.
Anti-LT ELISA.Detection of LT in H10407 culture supernatant was performed by GM1 ganglioside enzyme-linked immunosorbent assay (ELISA) as previously described (74).
Production of antisera.Plasmids TS18 and TS19 that encode the periplasmic domains of GspDα (amino acids 26 to 349 [GspDα26–349]) and GspDβ (amino acids 39 to 364 [GspDβ39–364]), respectively, were transformed into BL21(DE3). Culture was grown in 800 ml of LB broth containing 1 mM IPTG for expression of N-terminal His6-tagged GspDα26–349 and GspDβ39–364. Cells were pelleted and resuspended in 10 ml of nitrilotriacetic acid (NTA) buffer A (10 mM imidazole, 10% glycerol, 20 mM Tris, 250 mM NaCl, pH 8.0) to which 100 μg of RNase, 40 μg of DNase, 100 μl of 0.5 M MgCl2, and 400 μl of protease inhibitor (EDTA-free tablets; Roche) were added. The cell suspension was lysed by passage through a French press at 16,000 lb/in2, and cell debris was pelleted by centrifugation at 12,000 × g for 30 min at 4°C. The entire supernatant was applied to a HisTrap high-performance (HP) 5-ml affinity column (GE Health Care, Baie d'Urfe, Canada) and eluted with a gradient of NTA buffer A and NTA buffer B (500 mM imidazole, 10% glycerol, 20 mM Tris, 250 mM NaCl, pH 8.0). Selected fractions were applied to a desalting column (HiPrep 26/10 desalting column; GE Health Care) and resuspended in 20 mM Tris, pH 8.0. The preparation was then applied to an anion exchange column (6-ml Resource Q; GE Health Care) and eluted with a gradient of buffer A (20 mM Tris, pH 8.0) and buffer B (20 mM Tris, 500 mM NaCl, pH 8.0). Purified GspDα26-349 and GspDβ39-364 were sent to Immunoprecise Laboratories (Victoria, Canada) for production of rabbit anti-GspDα26-349 and anti-GspDβ39-364 antibodies. A 250-μg aliquot of each protein in complete Freund's adjuvant was injected into female New Zealand White rabbits. Equal amounts of protein in incomplete Freund's adjuvant were injected at 4, 7, and 10 weeks after the initial injection, followed by serum collection 2 weeks later. The anti-GspDα26-349 serum was depleted of antibodies capable of recognizing GspDβ39-364 by passing a 1-ml aliquot of the serum through a column comprised of 3 mg of GspDβ39-364 covalently coupled to a 1-ml HiTrap N-hydroxysuccinimide (NHS)-activated HP column (GE Health Care) following the manufacturer's instructions.
Immunoblot analysis.GspDα and GspDβ multimeric and monomeric proteins were detected by immunoblot analysis of whole-cell and cell-fractionated samples. Samples were electrophoresed in a 3 to 8% SDS gradient PAGE gel (Bio-Rad) until an 84-kDa protein standard was approximately 1 cm from the bottom of the gel and then transferred to polyvinylidene difluoride (PVDF) membrane. Proteins were visualized using primary rabbit anti-GspDα or anti-GspDβ antiserum, a peroxidase-conjugated anti-rabbit IgG secondary antibody (Sigma, Oakville, Canada), and chemiluminescent substrate (GE Health Care) detected with Hyperfilm (GE Health Care). myc-tagged proteins were detected with anti-myc antibody (Cell Signaling) following separation of proteins by 16% Tris-Tricine gel electrophoresis (80) and transfer to PVDF membrane.
Membrane fractionation and analysis.Separation of inner and outer membranes by isopycnic sucrose density gradient centrifugation was performed as described previously (41, 44). Cells were grown in 100 ml of CAYE medium, centrifuged, and resuspended in 10 ml of 10 mM HEPES, pH 8.0, containing 0.2 mM phenylmethylsulfonyl fluoride (PMSF), 4 μg/ml DNase, 4 μg/ml RNase, and 400 μl protease inhibitor cocktail (Roche, Mississauga, Canada), and cells were broken by passage through a French pressure cell. Following breakage, EDTA was added to a concentration of 3 mM, and whole cells were pelleted by centrifugation at 6,000 × g 10 min at 4°C. Supernatant was applied to the top of sucrose gradient 0 (44) composed of 0.2 ml of 60% sucrose and 2.8 ml of 10% sucrose in 10 mM HEPES containing 0.2 mM PMSF and centrifuged at 40,000 rpm for 3 h at 4°C. Pelleted cell membranes located on top of the 60% sucrose bed were applied to sucrose gradient 1 containing 0.2 mM PMSF and the following concentrations of sucrose in 10 mM HEPES: 1 ml of 55%, 3 ml of 50%, 1.8 ml of 45%, 3 ml of 40%, 1.2 ml of 35%, and 1.2 ml of 30% sucrose. The resultant SG1 gradient was centrifuged at 39,000 rpm for 16 h at 4°C. The SG1 was routinely separated into 14 or 15 fractions and analyzed for separation of inner and outer membrane material. Fractions containing inner membrane vesicles were identified by assay of the activity of the inner membrane protein NADH oxidase. NADH oxidase activity in fractions was assayed by oxidation of NADH measured by the decrease in absorbance (340 nm) over time and quantitated as μmol/min/ml as previously described (63). Fractions containing outer membrane material were identified by Coomassie blue staining following SDS-PAGE to locate major outer membrane porin proteins OmpC and OmpF (13, 19).
