Many pathogenic E. coli strains secrete virulence factors using type II secretory systems, homologs of which are widespread in Gram-negative bacteria. Recently, the enteropathogenic Escherichia coli strain E2348/69 was shown to secrete and surface-anchor SslE, a biofilm-promoting virulence factor, via a type II secretion system. Genes encoding SslE and its associated secretion system are conserved in some non-pathogenic E. coli, including the commonly-used W (Waksman) strain.
We report here that E. coli W uses its type II secretion system to export a cognate SslE protein. SslE secretion is temperature- and nutrient-dependent, being robust at 37°C in rich medium but strongly repressed by lower temperatures or nutrient limitation. Fusing either of two glycosyl hydrolases to the C-terminus of SslE prevented it from being secreted or surface-exposed. We screened mutations that inactivated the type II secretion system for stress-related phenotypes and found that inactivation of the secretion system conferred a modest increase in tolerance to high concentrations of urea. Additionally, we note that the genes encoding this secretion system are present at a hypervariable locus and have been independently lost or gained in different lineages of E. coli.
The non-pathogenic E. coli W strain shares the extracellular virulence factor SslE, and its associated secretory system, with pathogenic E. coli strains. The pattern of regulation of SslE secretion we observed suggests that SslE plays a role in colonization of mammalian hosts by non-pathogenic as well as pathogenic E. coli. Our work provides a non-pathogenic model system for the study of SslE secretion, and informs future research into the function of SslE during host colonization.
Keywords:Type II secretion; Surface display; Escherichia coli; Colonization factor
Gram-negative bacteria use diverse type II secretion systems (T2SS) to deliver a wide variety of proteins into the extracellular milieu [1,2]. Transport is effected by a membrane-spanning complex of 12–15 structural proteins, generically termed Gsp proteins (for
Secreted proteins serve many purposes, from electron transport to nutrient acquisition, and some are important pathogenicity factors for plant and animal pathogens in the Enterobacteraceae [5,6]. Type II secretion has been extensively studied in pathogenic strains of Escherichia coli, which collectively are known to use two distinct disease-promoting T2SS: the StcE secreting system encoded by the pO157 virulence plasmid , and the heat-labile enterotoxin (LT) secreting system common to many pathogenic strains . Recently the latter T2SS was shown for the first time to additionally secrete a non-LT protein, known as SslE, from the enteropathogenic strain E2348/69, thereby promoting biofilm maturation and rabbit colonization by E2348/69 [9,10]. The sslE gene sits immediately upstream of the T2SS-encoding secretory genes, and transcription of sslE and the gsp genes was shown to be co-regulated in E. coli strain H10407 . In E2348/69, SslE exists as a lipid-anchored, surface-exposed protein in the outer membrane and is also released into the culture supernatant. Strozen et al. termed the LT- and SslE-secreting system T2SSβ, to distinguish it from the chitinase-secreting T2SSα that co-occurs in several E. coli strains . Based on phylogenetic and structural analyses, Dunstan et al. recently determined that the E. coli T2SSβ is part of a larger group of T2SS that contain “Vibrio-type secretins”, making it a model for numerous type II secretion systems used to deliver toxic substrates by Vibrio and Escherichia species .
The SslE-secreting T2SSβ, unlike the StcE-secreting pO157 T2SS, is conserved in several non-pathogenic “safe” strains of E. coli (“safe” strains may colonize hosts, but have never been known to cause disease), including wild-type B and W isolates . To date, however, no report has described secretion of proteins by T2SSβ in any non-pathogenic strain. We were interested to determine whether non-pathogenic E. coli could also secrete the “virulence factor” SslE. Secretion of SslE by a safe strain would imply that SslE itself is not capable of promoting a disease state, and would invite comparisons of SslE function between pathogens and non-pathogens. Furthermore, if non-pathogenic E. coli could secrete SslE, the T2SSβ system could be studied using a non-pathogenic model organism.
