Plasmids from Shiga Toxin-Producing Escherichia coli Strains withRare Enterohemolysin Gene (ehxA) Subtypes Reveal PathogenicityPotential and Display a Novel Evolutionary Path

Sandra C. Lorenz,a,b Steven R. Monday,a Maria Hoffmann,a Markus Fischer,b Julie A. Kasea

U.S. Food and Drug Administration, Center for Food Safety and Applied Nutrition, Division of Microbiology, College Park, Maryland, USAa; University of Hamburg,Hamburg School of Food Science, Institute of Food Chemistry, Hamburg, Germanyb

ABSTRACT

Most Shiga toxin-producing Escherichia coli (STEC) strains associated with severe disease, such as hemolytic-uremic syndrome(HUS), carry large enterohemolysin-encoding (ehxA) plasmids, e.g., pO157 and pO103, that contribute to STEC clinical manifes-tations. Six ehxA subtypes (A through F) exist that phylogenetically cluster into eae-positive (B, C, F), a mix of eae-positive (E)and eae-negative (A), and a third, more distantly related, cluster of eae-negative (D) STEC strains. While subtype B, C, and Fplasmids share a number of virulence traits that are distinct from those of subtype A, sequence data have not been available forsubtype D and E plasmids. Here, we determined and compared the genetic composition of four subtype D and two subtype Eplasmids to establish their evolutionary relatedness among ehxA subtypes and define their potential role in pathogenicity. Wefound that subtype D strains carry one exceptionally large plasmid (>200 kbp) that carries a variety of virulence genes that areassociated with enterotoxigenic and enterohemorrhagic E. coli, which, quite possibly, enables these strains to cause disease de-spite being food isolates. Our data offer further support for the hypothesis that this subtype D plasmid represents a novel viru-lence plasmid, sharing very few genetic features with other plasmids; we conclude that these plasmids have evolved from a differ-ent evolutionary lineage than the plasmids carrying the other ehxA subtypes. In contrast, the 50-kbp plasmids of subtype E(pO145), although isolated from HUS outbreak strains, carried only few virulence-associated determinants, suggesting that theclinical presentation of subtype E strains is largely a result of chromosomally encoded virulence factors.

IMPORTANCE

Bacterial plasmids are known to be key agents of change in microbial populations, promoting the dissemination of varioustraits, such as drug resistance and virulence. This study determined the genetic makeup of virulence plasmids from rare entero-hemolysin subtype D and E Shiga toxin-producing E. coli strains. We demonstrated that ehxA subtype D plasmids represent anovel E. coli virulence plasmid, and although subtype D plasmids were derived from nonclinical isolates, they encoded a varietyof virulence determinants that are associated with pathogenic E. coli. In contrast, subtype E plasmids, isolated from strains re-covered from severely ill patients, carry only a few virulence determinants. The results of this study reemphasize the plasticityand vast diversity among E. coli plasmids. This work demonstrates that, although E. coli strains of certain serogroups may notbe frequently associated with disease, they should not be underestimated in protecting human health and food safety.

Shiga toxin-producing Escherichia coli (STEC) strains of vari-ous serotypes can cause severe illnesses, such as hemorrhagic

colitis (HC) and hemolytic-uremic syndrome (HUS). While E.coli O157:H7 represents the most prevalent serotype associatedwith severe human illness, non-O157 STEC strains are of equalconcern (1–4). Many pathogenic E. coli strains have been shown toproduce at least one Shiga toxin (stx1 or stx2), and many possess apathogenicity island, the locus of enterocyte effacement (LEE) (5–10). LEE encodes various proteins, including the intimin adhesion(eae) protein that enables E. coli to attach and colonize the hostintestinal epithelial cells and induce effacement of the brush bor-der microvilli (11, 12). However, some LEE-negative STEC strainshave caused severe diseases, including HUS, that were indistin-guishable from those caused by LEE-positive STEC strains, such asO157:H7 (13, 14). Evidently, LEE-negative strains have acquiredother mechanisms that enable these atypical STEC isolates to in-duce diseases, only some of which have been identified, for exam-ple, the subtilase toxin, SubAB, that can induce cell death or theproduction of the flagellin responsible for the bacterial invasion ofepithelial cells (15–17). The vast genetic heterogeneity of patho-

genic STEC strains makes it particularly difficult to establish mo-lecular criteria that can definitely identify STEC strains as infec-tious E. coli strains. The identification of emerging E. colipathotypes, like the German E. coli O104:H4 (2011), is particularlychallenging before an outbreak occurs (14). Interestingly, manyLEE-positive and LEE-negative disease-associated STEC strains

Received 20 June 2016 Accepted 11 August 2016

Accepted manuscript posted online 19 August 2016

Citation Lorenz SC, Monday SR, Hoffmann M, Fischer M, Kase JA. 2016. Plasmidsfrom Shiga toxin-producing Escherichia coli strains with rare enterohemolysingene (ehxA) subtypes reveal pathogenicity potential and display a novelevolutionary path. Appl Environ Microbiol 82:6367– 6377.doi:10.1128/AEM.01839-16.

Editor: M. J. Pettinari, University of Buenos Aires

Address correspondence to Julie A. Kase, julie.kase@fda.hhs.gov.

Supplemental material for this article may be found at http://dx.doi.org/10.1128/AEM.01839-16.

Copyright © 2016, American Society for Microbiology. All Rights Reserved.

crossmark

November 2016 Volume 82 Number 21 aem.asm.org 6367Applied and Environmental Microbiology

on June 20, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

carry the plasmid-encoded enterohemolysin, ehxA. Although theprecise role of EhxA in E. coli pathogenicity has not been fullyelucidated, EhxA is commonly used as a phenotypic marker todetect STEC strains, due to its hemolytic activity as observed onwashed sheep blood agar (18–20). Furthermore, ehxA nucleotidesequences have been shown to cluster into two main groups thatcorrespond to eae-positive and eae-negative STEC strains, mostlikely due to the evolution of two different virulence plasmids(21). In fact, Newton and colleagues sequenced the large virulenceplasmid of STEC O113:H21 and confirmed that eae-negativestrains do indeed carry a plasmid that is quite different from thatof the eae-positive strain O157:H7. For instance, pO113 was 165kbp in size and much larger than the 92-kbp plasmid of O157:H7and carried virulence factors, such as the adherence protein Iha orSubAB, that were unique to pO113 (22).