Cross-linking analysis.A culture volume of 1 ml was washed three times with phosphate-buffered saline (PBS), and dithiobis(succinylpropionate) (DSP) cross-linker (Pierce, Rockford, IL) was added to 100 μl of cells at a final concentration of 0.05 mM and incubated at 25°C for 30 min. The cross-linker was saturated by addition of Tris-HCl to a concentration of 50 mM and incubated at 25°C for 15 min. Selected samples were extracted once with phenol as previously described (55).
Statistical analysis.An unpaired two-sided Student's t test was used for all statistical analysis. Values were considered significantly different at a P value of <0.05.
RESULTS
The ETEC H10407 genome encodes two T2SSs.In silico analysis of the H10407 genome indicated that this strain encodes two T2SSs that we have designated alpha (T2SSα) and beta (T2SSβ) (Fig. 1). The T2SSα is encoded within a divergent operon composed of a minor operon that consists of genes gspABα and a major operon that consists of genes gspCα-Oα. Both the larger and smaller T2SSα operons are identical in nucleotide sequence to those previously described in E. coli K-12 (28).
Genetic organization of T2SS operons and surrounding genes in representative strains of pathogenic and nonpathogenic E. coli. The ETEC H10407 genome (21) encodes two T2SSs termed alpha (yellow) and beta (red). The corresponding area of the genome that encodes the T2SSα and T2SSβ of H10407 is given for the following strains: LT-positive ETEC strains E24377A and B7A (73), E. coli strain K-12 (7), enterohemorrhagic strain (EHEC) O157:H7 (66), adherent-invasive (AIEC) strain LF82 (68), enteropathogenic (EPEC) strain O127:H6 (43), enteroaggregative (EAEC) strain 55989 (92), uropathogenic (UPEC) strain UTI89 (14), avian pathogenic (APEC) strain O1 (47), and extraintestinal pathogenic (ExPEC) strain IHE3034 (56). Note that the T2SS encoded on plasmid pO157 is also shown (blue) (9). Genes flanking the T2SS operons are shown (white).
The T2SSβ is encoded by the operon consisting of genes gspCβ-Mβ that are readily identifiable as encoding proteins of a T2SS due to the extensive amino acid homology of the proteins encoded by these genes with T2SS proteins of other species. Whereas most T2SS operons encoded in other species are transcribed from a promoter upstream of gspC (27), the T2SSβ operon is atypical as it includes three genes (yghJ, pppA, and yghG) upstream of gspCβ whose protein products have not been characterized (95). The first gene, yghJ, encodes a large lipoprotein of unknown function that has extensive homology to AcfD (accessory colonization factor D) of V. cholerae (67), whose function has also not been elucidated. The second gene, pppA, encodes a putative prepilin peptidase that likely functions in processing of T2SSβ pseudopilins. The third hypothetical open reading frame (ORF), yghG, encodes a small protein of unknown function.
In silico analysis of the genomes of several E. coli pathotypes revealed that the T2SSα and/or T2SSβ operons are prevalent in pathogenic E. coli strains. As shown in Fig. 1, the genomes of LT-positive ETEC strains E24377A and B7A (73) encode the T2SSβ operon but do not encode the T2SSα operon. The nonpathogenic E. coli strain K-12 encodes a complete T2SSα but has a large deletion in the operon encoding T2SSβ that encompasses genes gspD to gspK. Enterohemorrhagic E. coli (EHEC) strain O157:H7 does not encode T2SSα or T2SSβ but does encode a T2SS within plasmid pO157 that is similar to the K. oxytoca T2SS (31). Adherent-invasive E. coli (AIEC) strain LF82, enteropathogenic E. coli (EPEC) strain O127:H6, and enteroaggregative E. coli (EAEC) strain 55989 harbor the gspOα gene of the T2SSα system and a complete T2SSβ system. Uropathogenic E. coli (UPEC) strain UTI89, the avian pathogenic E. coli (APEC) strain O1, and extraintestinal pathogenic E. coli (ExPEC) strain IHE3034 are similar to H10407 in encoding complete T2SSα and T2SSβ systems. This analysis revealed that the T2SSα and T2SSβ systems are prevalent but variably present among E. coli pathotypes.
The T2SSβ is assembled in H10407 under standard laboratory conditions.Since H10407 encodes two T2SSs, we sought to determine the relative level of assembly of each system by immunoblot detection of GspDα and GspDβ secretin multimers when cultures were grown under conditions previously shown to induce secretion of LT (CAYE medium) (33) since, presumably, under these conditions expression of the T2SS(s) would be induced. To detect GspDα or GspDβ independently by immunoblotting, antibodies were raised against the putative periplasmic domains of GspDα (amino acids 26 to 349) or GspDβ (amino acids 39 to 364) (47% identity between GspDα and GspDβ protein fragments) since the periplasmic domain exhibits a considerable degree of heterogeneity in secretin proteins, presumably due to its involvement in substrate selection (see Fig. S1 in the supplemental material) (reviewed in reference 49). Anti-GspDβ antibody was verified to not cross-react with GspDα, whereas anti-GspDα did cross-react with GspDβ and therefore required further purification (as described in Materials and Methods) that rendered the anti-GspDα antibody incapable of recognizing GspDβ (see Fig. S2). Immunoblot analysis of whole-cell samples revealed that GspDα could not be detected in multimeric or monomeric form (Fig. 2A). Inability to detect the GspDα protein was not due to use of an insufficient anti-GspDα antibody since the antibody was capable of detecting the GspDα secretin and GspDα monomer in strains in which the gspABα and gspC-Oα promoters were replaced by arabinose-inducible or IPTG-inducible promoters in strain TGS49 (Ptac-gspABα or Pbad-gspC-Oα) (Fig. 2B). These results suggested that under standard laboratory conditions the T2SSα is not assembled. Attempts to identify a condition in which GspDα could be detected in wild-type H10407 were not successful since variation of temperature, osmolarity, and growth medium did not result in detection of GspDα (data not shown).