We demonstrate here that the non-pathogenic E. coli strain W encodes a functional T2SSβ that secretes a cognate SslE protein. We found a strong effect of growth conditions on SslE secretion, which is relatively robust in rich medium at 37°C and undetectable when cells are cultured at 30°C or in minimal medium. Previous work suggested that the C-terminus of SslE might be a permissive site for sequence insertions with regards to T2SSβ recognition , but we found that C-terminal enzyme fusions to SslE blocked protein secretion and surface display.
As noted above, the T2SSβ was shown to promote mature biofilm formation in E. coli E2348/69. We searched for additional phenotypes in E. coli W by phenotypic microarray analysis of a mutant lacking T2SSβ-encoding genes on Biolog stress plates. The phenotypic microarray indicated a potential fitness effect of the mutation in high concentrations of urea. Using standard culture techniques, we found that deletion of T2SSβ-encoding genes, or the sslE gene, conferred a small survival advantage in medium containing high concentrations of urea.
Our findings make T2SSβ the only virulence-associated T2SS with shared functions in pathogenic and non-pathogenic E. coli. Considering our regulatory data and the clear homology between the T2SSβ-encoding operons of W and E2348/69, we propose that SslE is used by non-pathogenic as well as pathogenic strains of E. coli during host colonization.
E. coli W secretes SslE using T2SSβunder specific temperature and nutrient conditions
Prior to publication of the finished E. coli W genome sequence , a draft E. coli W genomic sequence generated by the U.S. Department of Energy Joint Genome Institute in collaboration with the Great Lakes Bioenergy Research Center (GenBank accession NZ_AEDF00000000) revealed the presence of the entire T2SSβ gene cluster, including a copy of the gene encoding SslE (see Figure 1 for a depiction of the locus). To determine whether E. coli W secreted endogenous SslE via T2SSβ, we analyzed the proteomes of the wild-type strain (WT) and a mutant lacking the genes encoding the conserved structural proteins of T2SSβ (ΔgspC-M). We grew strains in liquid culture, then harvested cells by centrifugation and compared the proteins present in cell lysates and cell-free supernatants (the latter containing any secreted proteins) by SDS-PAGE. We observed a ~180 kDa protein, the expected size for SslE, that was present in the supernatants of WT cultures but not Δgsp cultures (Figure 2A). The ~180 kDa protein band was absent from supernatants and cell extracts of a ΔsslE strain, but reappeared when we complemented the sslE deletion with plasmid-encoded sslE (Figure 2B). To further confirm that SslE was secreted and did not play an intracellular role in activating protein secretion, we attempted to complement the ΔsslE strain with a form of SslE lacking the Sec signal peptide (SslE-SP). Unlike wild-type SslE, SslE-SP could not complement the secretory defect in the ΔsslE strain. Taken together, our data demonstrate that SslE is secreted from wild-type E. coli W by T2SSβ.
Figure 1. Distribution of T2SSα and T2SSβ in non-pathogenic E. coli strains. Phylogeny is from Archer et al. , with O157:H7 as an outgroup lacking both T2SSα and T2SSβ. Loci encoding the two T2SS types (where present) are diagrammed for each strain. Branch lengths are arbitrary. T2SSαgsp genes are colored yellow, and T2SSβgsp genes are shown in red.
Figure 2. E. coli W secretes SslE using T2SSβ in a condition-dependent manner. All lanes are labeled by sample type: C = cell lysate, S = culture supernatant, M = molecular weight standards. A. Lysates and concentrated cell-free supernatants of wild-type and Δgsp strains showing SslE secretion by T2SSβ. B. Complementation of the ΔsslE mutation: WT = wild-type, VOC = vector-only control, SslE-SP = SslE lacking an N-terminal signal peptide. C. Complementation of the ΔpppA mutation. D. Condition-dependence of SslE secretion labeled by temperature and growth medium. Sizes of molecular weight standards are shown to the side of each gel in kDa. The presence of secreted SslE is marked with black triangles.