Bacterial plasmids play a key role in driving virulence evolutionand promoting the dissemination of various traits, such as drugresistance and virulence. The ability to exchange genetic materialbetween strains through, i.e., self-transmissible plasmids, enablesbacteria to adapt to various environments and promotes the oc-currence of emerging pathogens (14, 23, 24). E. coli has beenshown to possess a variety of plasmid types, many of which havebeen associated with virulence (25, 26). In fact, large enterohemo-lysin-encoding plasmids are found in most STEC isolates, includ-ing E. coli O157:H7 and non-O157 STEC strains, such as O26:H11, O103:H2, O113:H21, and O145:H28, strains commonlyassociated with diarrheal disease and HUS (22, 27–30). To date,six ehxA subtypes have been identified using PCR in combinationwith restriction fragment length polymorphism (RFLP) analysis.These ehxA subtypes have been shown to phylogenetically clusterinto eae-positive (subtypes B, C, and F), a mix of eae-positive(subtype E) and eae-negative (subtype A), and a third, most dis-tantly related, cluster of eae-negative (subtype D) STEC strains(31, 32). The fact that subtype D strains segregate away from all ofthe other subtypes suggests that these STEC strains may carrysignificantly different plasmids than those previously identifiedand characterized. Moreover, our previous analyses of 435 E. colistrains isolated from animal, food, environmental, and clinical(human) sources identified only four ehxA subtype D strains,which were all food isolates (32). These particular STEC strains

had not been implicated in human disease, but the rarity of theseisotypes in this population further suggested that these may beunique E. coli strains. The ehxA subtype E was carried by 2 of the435 strains, both of which were clinical isolates associated withHUS. Given the rarity of ehxA subtypes D and E, our objective wasto sequence the large plasmid of these six STEC strains for use in acomparative analysis with currently available plasmid sequencingdata representing the other four ehxA subtypes. Results from suchscrutiny may provide insight into the evolution of STEC and mayalso reveal additional virulence or drug resistance determinantscarried on their plasmids.

MATERIALS AND METHODSBacterial strains. The bacterial strains used in this study are listed in Table1. Strains CFSAN004176 to CFSAN004181 have been described previ-ously as 03-3375, 05-3014, 06-00048, 08-00022, 09-00049, andUSMARC_GB_STEC_089, respectively (32).

Whole-genome extraction. Bacterial strains were grown aerobicallyfor 18 to 24 h on tryptic soy agar at 37°C. One colony was transferred into50 ml of tryptic soy broth (TSB) and incubated for another 18 to 24 h at37°C in a shaking incubator. Genomic DNA was extracted using theDNeasy blood and tissue kit (Qiagen Inc., Valencia, CA) according to themanufacturer’s recommendations for Gram-negative bacteria. In order toincrease DNA concentration and volume, each strain was extracted threetimes using 4 ml of the bacterial culture. Samples were eluted off the samecolumn twice using 30 �l of AE buffer; the elutions for all three extractswere combined afterwards. When using the automated QIAcube extrac-tion system (Qiagen Inc., Valencia, CA), 2 ml of the bacterial culture wasused. The remaining bacterial culture was stored at �80°C.

Plasmid isolation and purification. Plasmid DNA was prepared byalkaline lysis as described previously (33), with the following modifica-tions. Bacterial cells, enriched in TSB, were harvested by centrifugationusing the Sorvall RC 6 Plus centrifuge (Thermo Fisher Scientific Inc.,Waltham, MA). The pellet was resuspended in an 18-ml glucose-Tris-EDTA solution mixed with 2 ml of a 20 mg/ml lysozyme solution and wasincubated for 10 min at room temperature. The lysate was treated with a40-ml NaOH-EDTA solution, kept on ice for 5 min, mixed well with 20 mlof 3 M sodium acetate (NaOAc) (pH 5.2), and kept on ice for an addi-tional 15 min. After the addition of 10 ml H2O, the solution was centri-fuged at 10,400 � g for 30 min at 4°C. Plasmid DNA was precipitatedaccording to the protocol (33), but centrifugation was performed at 8,000rpm for 20 min at 4°C. The supernatant was discarded, and the remaining

TABLE 1 E. coli strains used in the study and metadata

Strain Serotypea stx1b stx2

b eaeb ehxA subtype Source DiagnosisAccessionno./reference

CFSAN004176 O145:H25 � � � E Human HUS This study, 32CFSAN004177 O145:H25* � � � E Human HUS This study, 32CFSAN004178 O36:H14 � � � D Alfalfa sprouts This study, 32CFSAN004179 O136:H16 � � � D Bagged lettuce This study, 32CFSAN004180 O168:HNT � � � D Lettuce This study, 32CFSAN004181 O168:H8 � � � D Ground beef This study, 32EH41 O113:H21 � � � A Human HUS AY258503 (22, 64)94-3024 O104:H21 � � � A Human HC NZ_CP009107.1 (85)O157 Sakai O157:H7 � � � B Human HUS AB011549 (86, 87)11368 O26:H11 � � � C Human Diarrhea AP010954 (28)11128 O111:HNM � � � C Human Diarrhea/bloody AP010963 (28)RM13514 O145:H28 � � � C Human HUS NZ_CP006028.1 (27)RM13516 O145:H28 � � � C Human HUS NZ_CP006263.1 (27)12009 O103:H2 � � � F Human Diarrhea/bloody AP010959 (28)a NM, nonmotile; NT, nontypeable; *, H-type identified in silico, initially reported as O145:H28.b �, present; �, absent.

Lorenz et al.

6368 aem.asm.org November 2016 Volume 82 Number 21Applied and Environmental Microbiology

on June 20, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

pellet was briefly air dried and resuspended in 5 ml of 50 mM Tris HCl(pH 8.0)-1 mM EDTA buffer; buffer was added to a final weight of 8 g forthe purpose of plasmid purification using the CsCl gradient centrifuga-tion method (33). Different from the protocol, 8.5 g CsCl and 0.2 ml of a10-mg/ml ethidium bromide solution were added and mixed into theDNA-containing solution prior to being transferred to an ultracentrifu-gation tube. Any weight discrepancies were adjusted by the addition of thebuffer used previously. The solution was overlaid with mineral oil untilthe tube was filled and sealed, and centrifugation was performed at 40,000rpm for 24 h at room temperature in a type 70.1 Ti rotor using a BeckmanCoulter Optima L-90K ultracentrifuge (Beckman Coulter Inc., Fullerton,CA). Following centrifugation, the lower DNA band, as visualized bylongwave UV light, was removed with an 18-gauge needle, extracted 3�(or until the solution was colorless) with an equal volume of water-satu-rated butanol, and dialyzed 3 times against 2 liters of purified H2O at 4°C.