In wild-type ETEC strain H10407 the T2SSβ is assembled, whereas the T2SSα is not. (A) GspDα secretin multimer and monomer were not detected in whole-cell samples of wild-type H10407 (WT), gspDα, and gspDβ strains by immunoblotting with anti-GspDα antibody. (B) Replacement of gspABα and gspC-Oα promoters with inducible Ptac and Pbad promoters allowed detection of the GspDα secretin and monomer in strain TGS49 (Ptac-gspABα Pbad-gspC-Oα) as detected by anti-GspDα immunoblotting of whole-cell samples taken from wild-type H10407 (WT) and TGS49 cultures. The concentrations of arabinose (Ara) and IPTG used in cell culture for induction of Pbad and Ptac promoters, respectively, are given. (C) The GspDβ secretin (multimer) and monomer were observed by anti-GspDβ immunoblotting in whole-cell samples taken from wild-type (WT) H10407 and gspDα strains and not observed in a gspDβ mutant. The positions of GspDα and GspDβ secretin and monomer are labeled at the right side of the panel (where applicable), and the location of standard protein markers is given at the left side of each panel.
In contrast to T2SSα, T2SSβ is assembled and functional in ETEC under standard laboratory conditions since the GspDβ secretin multimer and monomer were detectable by immunoblotting with anti-GspDβ antibody (Fig. 2C) and since deletion of gspDβ prevented secretion of LT into culture supernatant (data not shown) as previously demonstrated (88). According to these data, although H10407 encodes two T2SSs, under standard laboratory conditions the T2SSβ is assembled and functional whereas the T2SSα is not.
Effect of yghJ, pppA, and yghG gene deletion on assembly of the GspDβ secretin and function of the T2SSβ.The T2SSβ operon is atypical in including genes yghJ, pppA, and yghG upstream of gspCβ (95). Since these genes are part of the T2SSβ operon, they likely encode structural components or substrates of the T2SSβ. Therefore, to more fully define the components of this system and the role these proteins perform in both assembly and function of T2SSβ, we constructed nonpolar deletion mutants of the yghJ, pppA, and yghG genes by use of the λRed recombination system and assessed the ability in these mutant strains to assemble the GspDβ secretin (indicative of T2SSβ assembly) and secrete LT (to assess the function of T2SSβ). As shown in Fig. 3, deletion of the yghJ gene had no effect on assembly of the GspDβ secretin or secretion of LT. Deletion of pppA also did not affect secretin assembly but did abrogate secretion of LT. This result suggested that PppA is the prepilin peptidase of T2SSβ, as expected based on its homology to prepilin peptidases and its ability to process the evolutionarily similar type IV prepilin subunits of Neisseria gonorrhoeae and K. oxytoca (29). Lastly, deletion of the hypothetical gene encoding YghG drastically decreased the amount of GspDβ secretin assembled and the amount of GspDβ monomer observed, accompanied by an inability to secrete LT.
Effect of deletion mutations of genes located upstream of the core gspCβ-Nβ genes of the T2SSβ operon on GspDβ secretin assembly and ability of the T2SSβ to secrete LT. (A) Immunoblot analysis of whole-cell samples of wild-type H10407 (WT), gspDβ, yghJ, pppA, and yghG strains with anti-GspDβ antibody. The positions of protein standard markers are given at the left side of the figure, and the positions of GspDβ secretin (multimer) and monomer are given at the right side of the figure. (B) The concentration of LT in supernatant assayed by ELISA from cultures of wild-type H10407 (WT), gspDβ, yghJ, pppA, yghG, and eltAB strains. Values are shown as the average amount of LT assayed in triplicate cultures for each strain. Significant difference (†, nonsignificant; *, P < 0.05) was calculated based on comparison of each value against that observed in the wild-type culture.
Expression of YghG and YghG-myc in trans complements secretin assembly and secretion of LT in the yghG strain.Deletion of yghG resulted in nearly complete lack of GspDβ in both multimeric (secretin) and monomeric forms (Fig. 3A), concomitant with abrogation of LT secretion. To confirm that the secretin assembly-negative and secretion-negative phenotypes resulted from the absence of YghG, YghG and a version of YghG that contains a 10-amino-acid myc sequence at the C-terminal end of the protein (to be used for immunodetection) were expressed in the yghG strain in trans, and the amount of secretin assembled and the concentration of LT in culture supernatant were examined. Expression of YghG or YghG-myc in trans in the yghG strain resulted in reestablishment of GspDβ secretin and monomer levels to those of the wild type in a gradient manner such that the amount of both multimer and monomer observed increased with the amount of arabinose used to induce expression of yghG and yghG-myc carried by the plasmids (Fig. 4A). Similarly, secretion of LT was reestablished to wild-type levels upon expression of YghG and YghG-myc in the yghG strain (Fig. 4B). These results verified that the secretin assembly-negative and secretion-negative phenotypes of the yghG strain are due to the lack of YghG. In addition, these results also indicated that the C-terminal myc tag of YghG-myc does not interfere with the function of YghG.