Intracellular SslE did not appear abundant in wild-type E. coli W, even under conditions where secretion of SslE was detectable. We observed accumulation of SslE in the cell when SslE was expressed from a multicopy plasmid, however. We postulate that in wild-type cells, the intracellular concentration of SslE is maintained at a relatively low level, and that SslE release from cells over time results in accumulation in the supernatant.
Type II secretion systems require prepilin peptidases to produce the mature, functional forms of their prepilin proteins , and the prepilin peptidase PppA is required for secretion of LT by T2SSβ in E. coli H10407 . To determine whether PppA is similarly required for SslE secretion by E. coli W, we compared SslE secretion in WT to a ΔpppA strain. SslE secretion was not detectable in the ΔpppA background, and the mutation could be complemented by plasmid-encoded PppA (Figure 2C). These results confirm that a fully-functional T2SSβ is required to secrete SslE, and indicate that expression of the gspC-M genes alone is not sufficient to allow SslE secretion.
We hypothesized that SslE secretion in E. coli W might play a role in host colonization, and that secretion might be regulated such that more SslE is secreted under conditions that resemble the mammalian gut. We assessed this conditionality by examining SslE secretion from cultures grown at different temperatures and nutrient conditions: 30°C vs. 37°C, and minimal MOPS-glycerol broth vs. rich LB (Figure 2D). We observed secretion of SslE only in cultures grown in LB at 37°C, indicating that either reduced temperature or nutrient limitations are sufficient to block SslE secretion.
C-terminal fusions to SslE prevent secretion
In their initial characterization of SslE surface display and secretion, Baldi et al. found that C-terminal fusion of a small tetracysteine-containing motif to SslE did not interfere with localization of SslE . This result suggested that the C-terminus of SslE might not be important for the recognition of SslE by T2SSβ, and thus might be a permissive site for polypeptide fusions. We were interested in testing C-terminal permissiveness for two reasons: first, because it might provide information about the targeting of SslE for secretion (as there are no defined secretory signals for type II secretion substrates), and second, because SslE fusions might be useful to anchor other proteins to the cell surface. We therefore independently fused two plant cell wall degrading enzymes, Cel45A and Pel10A from Cellvibrio japonicus, to the C-terminus of E. coli W SslE and assessed the capacity of these fusion proteins to be secreted or displayed on the cell surface. Both fusions resulted in stable, enzymatically active proteins when expressed in E. coli W. We did not generate fusions to the potentially lipidated N-terminus of SslE to avoid changes in lipidation that could affect protein localization.
We performed all secretion and display experiments side-by-side in wild-type and T2SS-deficient ΔpppA strains, and present the results in Table 1. By following activity of the enzymatic fusions, we found that neither fusion protein was released into the medium under conditions in which we found wild-type SslE to be released. Indeed, extracellular activity of SslE-Cel45A was difficult to detect, though lysed cells released highly active enzyme. Because the substrates for Cel45A (carboxymethyl cellulose) and Pel10A (polygalacturonic acid) are high molecular weight polysaccharides that cannot enter the E. coli cell, we were able to assess surface display of fusion proteins by measuring the enzymatic activity of intact cells as compared to cell lysates. These experiments further demonstrated that the fusion proteins were not displayed on the surface of the cell, but accumulated intracellularly.