The plasmid was concentrated by DNA precipitation using a 1:10 vol-ume of 3 M NaOAc (pH 5.2) and 0.6 volumes of isopropanol. DNA wascollected by centrifugation at 12,000 � g for 10 min at 4°C, and the pelletwas washed twice with 1 ml 70% ethanol. After centrifugation at 12,000 �g for 5 min, the ethanol was discarded and the pellet was resuspended in150 �l Tris-EDTA (TE) buffer. Quantity and purification were deter-mined by the electrophoresis of a 10-�l aliquot on a 0.60% Tris-borate-EDTA (TBE) agarose gel. If necessary, the remaining genomic DNAand/or RNA was removed using plasmid-safe ATP-dependent DNase(Epicentre, Madison, WI) and/or RNase A (20 mg/ml) (Life Technolo-gies, Grand Island, NY) per the manufacturer’s recommendations. Re-gardless of subsequent treatment, DNA underwent another phenol-chlo-roform-isoamyl alcohol (25:24:1) extraction and DNA precipitation, asdescribed earlier, to remove any residual proteins.

Sequencing and assembly. The plasmids were sequenced using third-generation, single-molecule, real-time (SMRT) DNA sequencing on thePacific Biosciences RS II sequencer (PacBio, Menlo Park, CA) as previ-ously described (34). SMRTbell 10-kbp template libraries were preparedaccording to the PacBio 10-kbp insert library protocol using the DNAtemplate prep kit 1.0. Template DNA was initially sheared to a target sizeof more than 10 kbp using Covaris g-Tubes (Covaris, Inc., Woburn, MA).Whole-genome libraries for strains CFSAN004177, CFSAN004178,CFSAN004180, and CFSAN004181 underwent an additional size-selectionstep using the BluePippin size-selection system (Sage Science, Inc., Beverly,MA) according to the PacBio 20-kbp template preparation protocol and theBluePippin user guide. Single 10-kbp libraries were sequenced usingP4/C2 chemistry kits on SMRT cells with a 180-min collection protocol.Sequence reads were analyzed by SMRT analysis 2.3.0 (PacBio, MenloPark, CA). De novo assembly of the 10-kbp continuous long-read data wasperformed using the PacBio hierarchical genome assembly process 3(HGAP 3.0)/Quiver software package (35). Plasmid DNAs of strainsCFSAN004176, CFSAN004178, and CFSAN004179 were sequenced onthe PacBio sequencing platform using the Genomics Resource Centerat the University of Maryland Institute of Genome Sciences in Balti-more, MD.

Plasmid closure and gene prediction/annotation. For plasmid clo-sure, dot plots were generated for each contig using the Gepard software(36) to identify overlapping, same-direction repeats at the contig ends. Toconfirm plasmid closure and ensure that the sequence did not carry arti-facts occurring as a consequence of repetitive sequences, a 20-kbp concat-enated sequence was generated that equally flanked the conjunctionpoint. Afterwards, the raw reads were mapped back to this artificial se-quence using SMRT analysis. Sequencing reads that covered the referenceconjunction point confirmed a circular contig. All PacBio-generated rawreads were once again mapped back to the closed, no-repeat-containingcontig, until a concordance consensus of 100% was achieved.

Sequences were annotated using the NCBI Prokaryotic Genome Au-tomatic Annotation Pipeline (PGAAP) (http://www.ncbi.nlm.nih.gov/genome/annotation_prok/) and the Bacterial Annotation System BASys(https://www.basys.ca/). Coding DNA sequences (CDSs) identified by

PGAAP and BASys were compared between the two annotation toolsregarding start and stop codons, as well as the annotated, predicted geneproducts. Moreover, start and stop codons, coding regions, and codingpotentials were further identified using GeneMarkS (37). Finally, we ver-ified each CDS and the predicted gene products by comparing the nucle-otide and translated protein sequences to publically available sequencedatabases using the NCBI nucleotide and protein Basic Local AlignmentSearch Tool (blastn and blastp) and UniProt BLAST tool (http://www.uniprot.org/). Protein sequences with more than 90% identity, coveringat least 90% of the matching protein, were considered a match. Proteinsequences with identities between 70% and 90% with at least 90% cover-age to the matching protein were considered a putative match (38, 39).Hypothetical proteins, proteins of unknown function, or protein prod-ucts that were assigned different protein functions by PGAAP and BASyswere further scrutinized using the Phyre2 server (40) and InterPro (http://www.ebi.ac.uk/InterProScan/) in order to define potential proteinfunctions and to identify protein domains. We reviewed pseudogenesthat were caused by nucleotide substitutions, insertions, and/or dele-tions within a CDS by remapping the sequencing raw reads to thepseudogene using CLC Genomics Workbench 7.5 (CLC bio, Boston,MA) to identify whether the cause was an actual mutation or an as-sembly error. Sequences were edited and updated in Sequin (http://www.ncbi.nlm.nih.gov/Sequin/).

Identification of ehxA subtypes and phylogenetic analysis. ehxAsubtypes of plasmids sequenced in this study were previously identified byPCR-RFLP (32, 41). ehxA subtypes of previously sequenced E. coli plas-mids used for comparative analysis in the current study (Table 1) wereidentified using the CLC Genomics Workbench restriction site analysistool. We used the TaqI restriction enzyme to create a virtual restrictionmap of the ehxA gene sequences obtained through the NCBI database andsequencing analysis performed herein. The resulting restriction sites andpatterns were compared to the six ehxA subtype reference sequences pre-viously published (31).

A phylogenetic tree was constructed using MEGA 6.06 with both theneighbor-joining (statistical method) and the maximum composite like-lihood methods (substitution method) to determine evolutionary dis-tances using 1,000 bootstrap replicates (42). Sequences were initiallyaligned in MEGA 6.06 using ClustalW 1.6 with default settings.

Accession number(s). The plasmid DNA sequences of strainsCFSAN004176 to CFSAN004181 were submitted to GenBank and havebeen assigned the following accession numbers: CP012492 andCP012493 (CFSAN004176), CP012495 and CP012496 (CFSAN004177),CP012498 (CFSAN004178), CP012499 (CFSAN004181), CP012500(CFSAN004180), and CP012501 (CFSAN004179). Strains showing morethan one accession number indicate multiple plasmids.