The secretin assembly-negative and LT secretion-negative phenotypes of the yghG strain can be complemented by expression of YghG and YghG-myc in trans. (A) Immunoblot analysis using anti-GspDβ antibody of whole-cell samples of the wild-type H10407 (WT) and yghG strains, a yghG strain complemented with pBAD/yghG (yghG::pBAD/yghG), and a yghG strain complemented with pBAD/yghG-myc (yghG::pBAD/yghG-myc). The concentrations of arabinose used for induction of yghG and yghG-myc expression are shown. The positions of the GspDβ secretin multimer and monomer are given at the right side of the panel. (B) The concentration of LT assayed by ELISA in supernatant from cultures of wild-type H10407 (WT), yghG, yghG::pBAD/yghG, and yghG::pBAD/yghG-myc strains containing various concentrations of arabinose (as shown at the bottom of the figure). Values are shown as the average of triplicate assays for each strain. Significant difference (†, nonsignificant; *, P < 0.05) was calculated based on comparison of each value against that observed in wild-type culture.
Evidence that YghG is an outer membrane lipoprotein.The secretin assembly-negative and LT secretion-negative phenotypes of the yghG strain suggested that YghG could be involved in assembly of the GspDβ secretin multimer in a manner similar to the PulS and OutS family of outer membrane lipoproteins that are required for assembly of the T2SS secretin in the outer membrane of K. pneumoniae and D. dadantii species, respectively (36, 61, 83). Analysis of the N-terminal sequence of YghG revealed that, like PulS and OutS of the T2SS and other outer membrane lipoproteins required for secretin assembly including YscW of the Y. enterocolitica T3SS and MxiM of the Shigella flexneri T3SS, YghG is likely a lipoprotein because it contains a lipoprotein signal sequence comprised of a lipobox motif with the consensus sequence L-A/S-G/A-C (79) (Fig. 5A). Maturation of lipoproteins occurs at the outer leaflet of the inner membrane by addition of diacylglycerol to the Cys residue of the lipobox motif by phosphatidylglycerol/prolipoprotein diacylglyceryl transferase (Lgt), cleavage of the signal peptide by signal peptidase II (LspA), and aminoacylation of the Cys residue by phospholipid/apolipoprotein transacylase (Lnt). Therefore, to further investigate the possibility that YghG is a lipoprotein, a substitution mutation was constructed to replace Cys 25 of the lipobox motif with alanine. This substitution would be expected to prevent processing of the preprotein by signal peptidase II due to the inability of LspA to recognize the unlipidated protein. As shown in Fig. 5B, anti-myc immunoblotting of whole-cell samples of the yghG strain expressing YghG(C25A)-myc in trans showed that the vast majority of YghG(C25A)-myc remains as a preprotein presumably due to the inability of LspA to cleave the signal sequence. In addition, high induction levels of YghG(C25A)-myc expression (0.5% arabinose) conferred a growth-defective phenotype, possibly due to sequestration of LspA complexes with the mutant protein (data not shown).
Evidence that YghG is an outer membrane lipoprotein. (A) Alignment of the N-terminal region of YghG with other lipoproteins known to be involved in secretin assembly including PulS from K. pneumoniae (P20440), OutS of Dickeya dadantii (Q01567), YscW of Y. enterocolitica (Q56851), and MxiM of Shigella flexneri (P0A1X2). The location of the signal sequence is indicated, with the lipobox (L-A/S-G/A-C) highlighted in gray. The locations of the substitution mutations C25A, A26D, and S27D in YghG are also indicated. (B) Myc-tagged proteins present in whole-cell samples of wild-type H10407 (lane 1), yghG (lane 2), yghG::pBAD/yghG-myc (lane 3), yghG::pBAD/yghG(C25A)-myc (lane 4), and yghG::pBAD/yghG(A26D S27D)-myc (lane 5) strains are detected by anti-myc immunoblotting. The location of the unprocessed YghG C25A variant that retains the signal peptide is indicated by an asterisk. (C) YghG-myc is localized to the outer membrane and YghG(A26D S27D)-myc is localized to the inner membrane when the protein is expressed in the yghG strain. YghG-myc and YghG(A26D S27D)-myc were detected by anti-myc immunoblotting of inner and outer membrane fractions (1 to 14) isolated by sucrose density gradient fractionation. Fractions containing inner membrane material were identified by assay of the inner membrane protein NADH oxidase, whereas fractions that contained outer membrane material were visualized by detection of outer membrane porin proteins OmpC and OmpF. The concentration of sucrose in fractions isolated from fractionation of a cell envelope preparation of the yghG::pBAD/yghG-myc strain is given.
The amino acid sequence of YghG suggested that the protein would be located in the inner leaflet of the outer membrane since residues +2 and +3 are not Asp residues (Fig. 5A). During maturation, lipoproteins are sorted to the inner or outer membrane according to the lipoprotein sorting signal located at positions 2 and 3 of the mature protein. In E. coli, Asp residues at positions 2 (94) and 3 (82) ensure retention of the lipoprotein in the inner membrane by evading the LolCDE complex responsible for lipoprotein transport to the outer membrane. Most other combinations of amino acids at positions 2 and 3 result in complex formation of the lipoprotein with LolCDE, transport across the periplasm by the chaperone LolA, interaction with the outer membrane lipoprotein LolB, and localization to the outer membrane via insertion of the N-terminal lipid modification in the outer membrane (91). To confirm that YghG is localized to the outer membrane, inner and outer membranes from a yghG strain expressing YghG-myc in trans were separated by sucrose gradient fractionation, and the location of YghG-myc was determined by immunoblotting (Fig. 5C). Although some YghG-myc was present in most fractions, including those of the inner membrane, the majority was observed in the intermediate and outer membrane fractions. In addition, replacing residues +2 and +3 of the mature lipoprotein with Asp residues (Fig. 5A) caused YghG(A26D S27D)-myc to strongly localize with the inner membrane (Fig. 5C). Taken together, the presence of the lipobox, the effect of the C25A substitution mutation on processing, and the effect of the +2 and +3 substitution mutations on localization all strongly suggest that YghG is an outer membrane lipoprotein.