Table 1. Extracellular and surface-displayed activity of SslE-Cel45A and SslE-Pel10A from liquid cultures
Inactivation of T2SSβ modestly increases urea tolerance
Baldi et al. demonstrated that inactivation of T2SSβ in E. coli E2348/69 inhibited biofilm maturation in confocal microscopic analysis of flow cell cultures, though it had no effect on early biofilm development in stationary plate assays . To uncover other phenotypes related to T2SSβ disruption, we used E. coli W as a non-pathogenic model system in a partial Biolog phenotypic microarray to compare wild-type and Δgsp strains grown with various stressors. The Biolog dye-reduction traces are presented in Additional file 1. Under most conditions the two strains were indistinguishable, but the screen indicated that elevated urea concentrations might differentially affect their growth. We examined this phenomenon in 96-well plate growth experiments under conditions in which our data showed SslE to be secreted (LB at 37°C). Compared to the wild-type control, Δgsp and ΔpppA strains maintained higher stationary-phase densities in the presence of 0.90 M and 1.15 M urea (Additional file 2: Figure S1), suggesting that inactivation of the T2SSβ system modestly increased urea tolerance even when the structural Gsp proteins were still expressed. We determined the role of SslE in this phenotype and verified modest urea tolerance by following the growth and viability of wild-type, Δgsp, and ΔsslE strains for 48 hours with or without 1.15 M urea under the standard culture conditions we used for SslE secretion experiments (in culture tubes on a rolling wheel for vigorous aeration). Culture absorbance readings and viable cell counts indicated that, without urea, the three strains grew equivalently up to 12 hours and slowly lost viability between 12 and 48 hours, with indistinguishable final viable counts at 48 hours (Figure 3 and Table 2). In the presence of 1.15 M urea all strains grew poorly, but Δgsp and ΔsslE strains maintained higher turbidity and viable cell counts than wild-type, with both mutants having > 60% more surviving cells than wild-type at 48 hours. We conclude that the inability to secrete SslE confers a small survival advantage in the presence of high concentrations of urea.
Format: PDF Size: 345KB Download file
This file can be viewed with: Adobe Acrobat Reader
Format: DOC Size: 43KB Download file
This file can be viewed with: Microsoft Word Viewer
Figure 3. Growth of wild-type and mutants lacking gsp genes or sslE with and without urea. A representative growth curve is shown for each strain grown under the conditions noted.
Table 2. Viable cell counts for cultures grown with and without urea
Discussion and conclusions
Strains within the species Escherichia coli encode different combinations of type II secretion systems, each of which secrete different effectors and presumably provide specific advantageous phenotypes to their host organisms. To this point, the only T2SS shown to be functional in non-pathogenic E. coli strains is the chitinase-secreting T2SSα, which is the sole T2SS encoded by E. coli K-12 [13,14] and whose role in natural environments is unknown. We demonstrate here that, surprisingly, the T2SSβ that promotes virulence of the enterotoxic strain H10407 and the enteropathogenic strain E2348/69 is conserved, and secretes a virulence factor homolog, in the non-pathogenic E. coli W strain. To our knowledge, this is the first time a virulence-associated type II secretion system has been shown to function in non-pathogenic E. coli. Deletion of sslE could be complemented in trans, indicating that an sslE disruption does not prevent expression or assembly of T2SSβ in E. coli W. We observed that E. coli W preferentially secretes SslE under nutrient-rich conditions at human body temperature (37°C), which suggests that SslE may be a colonization factor in non-pathogenic strains. The regulation of SslE secretion in other strains is unclear, but expression of genes encoding the LT-secreting T2SSβ in E. coli H10407 was also shown to be upregulated at host-associated temperatures . We hope that future experiments will elucidate the role of SslE in host colonization by non-pathogenic E. coli.
If secretion of SslE indeed aids diverse E. coli in gut colonization, it is perhaps surprising that some gut-derived isolates of E. coli, such as K-12 and O157:H7, lack the T2SS responsible for SslE secretion. Such strains may compensate for the loss of biofilm-forming propensity using other mechanisms; strains bearing the F plasmid (such as wild-type K-12) may rely on F pilus-mediated aggregation , for example. The genes encoding the SslE-secreting T2SSβ are present adjacent to the pheV tRNA gene, which appears to be a hypervariable locus in E. coli[16-18], so they may be randomly lost at a relatively high rate. Indeed, a comparison between phylogeny and T2SSα/T2SSβ presence suggests independent losses of T2SSβ in non-pathogenic strains (Figure 1). Notably, B and W encode the complete T2SSβ, while Crook’s and K-12 do not, in spite of the fact that Crook’s diverged from K-12 prior to the divergence of B. This indicates that either Crook’s and K-12 lost the T2SSβ-encoding genes independently, or that an ancestor of Crook’s, B, and K-12 lost the genes, which were subsequently re-acquired by strain B. An examination of the T2SSβ-encoding loci in Crook’s and K-12 strongly supports the former explanation. In K-12, the T2SSβ-encoding gsp operon clearly experienced an internal deletion that removed the gspD-Kβ genes, inactivating the T2SS. In Crook’s, however, the homologous genomic locus appears entirely different: all gsp genes are absent, and in their place is the fec operon (encoding a ferric citrate transport system) and a variety of putative ORFs. We infer that the most parsimonious explanation of the phylogenetic distribution of T2SSβ is that K-12 and Crook’s both lost the T2SS at different points in their evolutionary histories. It remains an open question what pattern of gene gains and losses best explains the distribution of T2SSβ across the diversity of E. coli strains not considered in our analysis.