RESULTS AND DISCUSSIONGenetic features of ehxA subtype D plasmids and pathogenicpotential. Strains CFSAN004178, CFSAN004179, CFSAN004180,and CFSAN004181 each harbored one large virulence plasmid of213,847 bp (pCFSAN004178), 242,187 bp (pCFSAN004179),225,292 bp (pCFSAN004180), and 223,952 bp (pCFSAN004181)(Table 2). These plasmids shared 99% sequence identity, with var-ious degrees of coverage (70% to 99%), primarily due to differ-ences in plasmid size rather than sequence composition (Fig. 1A;see also Table S1 in the supplemental material).

In general, almost all virulence-associated E. coli plasmids fallinto the IncF incompatibility group and range in size from 50 to170 kbp (Table 2) (24, 26). While ehxA subtype D plasmids werefound to carry the replicons RepFIB and RepFII of the IncF plas-mids, these 210- to 240-kbp plasmids are exceptionally large. Shi-gella and enteroinvasive E. coli carry large 180- to 230-kbp inva-sion plasmids (pINV); however, none of the pINV characteristicgenetic features, such as transcriptional activators virB and

Analysis of Virulence Plasmids from STEC

November 2016 Volume 82 Number 21 aem.asm.org 6369Applied and Environmental Microbiology

on June 20, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

mxiEab or components of the needle complex mxiGHILMD, wereidentified on subtype D plasmids (26, 43, 44). In fact, the geneticcompositions of subtype D plasmids were found to be unlike anycharacterized E. coli plasmid available in GenBank; the bestBLAST match for pCFSAN004178 to pCFSAN004181 was sorbi-tol-fermenting (SF) E. coli O157:H� pSFO157 (GenBank acces-sion number AF401292.1), sharing 95% nucleotide sequenceidentity and 28%, 32%, 30%, and 31% coverage, respectively.

Subtype D plasmids exhibited a mosaic-like structure, encod-ing a large number of mostly remnant transposases (Fig. 2) and aunique combination of putative and known virulence factors(see Table S1 in the supplemental material). All subtype Dplasmids were found to carry multiple copies of the heat-stableenterotoxin gene sta1, and pCFSAN004179 to pCFSAN004181also carried the enteroaggregative heat-stable enterotoxin (EAST

1) gene astA (Fig. 1A), both of which are associated with the de-velopment of diarrhea (45–47). Furthermore, pCFSAN004180and pCFSAN004181 carried a gene encoding putative tempera-ture-sensitive hemagglutinin autotransporter tsh, which is a viru-lence factor commonly associated with avian pathogenic E. coli(APEC) (48, 49). However, numerous deletions within the pertac-tin-like passenger domain, involved in virulence, and the auto-transporter barrel domain were identified which likely render thisTsh-like homolog biologically inactive. Interestingly, both plas-mids carried a second gene belonging to the family of the serineprotease autotransporter of Enterobacteriaceae (SPATE) (see Ta-ble S1), an autotransporter family that is commonly associatedwith virulence (50–52). Unlike the Tsh-like homolog, this proteinappears to have sustained a single mutational event that elimi-nated only the pertactin-like passenger domain, while all of the

TABLE 2 Overview of the virulence plasmids representing the six ehxA subtypes

StrainSize(kbp)

No.CDS GC %

Virulence factors and other genetic featuresa

RepFII RepFIB sta1 astA cma/cba espI T6SS stcE K88 katP toxB espP Ecf subAB iha saa epeA

CFSAN004176 52.3 58 46.6 � � � (�) � � (�/�) � � � � � � � � � �CFSAN004177 52.3 58 46.6 � � � (�) � � (�/�) � � � � � � � � � �CFSAN004178 213.8 208 45.8 � � � (�) � � � � � � � � � � � � �CFSAN004179 242.2 261 47.0 � � � � � � � � � � � � � � (�) � �CFSAN004180 225.3 221 46.3 � � � � � � � � � � � � � � (�) � �CFSAN004181 224.0 232 46.3 � � � � � � � � � � � � � � � � �pO113_EH41b 165.5 155 49.6 � � � � � � � � � � � � � � � � �pO104_94-3024 161.4 158 49.5 � � � (�) � � � � � � � � � � � � �pO157 Sakai 92.7 92 47.6 � � � (�) � � (�) � � � � � � � (�) � �pO26_11368 85.2 93 47.5 � � � (�) � � (�/�) � � � � � � � (�) � �pO111_11128 77.7 90 50.0 � � � (�) � � (�) � � � � � � � (�) � �pO145_RM13514 87.1 95 47.6 � � � (�) � (�) � � � � � � � � (�) � �pO145_RM13516 98.1 114 49.7 � � � (�) � � � � � � � � � � (�) � �pO103_12009 75.5 90 49.1 � � � � � (�) (�) � � � � � � � � � �a �, present; �, absent; (�), present on the chromosome only; (�/�), partial T6SS encoded on the chromosome.b Whole-genome sequence not available on NCBI.

FIG 1 Comparative nucleotide sequence analysis of virulence plasmids representing the six ehxA subtypes. The color keys for each plasmid used are shown oneach side of the circular map with nucleotide sequence identity compared to the reference plasmid. The outer most ring shows genetic features found on thereference plasmid. (A) BLAST comparison of 12 plasmids representing the six ehxA subtypes against the ehxA subtype D plasmid of strain CFSAN004179. (B)BLAST comparison of 11 plasmids representing the six ehxA subtypes against ehxA subtype E plasmid of strain CFSAN004176. Circular maps were generatedusing BLAST Ring Image Generator (BRIG).

Lorenz et al.

6370 aem.asm.org November 2016 Volume 82 Number 21Applied and Environmental Microbiology

on June 20, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

other SPATE characteristic domains are retained and intact (53).This suggests that this protein may still be capable of forming abeta barrel in the outer membrane that, either individually or incombination with the peptidase activity present on this protein,acts to serve a function critical to cell survival.