YghG is a novel pilot protein.Since YghG is encoded within the T2SSβ operon, is required for assembly of the GspDβ secretin, and is mainly localized to the outer membrane, we hypothesized that YghG likely functions in a manner similar to the T2SS PulS/OutS family of pilot proteins in protection and localization of the secretin in the outer membrane. Interestingly, however, amino acid alignment of YghG with the T2SS proteins PulS, OutS, and EtpO revealed that YghG does not share significant sequence identity with these proteins, for example, displaying low sequence identity (16.9%), similarity (25.3%), and global alignment (25.5) scores with PulS (Fig. 6).
YghG does not exhibit significant amino acid sequence similarity with members of the PulS/OutS family of T2SS pilotin proteins. Full-length sequence alignment of members of the PulS/OutS family of proteins including PulS from K. pneumoniae (P20440), OutS of D. dadantii (Q01567), and EtpO from E. coli O157:H7 (O32577) with YghG of H10407 (Q8VPD0) are shown. Percent identity, percent similarity, and score were determined for each sequence in comparison to PulS by a global sequence alignment using the Needleman-Wunsch global alignment algorithm according to the program EMBOSS (European Bioinformatics Institute). Identical amino acids are highlighted in a black background, and similar amino acids are highlighted in a gray background.
Pilotin proteins function in localizing the secretin protein to the outer membrane; therefore, to determine if YghG is a pilotin, we analyzed the location of GspDβ by immunoblotting inner and outer membrane fractions when outer membrane-localized (YghG-myc) and inner membrane-localized [YghG(A26D S27D)-myc] versions of YghG were expressed in a yghG strain. As shown in Fig. 7, sucrose gradient fractionation separated cell envelope preparations into 14 fractions. Fractions that contained inner membrane vesicles (fractions 3 and 4) were identified by assay of the activity of the inner membrane protein NADH oxidase (Fig. 7A), whereas fractions that contained outer membrane material (fractions 11 and 12) were identified by detection of outer membrane porin proteins OmpC and OmpF in samples of the fractions separated by SDS-PAGE and visualized by Coomassie blue staining (Fig. 7B). It was evident from the protein profile shown in Fig. 7B that the intermediate fractions 5 to 10 contain a mixture of both inner and outer membrane material, with more inner membrane material present in fractions 5, 6, and 7 and mostly outer membrane material in fractions 8, 9, and 10. In the GspDβ immunoblot shown in Fig. 7C, it was observed that GspDβ in wild-type H10407 mostly fractionated with the outer membrane; however, there is a minor amount of GspDβ present in inner membrane fraction 4. In a yghG mutant, however, GspDβ fractionated strongly with the inner membrane fractions. Expression of YghG-myc in the yghG strain reestablishes localization of GspDβ to the outer membrane fractions, thereby modeling the situation observed in the wild type. When the inner membrane-localized YghG(A26D S27D) was expressed in a yghG strain, no effect was observed in comparison to the yghG strain. These results showed that GspDβ can localize to the outer membrane only in the presence of outer membrane-localized YghG. In the absence of YghG, GspDβ is detectable and present in the inner membrane fractions. These data indicated that the outer membrane-localized YghG is required for localization of GspDβ to the outer membrane.
Localization of GspDβ in the outer membrane requires expression of outer membrane-localized YghG. NADH oxidase activity (A) and (B) whole-protein profile of fractions isolated by sucrose gradient fractionation of a cell envelope preparation from wild-type H10407 are shown. The concentration of sucrose in each fraction is given. The locations of outer membrane proteins OmpC and OmpF are designated (*). (C) Immunoblot analysis of fractions isolated from wild-type H10407 (WT), yghG, yghG::pBAD/yghG-myc, and yghG::pBAD/yghG(A26D S27D)-myc strains with anti-GspDβ antibody. Only the top portion of each immunoblot that corresponds to the location of the secretin multimer is included due to the negligible amount of GspDβ monomer observed after French press cell breakage (refer to the Results section). Fractions that contain predominantly inner (fractions 3 and 4) or outer (fractions 11 and 12) membrane material are outlined.
The immunoblots shown in Fig. 7C are of the GspDβ secretin multimer only and do not include the area of the immunoblot that shows the GspDβ monomer. This is because we observed that French pressure cell breakage caused spontaneous GspDβ multimerization, and therefore immunoblot analysis detected nearly all GspDβ protein present as a multimeric complex only (data not shown). As a result, this cell fractionation data could be used to determine the location of the protein as present in the inner membrane, outer membrane, or neither but could not provide information regarding the relative amounts of secretin multimer and monomer in the cell. For investigation of the relative amount of GspDβ in monomeric and multimeric forms, we used immunoblotting of whole-cell samples, as described below. Also note that although GspDβ is barely detectable as monomer and multimer by immunoblotting of whole-cell samples in the yghG strain (Fig. 3), preparation of inner and outer membranes concentrates inner and outer membrane proteins considerably. As a result, the relatively minor amount of GspDβ present in the yghG strain was readily detectable by immunoblotting of inner and outer membrane fractions isolated from cell envelope preparations, as shown in Fig. 7.
The majority of GspDβ is degraded when not localized to the outer membrane.Comparison of the amount of GspDβ present in whole-cell samples in yghG strains expressing inner [YghG(A26D S27D)-myc] or outer (YghG-myc) membrane versions of YghG revealed that the amount of GspDβ observed in multimeric and monomeric forms is much greater when GspDβ is localized to the outer membrane than when it is present in the inner membrane (Fig. 8A). In addition, expression of the inner membrane version of YghG does not alter the small amounts of secretin multimer and monomer observed in the yghG strain. These results were not explained by a difference in amount of inner and outer membrane-localized YghG present in the cell since immunoblot detection revealed similar levels of each protein in its respective strain upon induction with arabinose (Fig. 8B). Consistent with the requirement for outer membrane-localized YghG in assembly of the secretin, LT was not detected in culture supernatant if YghG was localized to the inner membrane due to an absence of assembled secretin and thus functional T2SSβ (Fig. 8C).