It is of interest to note that a non-polar deletion of the pppA gene, encoding a prepilin peptidase, prevents secretion of SslE by E. coli W. This result agrees with a similar experiment performed by Strozen et al. to assess effects of PppA on LT secretion in H10407 . Both W and H10407 also encode a second prepilin peptidase (GspO) whose homolog is functional in facilitating ChiA secretion via T2SSα in K-12 . Whether the GspO peptidase is not expressed under conditions associated with SslE secretion in both W and H10407, or whether the two peptidases display different substrate specificities, remains to be determined.
Strikingly, in the presence of the otherwise intact gsp operon, deletion of sslE was effective in promoting modest urea tolerance. When we first observed the urea-tolerant phenotype of the Δgsp strain, we hypothesized that the mutant’s advantage stemmed from lacking the transmembrane components of the T2SS, particularly the secretin pore in the outer membrane, which might be denatured by urea. The urea tolerance of the ΔsslE mutant rules out this hypothesis, however, and indicates that secretion of SslE by T2SSβ renders cells modestly more sensitive to urea. Relative urea sensitivity is likely due to indirect effects on cell physiology of bearing surface-displayed SslE or of releasing of SslE into the culture medium.
We report here that enzymatic fusions to the C-terminus of SslE interfere with its targeting to the T2SS, as measured by release of fusion proteins and by display of fusion proteins on the outer leaflet of the outer membrane. Previously, Baldi et al. fused a tetracysteine motif to the C-terminus of E2348/69 SslE and saw that the fusion protein was still displayed on the cell surface . We do not think these results contradict ours, due to the significant structural differences between the fusion proteins in question. We propose that the six amino acids appended to the C-terminus of SslE in the study by Baldi et al. did not affect secretion of SslE, but that our fusions of SslE to large tightly-folded proteins (plant cell wall degrading enzymes from Cellvibrio japonicus) occluded important targeting motifs recognized by the T2SS. The uncharacterized nature of T2SS recognition of substrates  unfortunately limits our ability to speculate further as to what these motifs might be. Future dissection of the SslE protein with internal deletions and protein fusions may yield new insights into the targeting motif(s) of SslE, and determine whether SslE fusions can be used in the surface display of other proteins.
Growth media, strains and plasmids
E. coli strains and plasmids used in this study are summarized in Table 3, and sequences of the plasmids are provided in Additional file 3. The rich (LB) and minimal (Neidhardt MOPS minimal with 0.2% glycerol) media [21,22] contained supplements at the following concentrations: 25 μg/ml kanamycin, 100 μg/ml ampicillin, and 30 μg/ml chloramphenicol. Mutant strains were constructed by replacing various loci with a FRT-kan-FRT cassette via the λ Red method, and kan cassettes were then removed by FLP excision as described [23,24]. The FRT-kan-FRT cassette used for gene disruptions of gspC-M, pppA, and sslE was amplified from Keio mutant genomic DNA  using the primer pairs noted in Table 4. To ensure our phenotypes did not result from second-site mutations, we generated all mutant strains twice in parallel and performed assays with two independent isolates, which behaved similarly in all cases.