Subtype D plasmids carried a 15-gene cluster related to thetype VI secretion system (T6SS) (Fig. 1A), a typically chromo-somally encoded secretion system found to be present in manypathogenic Gram-negative bacteria, including Vibrio cholerae,Salmonella enterica, and E. coli, but not in their avirulent counter-parts (54–57). While we found that subtype D plasmids en-coded most of the 13 proposed core components of a fullyfunctional T6SS, the hemolysin-coregulated protein, Hcp (54–58), appeared to be missing (Fig. 3A; see also Table S1 in thesupplemental material). Although this T6SS appears to be non-functional, it seems to be conserved across a variety of pathogenicE. coli strains, including APEC O2:K1 and the German HUSO104:H4 strain (Fig. 3A) (59), suggesting that it may haveretained a yet undetermined but critical biological function.Notably, both APEC O2:K1 and O104:H4, as well as strainsCFSAN004178, CFSAN004180, and CFSAN004181, carry a sec-ond, apparently complete T6SS on their chromosomes, with theT6SSs of CFSAN004180 and CFSAN004181 being nearly identicalto those of O157 Sakai, O103:H2, and the second T6SS inO104:H4 (Fig. 3B) (60). In fact, many pathogens have been foundto carry multiple T6SSs, and there is emerging evidence that var-ious T6SSs within a single strain may serve different functions,such as mediating virulence toward eukaryotic cells or bacterialcompetitors (61–63). In contrast, CFSAN004179 was found tocarry a single, chromosomally encoded hcp gene (60). Such “or-phan” hcp genes are frequently found scattered throughout thebacterial chromosome and have been shown to play an active rolein the T6SS, despite being encoded far outside the main T6SSlocus (62). Thus, the chromosomally encoded hcp found inCFSAN004179 may restore T6SS functionality. Finally, given theclose proximity of the T6SS loci to a number of transposases andthe distinctly different GC content of the T6SS (�35%) in com-

FIG 2 Circular map of virulence plasmid pO136 (CFSAN004179) generatedwith CGView Server 1.0 (http://stothard.afns.ualberta.ca/cgview_server/).Genes on the leading and lagging strand are shown in the outer two rings,respectively, with arrow direction indicating the direction of transcription. GCcontent is represented by the black ring with high GC content as the outermostportion. GC skew is shown on the innermost ring. Genes displayed are cate-gorized by color as shown on the bottom. Red includes known and putativevirulence factors; pink indicates methylase-, resolvase-, and endonuclease-encoding genes; and yellow includes replication-, maintenance-, and transfer-associated genes.

FIG 3 Comparative analysis of the T6SS loci found in various pathogenic E. coli strains. Conserved T6SS genes are represented by the same colors with whitearrows indicating genes unrelated to the T6SS and gray arrows indicating T6SS-associated genes with unknown function. Arrow direction is indicative of thedirection of transcription. The color keys shown on the bottom indicate the functional gene classes of the T6SS loci (59). (A) Comparison between T6SS locicarried on subtype D plasmids as represented by CFSAN004179 and the chromosomally encoded T6SS loci found in pathogenic strains O104:H4 and APECO2:K1. (B) Comparison between T6SS loci found on the chromosome of subtype D strains compared to O157 Sakai, O103:H2, and the second T6SS loci foundin O104:H4. The alignment was generated with Geneious 9.1.2 using MAFFT alignment with default settings.

Analysis of Virulence Plasmids from STEC

November 2016 Volume 82 Number 21 aem.asm.org 6371Applied and Environmental Microbiology

on June 20, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

parison to the overall GC content of the subtype D plasmids(�47%), it appears that this T6SS was acquired via horizontalgene transfer (Table 2; Fig. 2).

Colonization and ultimately the adherence of bacterial cells tothe host’s intestinal tract are essential steps in bacterial pathogen-esis. The LEE pathogenicity island is central to this process and hasbeen found in the majority of pathogens associated with severedisease (6, 7, 10). Although subtype D plasmids were derived fromeae-negative, nonclinical STEC strains, this detail does not elimi-nate the pathogenic potential of these strains, as LEE-negativeSTEC isolates also have been recovered from severely ill patients(13, 14). Generally, LEE-negative clinically isolated STEC strainshave been shown to carry other adhesins and colonization-con-tributing factors (CCFs), such as Saa (STEC autoagglutinating ad-hesin), Iha (IrgA homologue adhesin), EpeA (enterohemorrhagicE. coli [EHEC] plasmid-encoded autotransporter), and Lpf (longpolar fimbria), that compensate for the lack of LEE (15, 22, 64). Infact, we found that subtype D strains carried a variety of chromo-somally encoded fimbrial operons, including the lpf locus, andstrains CFSAN004179 and CFSAN004180 also carried iha (60).Furthermore, pCFSAN004179 carried a gene encoding a putativezinc metalloprotease StcE and an associated 13-gene cluster of atype II secretion system (T2SS), which is also present on pO157,pO145 (RM13516), and pO103 of eae-positive STEC strains(Fig. 1A). StcE, like EpeA, exhibits mucinolytic activity thatmediates increased intimate adherence by facilitating a closerinteraction between bacteria and host cells (65–67). Addition-ally, pCFSAN004179 to pCFSAN004181 were found to carry thevirulence-associated fimbrial adhesin K88 gene cluster (faeC-faeJ)(68–70) that has been shown to have the ability to adhere to hu-man intestinal cells (71, 72). However, the major subunit, FaeG, ofthe K88 adhesion apparatus shared only 32% protein sequenceidentity with the FaeG protein expressed by the prototypical K88gene cluster, and pCFSAN004179 carried the gene encoding theF41 antigen instead of FaeG (73). Despite the absence of the K88antigen FaeG, subtype D plasmids likely encode functional K88complexes comprised of antigenically different adhesins. Specifi-cally, it has been shown that the seven genes encoding the minorsubunits, ushers, and chaperones of the K88 and F41 systems shareextensive DNA sequence homology. Although, the major anti-genic subunits F41 and FaeG are less homologous (�30% proteinidentity), the K88 system expresses a functionally active fimbriaewhen the K88 antigen is exchanged for the F41 antigen (73, 74).

The adhesins and CCFs encoded on subtype D strains lead us toconclude that these strains may potentially be capable of coloniz-ing the human gut and, given the virulence factors noted on sub-type D plasmids (astA, sta1, T6SS, stcE) and the presence of Shigatoxin, may possibly cause disease. In fact, while CFSAN004178carries stx2g, CFSAN004179 carries stx1a and CFSAN004180 andCFSAN004181 carry the highly HUS-associated stx2a variant (7,14). Although these subtype D strains were isolated from foodsthat have been frequently implicated in foodborne E. coli out-breaks (leafy greens, ground beef), they belong to serogroups(O36, O136, and O168) that rarely have been recovered from illindividuals (3). It is possible that subtype D strains may causediarrhea, as they carry the diarrhea-associated enterotoxins Staand EAST1 (45–47, 75), but maybe not to the extent where med-ical care is sought or infections were treated without identificationof the causative agent. Perhaps, the adhesins and CCFs carried ontheir genomes are not efficient enough for Shiga toxin delivery,

thus preventing the occurrence of severe diseases, such as HUS.Importantly, the LEE pathogenicity island is encoded on a mobilegenetic element and has been shown to integrate into the tRNAgenes selC, pheU, and pheV (76), which all appear to be unoccu-pied in CFSAN004180 and CFSAN004181 while only the pheUgene in CFSAN004179 appears to be intact (data not shown). Thisopens the possibility of strains CFSAN004179 to CFSAN004181acquiring LEE via horizontal gene transfer, which may ultimatelyresult in emerging pathogens.