Assembly of the secretin and protection from degradation require localization of GspDβ to the outer membrane. Immunoblotting was performed of whole-cell samples of the wild-type H10407 (WT) or yghG strain or a yghG strain expressing outer membrane-localized YghG (yghG::pBAD/yghG-myc strain) or inner membrane-localized YghG [yghG::pBAD/yghG(A26D S27D)-myc strain] with anti-GspDβ antibody (A) or anti-myc antibody (B). The concentration of arabinose used for induction of plasmid-encoded protein expression is shown at the top of the panel. The locations of GspDβ multimer and monomer are shown. (C) The concentration of LT assayed by ELISA in supernatant from cultures of wild-type H10407 (WT), yghG, yghG::pBAD/yghG-myc, and yghG::pBAD/yghG(A26D S27D)-myc strains containing various concentrations of arabinose (as shown at the bottom of the figure). Values are shown as the average of triplicate assays for each strain. Significant difference (†, nonsignificant; *, P < 0.05) was calculated based on comparison of each value against that observed in wild-type culture.
Localization of GspDβ to the outer membrane is required for secretin assembly.The results shown in Fig. 7 and 8 revealed that outer membrane localization of GspDβ is dependent upon YghG and that localization to the outer membrane is required for GspDβ secretin assembly since only in the wild-type H10407 strain or in the yghG strain complemented with YghG-myc is GspDβ present in the outer membrane fractions (Fig. 7) and detected as a multimer (Fig. 8). Therefore, to confirm that localization of GspDβ to the outer membrane is required for secretin assembly and to determine if the inability to detect secretin multimer when GspDβ is localized to the inner membrane could be due to degradation of an inner membrane secretin that is more prone to degradation during preparation of whole-cell samples, secretin multimers of GspDβ were cross-linked in vivo and analyzed following treatment with phenol. As shown in Fig. 9, addition of phenol to whole-cell samples of wild-type H10407 degrades the GspDβ secretin (Fig. 9, lane 1) into its monomeric components (Fig. 9, lane 2). However, addition of the cross-linker DSP to whole cells prior to treatment with phenol caused multiple bands that correspond to protomers of GspDβ (likely dimers and tetramers) to be observed (Fig. 9, lane 3). In the yghG strain these protomers of GspDβ were not observed (Fig. 9, lane 4). Expression of YghG-myc in the yghG strain reestablished detection of GspDβ protomers in a gradient fashion, whereby increased expression of YghG-myc (induced with arabinose) resulted in an increased amount of GspDβ protomers observed (Fig. 9, lanes 5 to 8). In the absence of outer membrane-localized YghG such as in the yghG strain (Fig. 9, lane 4) or when the inner membrane variant YghG(A26D S27D)-myc was expressed (Fig. 9, lanes 9 to 12), GspDβ protomers were not detected, and only GspDβ monomer was evident. These data therefore show that outer membrane localization of GspDβ is required for secretin formation and that the GspDβ that is present in the absence of outer membrane-localized YghG is not present in multimeric form; that is, GspDβ is unable to form a multimer when localized to the inner membrane.
Cross-linked protomers indicative of GspDβ secretin assembly are observed only when GspDβ is localized to the outer membrane. Immunoblot analysis with anti-GspDβ antibody of whole-cell samples of the wild-type H10407 (WT) or yghG strain or a yghG strain expressing outer membrane-localized YghG in trans (yghG::pBAD/yghG-myc strain) or inner membrane-localized YghG [yghG::pBAD/yghG(A26D S27D)-myc strain] in trans. Samples were treated with the cross-linker DSP and phenol as indicated. The locations of GspDβ secretin multimer and monomer are given; asterisks indicate the positions of GspDβ protomers. The positions of protein standard markers are given at the left side of the figure.
DISCUSSION
In numerous E. coli pathotypes the T2SS has been shown to be required for pathogenesis of human and animal hosts, thereby identifying this system as a major virulence factor in this species (1, 35, 38, 51). In silico analysis of the H10407 genome revealed that this ETEC strain encodes two T2SSs, designated alpha (T2SSα) and beta (T2SSβ) (Fig. 1), both of which have been identified as capable of secreting the major ETEC virulence factor LT, albeit by different mechanisms (39, 88). The T2SSα, when expressed in an hns (histone-like nucleoid structuring protein) strain of E. coli K-12, is capable of secreting LT (the eltAB genes that encode LT are carried by a plasmid) that remains associated with the outer membrane and becomes loaded onto outer membrane vesicles (39). The T2SSβ has been shown in ETEC strain H10407 to be responsible for secretion of LT into culture supernatant (88), a function that is required for in vivo colonization since deletion of gspMβ renders ETEC incapable of colonizing the mouse intestine (1).