Table 3. Strains and plasmids used in this study
Format: ZIP Size: 95KB Download file
Table 4. Primers used in this study
All synthetic DNA for plasmid constructions described below was provided by Geneart (Regensburg, Germany). Plasmid pRH21 was constructed from pEG100  by replacing the multiple cloning site (MCS) with a synthetic variant including a tandem His-FLAG tag and by adding the rrnB-derived terminators from pTrc99A downstream of the MCS. pRH31 was constructed from pTrc99A  by replacing the MCS with the same synthetic variant as in pRH21. pMSD6 was constructed using sslE amplified from E. coli W genomic DNA with the sslE-up and sslE-dn primers (Table 4). pMSD7 was constructed using sslE similarly amplified with the sslE-noSP-up and sslE-dn primers. pMSD8 was constructed using pppA similarly amplified with the pppA-up and pppA-dn primers. For construction of pMSD6, pMSD7, and pMSD8, the PCR products were digested with Acc65I and BamHI and ligated into the large Acc65I/BamHI fragment of pRH21.
For construction of pRH153, sslE was amplified from E. coli W genomic DNA using primers sslE-up and sslE-dn-nostop, and the PCR product was digested with Acc65I and BamHI. A gene encoding the mature form of Cel45A from Cellvibrio japonicus Ueda107 was synthesized and codon optimized for E. coli expression, then amplified using the cel45A-noSP-up and cel45A-dn primers, and the PCR product was digested with BamHI and HindIII. The two digested PCR products (sslE and cel45A) were ligated into the large Acc65I/HindIII fragment of pRH31. pRH154 was constructed as pRH153, with a synthetic gene encoding mature Pel10A from C. japonicus Ueda107 (with altered codon usage for expression in E. coli) being amplified using the pel10A-noSP-up and pel10A-dn primers prior to digestion and ligation.
Protein expression and detection
For assessing secretion of wild-type SslE, cultures of indicated strains (mutants were all kan-marked, except ΔpppA mutants, which were unmarked) were grown in liquid media (LB at 37°C unless otherwise noted) with aeration for 16–20 hours. For complementation of the ΔsslE mutation, gene expression from plasmids was induced with 1 μM isopropyl-β-D-galactopyranoside (IPTG). Cells were harvested by centrifugation and resuspended in SDS sample buffer (SSB)  according to the following formula: resuspension volume (in μl) = 100 × A600 × vol harvested (in ml). These concentrated cell lysates were diluted 1:100 in SSB for SDS-PAGE. Cell-free supernatants were concentrated ~10-fold by filtration using Centricon spin columns (Millipore, Billerica, MA, USA), and added to concentrated SSB for SDS-PAGE. Samples were separated on 4-12% SDS-polyacrylamide gels and stained with silver to visualize protein bands . SslE secretion experiments were repeated 2–4 times, and single representative gels are shown.
To produce the images in Figure 2, the stained gels were digitally photographed and gel images were enhanced using Adobe Photoshop software. Linear transformations (contrast and brightness adjustments) were applied to the images for clarity; such transformations were applied uniformly across any given gel image.
Fusion protein localization by enzyme activity
To measure secretion and surface display of SslE-enzyme fusions, cultures of WT and ΔpppA::FRT strains bearing the indicated plasmids were grown in LB at 37°C with aeration for 16–20 hours. Cells were harvested by centrifugation, and cell-free supernatants were removed; an aliquot of collected cells was removed and lysed using the PopCulture reagent from Novagen (Madison, WI, USA). Enzymatic activities associated with intact cells, lysed cells, and cell-free supernatants were then immediately measured. SslE-Cel45A activity was measured using the CRACC assay , and SslE-Pel10A activity was measured using the pectate lyase assay described by Collmer .
Phenotypic microarray experiments were performed using an OmniLog reader (Biolog, Hayward, CA, USA) as per the manufacturer’s instructions using plate types PM-9 and PM-10. Cultures were grown at 37°C for 48 hours, and respiration data were analyzed using the PM software provided with the OmniLog reader. Strains used were wild-type W and Δgsp::FRT (unmarked deletion of gspC-M).