Subtype D plasmids encode a number of virulence factors thatare associated with APEC (tsh), enterotoxigenic E. coli (ETEC)(sta1, astA, K88), and EHEC (stcE) strains, suggesting that severalhorizontal gene transfers may have already occurred. In fact, allvirulence factors are located in relatively close proximity to rem-nant transposases; most (other than T6SS) have been acquiredupstream of the RepFIB locus in a region that tends to containmost of the acquired virulence genes (Fig. 2) (26). The reportedrarity of subtype D plasmids is surprising given that they all carryplasmid maintenance and stability genes and also possess the abil-ity to participate in conjugal transfer, carrying nearly a complete35-kbp F transfer region (see Table S1 in the supplemental mate-rial) (77). More importantly, the acquisition of these plasmidtypes may enhance or confer virulence to other bacteria. Thus,subtype D plasmids are probably more widespread than initiallyassumed, and we anticipate that with the rapid development ofnext-generation sequencing, more subtype D plasmids will beidentified.

Genetic features of ehxA subtype E plasmids and pathogenicpotential. Strains CFSAN004176 and CFSAN004177 harborednearly identical (99% sequence identity) 52,297-bp virulenceplasmids that shared the highest nucleotide sequence identity(98%) and coverage (43%) with the 194,170-bp plasmid of E. colipO26-Vir (GenBank accession number FJ386569.1). Unlike otherehxA subtypes, subtype E plasmids were found to carry the colicinB and M operons (Table 2), providing them with a selective ad-vantage over competing, colicin-sensitive E. coli strains (26). Sub-type E plasmids also carry a gene encoding the typically island-encoded EspI, a potentially virulence-associated E. coli-secretedprotease of the SPATE family (Fig. 4A). Although the physiologi-cal function of EspI remains to be determined (78), a virulence-associated function appears to be likely as (i) most SPATE havebeen found to contribute to the pathogenicity of strains expressingthem and (ii) EspI, like EspP, has been shown to have optimalactivities at conditions that resemble the human bloodstream,suggesting a function during human STEC infection (50–53). Fur-thermore, espI appears to be widely conserved across a variety of E.coli strains, including the chromosomes of clinical strainsO103:H2 (strain 12009) and O145:H28 (strain RM13514), asso-ciated with bloody diarrhea and HUS, respectively (27, 28). Inter-estingly, espI is also present on the chromosome of ehxA subtype Estrains, potentially providing a greater benefit in fitness, particu-larly under stress conditions when high levels of gene expressionmay be needed. The possession of multiple gene copies, as well asthe lack of other known virulence factors carried on plasmids ofthe other ehxA subtypes, may provide additional evidence for thevirulence-associated function of EspI, particularly since subtype Eplasmids were isolated from HUS outbreak strains and the viru-lence traits (i.e., toxB, espP, katP) encoded on the plasmids of theother ehxA subtypes have been shown to contribute to the clinicalpresentation of these pathovars (Table 2) (17, 52, 65, 66, 79, 80).

Lorenz et al.

6372 aem.asm.org November 2016 Volume 82 Number 21Applied and Environmental Microbiology

on June 20, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

Although subtype E plasmids were derived from clinical iso-lates that are associated with HUS, unexpectedly these plasmidsonly seem to confer rather low selective advantage to the bacteriacompared to other such plasmid types: (i) the only virulence as-sociated factor, EspI, was also present on the chromosome, and(ii) many strains were found to be resistant to colicins (26, 81).Furthermore, subtype E plasmids were derived from strains ofserogroup O145, a serogroup that is frequently recovered fromseverely ill patients (3, 27), yet these particular plasmid types wereonly rarely identified (31, 32). Instead, we found that strainsRM13514 and RM13516 of serotype O145:H28 harbor plasmidsof ehxA subtype C, which carry quite a different repertoire of vir-ulence genes compared to subtype E plasmids (Fig. 1B, Table 2)from O145:H25.

Despite the allegedly low contribution of subtype E plas-mids to the clinical presentation of this pathovar, genomicallysubtype E strains are better equipped to cause severe diseasethan subtype D strains. Like subtype D strains (CFSAN004180and CFSAN004181), subtype E strains carry the HUS-associatedstx2a variant; however, contrary to subtype D, subtype E plasmidswere derived from LEE-positive STEC strains, and thus possess anefficient colonization and adhesion mechanism. Furthermore,subtype E strains possess a second 34,714-bp non-ehxA-encodingplasmid that carried an intact sfp fimbrial gene cluster, which quitepossibly contributes to the overall pathogenicity of this pathovar(Fig. 4B) (82, 83). Noteworthy, the complete sfp cluster was ini-tially thought to be a unique pSFO157 characteristic among hu-

man-pathogenic E. coli and has only recently been identified onpO165 from EHEC strains associated with diarrhea and HUS (82,84). It is likely, that the sfp cluster noted on subtype E plasmids wasacquired via lateral gene transfer, as it was surrounded by a num-ber of transposases (Fig. 4B).

Comparative analysis with ehxA subtype A, B, C, and F plas-mids and phylogenetic analysis. While the 200-kbp virulenceplasmids of subtype D were considerably larger than the plasmidsof the other subtypes, subtype E plasmids were much smaller asthey were only 50 kbp in size (Table 2). Furthermore, the majorityof genes noted on subtype D plasmids were absent on the otherfive subtype plasmids, and only a few genetic features wereshared with pO157, pO145 (strain RM13516), and pO103, be-longing to ehxA subtypes B, C, and F (Table 2, Fig. 1A). Theseincluded portions of the F conjugal transfer apparatus, al-though subtype D plasmids carried, by far, the largest and mostcomplete transfer region, as well as stcE and the associated T2SScarried by pCFSAN004179 (Fig. 1A). On the contrary, pO113 andpO104 of the subtype A strains carried the transfer apparatus ofIncI1 plasmids. Despite these similarities, subtype D plasmids alsolacked characteristic features that were common to plasmids ofehxA subtypes B, C, and F, such as the 4-gene ecf cluster, espP,toxB, and katP (Table 2). Thus, the highly divergent sequencebackbone of the subtype D plasmids compared to those of theother ehxA plasmids supports the idea that subtype D plasmids, infact, have evolved from a different evolutionary lineage (Fig. 5).