The T2SSα operon of H10407 is identical at the nucleotide level to that of E. coli K-12, and as in the K-12 strain, the T2SSα was not assembled in H10407 under standard laboratory conditions (Fig. 2A), presumably due to repression of the gspABα and gspC-Oα promoters by HNS (28). Replacement of gspABα and gspC-Oα promoters with IPTG or arabinose-inducible promoters enabled detection of GspDα in both monomeric and multimeric forms (Fig. 2B). It is therefore most likely that the specific in vivo or environmental conditions required for T2SSα expression have not been duplicated in the laboratory. It is possible that a specific condition encountered during intestinal colonization by ETEC is required for T2SSα induction. However, although in addition to H10407 the ETEC strain TW10598 also contains both T2SS operons (analysis not shown), the absence of a T2SSα operon in ETEC strains E24377A and B7A (Fig. 1), in addition to strain TW10828 (analysis not shown), indicates that the T2SSα is not involved in LT secretion in at least some ETEC strains (Fig. 1). Genomic in silico analysis of representative strains of other E. coli pathotypes revealed that the presence of a complete T2SSα operon in the genome is not highly conserved since EHEC strain O157:H7, AIEC strain LF82, EPEC strain O127:H6, and EAEC strain 55989 do not encode this system, whereas UPEC strain UTI89, APEC strain O1 strains, and ExPEC strain IHE3034 do (Fig. 1).
The T2SSβ is assembled and functional in ETEC when bacteria are grown under laboratory conditions since GspDβ was readily detectable in both multimeric (secretin) and monomeric forms (Fig. 2C) and since deletion of gspDβ rendered the strain incapable of secreting soluble LT into culture supernatant, as previously described (88; also data not shown). In silico analysis revealed that T2SSβ is prevalent among other E. coli pathotypes because it is encoded in the genomes of ETEC strains E24377A and B7A, AIEC strain LF82, EPEC strain O127:H6, EAEC strain 55898, UPEC strain UTI89, APEC strain O1, and ExPEC strain IHE3034 (Fig. 1).
The cryptic nature of T2SSα expression in comparison to that of T2SSβ in ETEC is similar to that described in other species that encode multiple T2SSs such that one system is easily detectable as functional whereas the other is not. For instance, Pseudomonas aeruginosa encodes two T2SSs designated Xcp and Hxc. The Xcp system is expressed in standard laboratory medium (12, 75) and functional in secretion of a variety of proteins including elastase, lipase, and chitin-binding proteins, among others (16, 54). The Hxc (homologous to Xcp) system, however, is expressed only under phosphate-limiting conditions and functional in secretion of a single protein, the alkaline phosphatase protein LapA (4). In Xanthomonas campestris pv. vesicatoria, the Xps T2SS is involved in secretion of plant cell wall-degrading enzymes and required for virulence, whereas the Xcs system is not required for virulence and has not been identified as involved in secretion of any substrates (87). As a result, similar to the situation in other species that encode multiple T2SSs, expression of the ETEC T2SSα likely requires distinct environmental conditions, whereas T2SSβ expression may not.
The T2SSβ operon is atypical in comparison to T2SS operons of other species by including three genes (yghJ, pppA, and yghG) upstream of gspCβ (95) (Fig. 1). The functions of the proteins encoded by these genes have not been previously studied, and their presence suggested that additional uncharacterized factors could be involved in assembly and function of T2SSβ in comparison to other T2SSs. The yghJ gene encodes a large 1,520-amino-acid putative lipoprotein that is homologous to the accessory colonization factor D (AcfD) of V. cholerae, required for colonization of the mouse intestine (67) and resistance to the vibriocidal activity of anti-vibrio whole-cell antibodies and complement (64). Interestingly in a study by Moriel et al. (56), anti-YghJ antibodies were found to be highly immunoprotective in mice upon challenge with ExPEC strain IHE3034, and the T2SSβ was shown to be capable of secreting YghJ into culture supernatant. These data suggest that YghJ is a substrate of the T2SSβ and is not likely a structural component of it. The putative designation of YghJ as a secreted protein is consistent with the results described in this study since deletion of yghJ did not affect the assembly or function of T2SSβ (Fig. 3).
The T2SSβ pppA gene encodes a protein homolog of T2SS prepilin peptidases that are required for processing of pseudopilin subunits GspG, -H, -I, -J, and -K. Deletion of PppA did not affect assembly of the secretin multimer (Fig. 3A) but did prevent secretion of LT (Fig. 3B). Presumably, the pseudopilus was not assembled in the pppA mutant, thereby rendering the T2SSβ nonfunctional. In fact, the vestigial T2SSβ PppA of E. coli K-12 has been shown to be capable of processing the evolutionarily similar type IV prepilin subunits of Neisseria gonorrhoeae and K. oxytoca (29). This finding is not surprising, given that some species including P. aeruginosa and A. hydrophila encode one prepilin peptidase (GspO) that is required for processing of both T2SS pseudopilins and type IV pilins (65).
The third hypothetical open reading frame encoded upstream of gspCβ in the T2SSβ operon, yghG, encodes a 136-amino-acid protein likely localized to the inner leaflet of the outer membrane due to the presence of an N-terminal signal sequence, a lipobox domain, and amino acids that are not Asp residues at positions +2 and +3 of the mature lipoprotein (Fig. 5A). Immunoblot detection of YghG-myc in outer membrane fractions when the protein was expressed in a yghG strain (Fig. 5C) confirmed that YghG-myc is mainly localized to the outer membrane. Replacing the lipid-modified Cys residue of the mature YghG-myc lipoprotein with Ala (mutation C25A) prevented processing of the preprotein, likely due to the inability of signal peptidase II (LspA) (34) to cleave the signal peptide (Fig. 5B). Lastly, YghG could be localized to the inner membrane by replacing amino acids 2 and 3 of the mature lipoprotein with Asp residues (Fig. 5C), an alteration known to prevent detection of the protein by the LolCDE system required for transport of lipoproteins to the outer membrane. Together, these results strongly suggest that YghG is an outer membrane lipoprotein.