To compare urea tolerances in 96-well plates, wild-type, Δgsp::FRT, and ΔpppA::FRT strains were cultured in 200 μl aliquots of LB containing 0, 0.9 M, or 1.15 M urea in 96-well plates (inoculated as 1:100 dilutions from LB overnight cultures). Plates were grown with shaking at 37°C in a Tecan M1000 plate reader (Durham, NC, USA). Growth and survival were followed by regular measurement of A595 for each culture.
To compare urea tolerances in glass culture tubes, wild-type, Δgsp::FRT, and ΔsslE::FRT strains were cultured in 8 ml volumes of LB containing no urea or 1.15 M urea on a rolling wheel at 37°C. Biological duplicate cultures of each strain were inoculated with 1:1000 dilutions from LB overnight cultures after verification that all overnight cultures grew to equivalent A600 turbidity readings. Turbidity in growing cultures was measured by reading A600 using a Spectronic 20D digital spectrophotometer; for cultures with high densities (A600 > 1.5), aliquots of the culture were diluted 1:10 or 1:20 prior to measurement of A600. Viable cells were enumerated by 10-fold serial dilution of cultures into sterile 0.9% NaCl followed by plating of dilutions on non-selective media and colony counting.
Availability of supporting data
Biolog cultivation data are included as Additional file 1. Data from microtiter plate growth experiments of cells under urea stress are included as in Additional file 2: Figure S1. The sequences of all plasmids described in this study are included as Additional file 3.
T2SS: Type II secretion systems; LT: Heat-labile enterotoxin; Gsp: General secretory pathway; WT: Wild-type; MCS: Multiple cloning site.
The authors declare no competing interests.
MD, RL, and RH designed experiments and contributed to writing the manuscript. MD and RH performed experiments and analyzed data. All authors read and approved the final manuscript.
We would like to thank David Keating for thoughtful discussions and critical review of the manuscript. This work was funded by the DOE Great Lakes Bioenergy Research Center (DOE BER Office of Science DE-FC02-07ER64494). Sequencing of E. coli W by the U.S. Department of Energy Joint Genome Institute is supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231.
Hobbs M, Mattick JS: Common components in the assembly of type 4 fimbriae, DNA transfer systems, filamentous phage and protein-secretion apparatus: a general system for the formation of surface-associated protein complexes.
Baldi DL, Higginson EE, Hocking DM, Praszkier J, Cavaliere R, James CE, Bennett-Wood V, Azzopardi KI, Turnbull L, Lithgow T, et al.: The type II secretion system and its ubiquitous lipoprotein substrate, SslE, are required for biofilm formation and virulence of enteropathogenic Escherichia coli.
Dunstan RA, Heinz E, Wijeyewickrema LC, Pike RN, Purcell AW, Evans TJ, Praszkier J, Robins-Browne RM, Strugnell RA, Korotkov KV, Lithgow T: Assembly of the type II secretion system such as found in Vibrio cholerae depends on the novel pilotin AspS.
Yang J, Baldi DL, Tauschek M, Strugnell RA, Robins-Browne RM: Transcriptional regulation of the yghJ-pppA-yghG-gspCDEFGHIJKLM cluster, encoding the type II secretion pathway in enterotoxigenic Escherichia coli.
Rumer L, Jores J, Kirsch P, Cavignac Y, Zehmke K, Wieler LH: Dissemination of pheU- and pheV-located genomic islands among enteropathogenic (EPEC) and enterohemorrhagic (EHEC) E. coli and their possible role in the horizontal transfer of the locus of enterocyte effacement (LEE).
Schneider G, Dobrindt U, Bruggemann H, Nagy G, Janke B, Blum-Oehler G, Buchrieser C, Gottschalk G, Emody L, Hacker J: The pathogenicity island-associated K15 capsule determinant exhibits a novel genetic structure and correlates with virulence in uropathogenic Escherichia coli strain 536.