The current study mostly confirmed the evolutionary distances

FIG 4 Circular maps of virulence plasmids pO145_ehx (A) and pO145_sfp (B) generated with CGView Server 1.0 (http://stothard.afns.ualberta.ca/cgview_server/). Genes on the leading and lagging strand are shown on the outer two rings, respectively, with arrow direction indicating direction of transcription. GCcontent is represented by the black ring with high GC content as the outermost portion. GC skew is shown on the innermost ring. Genes displayed are categorizedby color as shown on the bottom. Red includes known and putative virulence factors; pink includes methylase-, resolvase-, and endonuclease-encoding genes;and yellow includes replication-, maintenance-, and transfer-associated genes (disrupted genes are not displayed).

Analysis of Virulence Plasmids from STEC

November 2016 Volume 82 Number 21 aem.asm.org 6373Applied and Environmental Microbiology

on June 20, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

among ehxA subtypes, as previously determined by phylogeneticanalysis; however, contrary to previous studies (31, 32), we foundthat subtype E clustered with subtypes B, C, and F and apart fromsubtype A (Fig. 5). The enhanced algorithms of the newer MEGAsoftware, associated with a higher number of bootstrap replicatesused in this study, most likely resulted in the construction of aslightly different phylogenetic tree (42). Nevertheless, the currentphylogeny appears rather likely, as subtype E plasmids share moregenetic features with the plasmids of subtypes B, C, and F thanwith subtype A plasmids (Table 2; Fig. 1B).

Comparison of sequencing plasmid DNA-only versus whole-genome DNA templates. Large virulence plasmids are, generally,low copy number with one or two plasmids per cell, ranging in sizefrom 50 to 200 kbp (24). Sequencing low-copy-number plasmidsagainst the background of the much larger, approximately 5-MbpE. coli chromosome may result in decreased sequencing coverageof the plasmids, potentially leading to erroneous sequence assem-blies. Thus, whole-genome preparations containing plasmids and

plasmid-only DNA preparations were sequenced, and the resultswere compared for accuracy.

Expectedly, sequencing plasmid DNA-only templates pro-duced significantly more mapped sequencing reads and higheraverage sequencing coverage/depth across the plasmids than thesequencing results obtained from whole-genome preparations(see Table S2 in the supplemental material). However, despite thesignificant differences in efficacy observed using different tem-plates, overall sequence data shared 99% nucleotide sequenceidentity. Presumably, the long sequencing reads produced by thePacBio sequencing platform minimized the number of erroneousassemblies.

Conclusion. We demonstrated that ehxA subtype D plasmidsrepresent a novel E. coli virulence plasmid that most likely evolvedfrom a different evolutionary lineage than plasmids of the otherfive ehxA subtypes. Despite being isolated from foods, subtype Dplasmids encoded a variety of virulence determinants that are as-sociated with pathogenic ETEC (sta1, astA, K88) and EHEC (stcE,

FIG 5 Neighbor-joining phylogenetic tree based on nucleotide sequences of ehxA. Evolutionary distances were computed using the maximum compositelikelihood method with 1,000 bootstrap replicates in MEGA 6.06. The scale bar indicates the number of substitutions per base; bootstrap values are displayed atbranch points. Strains AGR053, AGR119, ER03 4238, AGR158, AGR151, and AGR670 served as reference sequences for the six ehxA subtypes (31).

Lorenz et al.

6374 aem.asm.org November 2016 Volume 82 Number 21Applied and Environmental Microbiology

on June 20, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

ehxA) strains and, thus, carry the potential of causing disease. Incontrast, subtype E plasmids, isolated from HUS outbreak strains,carry only few virulence determinants, including the potentiallyvirulence-associated espI gene. However, genomically subtype Estrains are better equipped to cause severe disease than those ofehxA subtype D, particularly as subtype E plasmids were derivedfrom LEE-positive STEC strains. Our insights serve as an impor-tant reminder that, due to the plasticity and vast diversity amongE. coli virulence plasmids, E. coli strains that have not yet demon-strated virulence should still be under surveillance to best protectfood/feed safety and human health.

ACKNOWLEDGMENTS

We thank Tim Muruvanda, Justin Payne, and Ruth E. Timme for techni-cal assistance in software analysis and genome sequencing on the PacBioplatform, as well as for the NCBI submission process. We also thank Lili F.Vélez for technical review of the manuscript.

This project was supported by the appointment of S.C.L. to the Re-search Fellowship Program for the Center for Food Safety and AppliedNutrition administered by the Oak Ridge Associated Universities througha contract with the FDA. This project is further part of the Ph.D. program(S.C.L.) at the University of Hamburg, Germany under the supervision ofMarkus Fischer.

REFERENCES1. Scallan E, Hoekstra RM, Angulo FJ, Tauxe RV, Widdowson M, Roy SL,

Jones JL, Griffin PM. 2011. Foodborne illness acquired in the UnitedStates—major pathogens. Emerg Infect Dis 17:7–15. http://dx.doi.org/10.3201/eid1701.P11101.

2. Rangel JM, Sparling PH, Crowe C, Griffin PM, Swerdlow DL. 2005.Epidemiology of Escherichia coli O157:H7 outbreaks, United States, 1982–2002. Emerg Infect Dis 11:603– 609. http://dx.doi.org/10.3201/eid1104.040739.

3. Bettelheim KA. 2007. The non-O157 Shiga-toxigenic (verocytotoxigenic)Escherichia coli; under-rated pathogens. Crit Rev Microbiol 33:67– 87.http://dx.doi.org/10.1080/10408410601172172.

4. Johnson KE, Thorpe CM, Sears CL. 2006. The emerging clinical impor-tance of non-O157 Shiga toxin-producing Escherichia coli. Clin Infect Dis43:1587–1595. http://dx.doi.org/10.1086/509573.

5. Konowalchuk J, Speirs JI, Stavric S. 1977. Vero response to a cytotoxin ofEscherichia coli. Infect Immun 18:775–779.

6. McDaniel TK, Jarvis KG, Donnenberg MS, Kaper JB. 1995. A geneticlocus of enterocyte effacement conserved among diverse enterobacterialpathogens. Proc Natl Acad Sci U S A 92:1664 –1668. http://dx.doi.org/10.1073/pnas.92.5.1664.

7. Paton JC, Paton AW. 1998. Pathogenesis and diagnosis of Shiga toxin-producing Escherichia coli infections. Clin Microbiol Rev 11:450 – 479.

8. Karmali M, Petric M, Steele B, Lim C. 1983. Sporadic cases of haemo-lytic-uraemic syndrome associated with faecal cytotoxin and cytotoxin-producing Escherichia coli in stools. Lancet i:619 – 620. http://dx.doi.org/10.1016/S0140-6736(83)91795-6.