A nonpolar chromosomal deletion of yghG prevented assembly of the GspDβ secretin (Fig. 3A), thereby rendering the T2SSβ nonfunctional, and resulted in the inability to secrete LT (Fig. 3B). Confirmation that the secretin assembly-negative and secretion-negative phenotypes of the yghG strain are attributable to the lack of YghG was provided by complementation with YghG or YghG-myc expressed in trans (Fig. 4). The requirement for outer membrane-localized YghG in assembly of the GspDβ secretin suggested that YghG functions as a pilotin protein. Consistent with the function of pilotin proteins, YghG is required for localization of GspDβ to the outer membrane; in the absence of YghG, GspDβ was localized to the inner membrane (Fig. 7).
Mislocalization of a secretin protein due to the absence of its pilotin has been previously shown to result in multimeric complexes observed in the inner membrane (37). In the ETEC strain H10407, however, the vast majority of the GspDβ is degraded when not localized to the outer membrane, and the remainder does not form a secretin in the inner membrane in the absence of its pilotin YghG. In whole-cell samples of strains expressing the inner membrane or outer membrane versions of YghG (Fig. 8A), it is evident that the total amount of GspDβ present in secretin and monomeric forms was much greater when the outer membrane version of YghG was expressed (which localizes GspDβ to the outer membrane) (Fig. 7) than when the inner membrane version of YghG was expressed. Although it was shown that PulS was capable of preventing PulD degradation when it was localized to the inner membrane (36), expression of the inner membrane YghG(A26D S27D)-myc did not change the negligible amount of multimer and monomer GspDβ that was observed in the yghG mutant (Fig. 8A). This would suggest that localization of GspDβ to the outer membrane and not YghG per se is required to prevent degradation of GspDβ although we cannot be certain that the YghG mutant, despite being normally processed, is present in a state that would allow it to bind to the inner membrane GspDβ. In any case, these results are similar to those observed for the Yersinia enterocolitica T3SS pilot protein YscW, which does not prevent proteolytic degradation of YscC unassembled monomers prior to secretin assembly but is solely required for localization of YscC to the outer membrane (8).
GspDβ was not observed in multimeric form in the inner membrane in the absence of the pilotin YghG. As shown in Fig. 9, cross-linked protomers of GspDβ were observed only when GspDβ was localized to the outer membrane and were observed solely in monomeric form when GspDβ was not localized to the outer membrane. This situation is similar to the state of the type IV pilus secretin PilQ of Pseudomonas aeruginosa that in the absence of its pilotin PilF does not form the secretin in the inner membrane (48).
The mechanism by which GspDβ is localized to the outer membrane by YghG likely involves the Lol system that is required for transfer of lipoproteins to the outer membrane. A model proposed by Okon et al. (62) describes the putative mechanism used in localization of the T3SS secretin MxiD to the outer membrane by the pilotin protein MxiM in Shigella flexneri. This mechanism is based on structural changes that are observed upon binding of the “cracked” β-barrel-type MxiM to the C-terminal 18 residues of MxiD that presumably lead to formation of a ternary complex composed of LolA-MxiM-MxiD and subsequent localization of MxiD to the outer membrane. This model is intriguing because it incorporates the known function of LolA and LolB in localization of lipoproteins to the outer membrane to explain how the secretin protein is localized by a pilotin protein. Since YghG is an outer membrane lipoprotein that would require the function of the Lol system for localization in the outer membrane, YghG could localize GspDβ in the outer membrane by the same mechanism. In fact, in a recent study by Collin et al. (17), the K. oxytoca T2SS secretin PulD could not be localized in the outer membrane of E. coli if a mutant LolA protein that is incapable of lipoprotein transfer to LolB was expressed.
Secretin proteins exhibit significant sequence conservation, with exception of the C-terminal variable domain recognized by the pilotin protein (8, 22, 62). Due to the heterogeneous nature of the pilotin binding site of secretins, pilotin proteins themselves exhibit a large degree of heterogeneity, thereby making in silico identification of these proteins difficult. As a result, pilotin proteins are classified as such based on the similarity of the function of the protein and not solely on amino acid similarity. Consistent with the divergent nature of the amino acid sequences of pilotin proteins, comparison of the PulS/OutS family of T2SS pilotin proteins with YghG showed that YghG exhibits limited amino acid sequence identity with this family of proteins (Fig. 6). However, since YghG performs a pilotin function similar to PulS and OutS in being required for localization of their respective T2SS secretin proteins to the outer membrane, we suggest redesignation of the hypothetical protein YghG as the T2SSβ pilotin protein GspSβ.
In summary, we have demonstrated that the wild-type pathogenic ETEC isolate strain H10407 encodes two T2SSs (termed alpha and beta), each of which is variably present within the genome of E. coli pathotypes. Mutational and physiological analyses demonstrated that the previously hypothetical protein YghG encoded within the T2SSβ operon is a pilotin protein required for localization of GspDβ to the outer membrane, a function integral for assembly of the GspDβ secretin utilized by the T2SSβ to secrete the major ETEC virulence factor, the heat-labile enterotoxin.
ACKNOWLEDGMENTS
We thank James Fleckenstein for providing the H10407 eltAB strain and John Cronan for kindly providing plasmid pBAD322C.
This research was supported by a grant from the Natural Sciences and Engineering Research Council of Canada.
FOOTNOTES
- Received 1 January 2012.
- Returned for modification 5 March 2012.
- Accepted 6 May 2012.
- Accepted manuscript posted online 14 May 2012.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.06394-11.
- Copyright © 2012, American Society for Microbiology. All Rights Reserved.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.
- 21.↵
- 22.↵
- 23.
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.↵
- 97.↵