9. Levine MM. 1987. Escherichia coli that cause diarrhea: enterotoxigenic,enteropathogenic, enteroinvasive, enterohemorrhagic, and enteroadher-ent. J Infect Dis 155:377–389. http://dx.doi.org/10.1093/infdis/155.3.377.

10. Knutton S, Baldwin T, Williams P, Manjarrez-Hernandez A, Aitken A.1993. The attaching and effacing virulence property of enteropathogenicEscherichia coli. Zentralbl Bakteriol 278:209 –217.

11. Jarvis KG, Giron JA, Jerse AE, McDaniel TK, Donnenberg MS, KaperJB. 1995. Enteropathogenic Escherichia coli contains a putative type IIIsecretion system necessary for the export of proteins involved in attachingand effacing lesion formation. Proc Natl Acad Sci U S A 92:7996 – 8000.http://dx.doi.org/10.1073/pnas.92.17.7996.

12. Jerse AE, Yu J, Tall BD, Kaper JB. 1990. A genetic locus of enteropatho-genic Escherichia coli necessary for the production of attaching and effac-ing lesions on tissue culture cells. Proc Natl Acad Sci U S A 87:7839 –7843.http://dx.doi.org/10.1073/pnas.87.20.7839.

13. Paton AW, Woodrow MC, Doyle RM, Lanser JA, Paton JC. 1999.Molecular characterization of a Shiga toxigenic Escherichia coli O113:H21

strain lacking eae responsible for a cluster of cases of hemolytic-uremicsyndrome. J Clin Microbiol 37:3357–3361.

14. Beutin L, Martin A. 2012. Outbreak of Shiga toxin-producing Escherichiacoli (STEC) O104:H4 infection in Germany causes a paradigm shift withregard to human pathogenicity of STEC strains. J Food Prot 75:408 – 418.http://dx.doi.org/10.4315/0362-028X.JFP-11-452.

15. Luck SN, Bennett-Wood V, Poon R, Robins-Browne RM, Hartland EL.2005. Invasion of epithelial cells by locus of enterocyte effacement-negative enterohemorrhagic Escherichia coli. Infect Immun 73:3063–3071.http://dx.doi.org/10.1128/IAI.73.5.3063-3071.2005.

16. Luck SN, Badea L, Bennett-Wood V, Robins-Browne R, Hartland EL.2006. Contribution of FliC to epithelial cell invasion by enterohemor-rhagic Escherichia coli O113: H21. Infect Immun 74:6999 –7004. http://dx.doi.org/10.1128/IAI.00435-06.

17. Wang H, Paton JC, Paton AW. 2007. Pathologic changes in mice inducedby subtilase cytotoxin, a potent new Escherichia coli AB5 toxin that targetsthe endoplasmic reticulum. J Infect Dis 196:1093–1101. http://dx.doi.org/10.1086/521364.

18. Beutin L, Montenegro MA, Orskov I, Orskov F, Prada J, ZimmermannS, Stephan R. 1989. Close association of verotoxin (Shiga-like toxin)production with enterohemolysin production in strains of Escherichia coli.J Clin Microbiol 27:2559 –2564.

19. Beutin L, Zimmermann S, Gleier K. 1996. Rapid detection and isolationof Shiga-like toxin (verocytotoxin)-producing Escherichia coli by directtesting of individual enterohemolytic colonies from washed sheep bloodagar plates in the VTEC-RPLA assay. J Clin Microbiol 34:2812–2814.

20. Beutin L, Aleksic S, Zimmermann S, Gleier K. 1994. Virulence factorsand phenotypical traits of verotoxigenic strains of Escherichia coli isolatedfrom human patients in Germany. Med Microbiol Immunol 183:13–21.

21. Boerlin P, Chen S, Colbourne JK, Johnson R, De Grandis S, Gyles C.1998. Evolution of enterohemorrhagic Escherichia coli hemolysin plas-mids and the locus for enterocyte effacement in Shiga toxin-producing E.coli. Infect Immun 66:2553–2561.

22. Newton HJ, Sloan J, Bulach DM, Seemann T, Allison CC, Tauschek M,Robins-Browne RM, Paton JC, Whittam TS, Paton AW, Hartland EL.2009. Shiga toxin-producing Escherichia coli strains negative for locus ofenterocyte effacement. Emerg Infect Dis 15:372–380. http://dx.doi.org/10.3201/eid1503.080631.

23. Frost LS, Leplae R, Summers AO, Toussaint A. 2005. Mobile geneticelements: the agents of open source evolution. Nat Rev Microbiol 3:722–732. http://dx.doi.org/10.1038/nrmicro1235.

24. Funnell BE, Phillips GJ (ed). 2004. Plasmid biology. ASM Press, Wash-ington, DC.

25. Kaper JB, Nataro JP, Mobley HL. 2004. Pathogenic Escherichia coli. NatRev Microbiol 2:123–140. http://dx.doi.org/10.1038/nrmicro818.

26. Johnson TJ, Nolan LK. 2009. Pathogenomics of the virulence plasmids ofEscherichia coli. Microbiol Mol Biol Rev 73:750 –774. http://dx.doi.org/10.1128/MMBR.00015-09.

27. Cooper KK, Mandrell RE, Louie JW, Korlach J, Clark TA, Parker CT,Huynh S, Chain PS, Ahmed S, Carter MQ. 2014. Comparative genomicsof enterohemorrhagic Escherichia coli O145:H28 demonstrates a commonevolutionary lineage with Escherichia coli O157:H7. BMC Genomics 15:17.http://dx.doi.org/10.1186/1471-2164-15-17.

28. Ogura Y, Ooka T, Iguchi A, Toh H, Asadulghani M, Oshima K,Kodama T, Abe H, Nakayama K, Kurokawa K, Tobe T, Hattori M,Hayashi T. 2009. Comparative genomics reveal the mechanism of theparallel evolution of O157 and non-O157 enterohemorrhagic Escherichiacoli. Proc Natl Acad Sci U S A 106:17939 –17944. http://dx.doi.org/10.1073/pnas.0903585106.

29. Schmidt H, Karch H. 1996. Enterohemolytic phenotypes and genotypesof Shiga toxin-producing Escherichia coli O111 strains from patients withdiarrhea and hemolytic-uremic syndrome. J Clin Microbiol 34:2364 –2367.

30. Schmidt H, Beutin L, Karch H. 1995. Molecular analysis of the plasmid-encoded hemolysin of Escherichia coli O157:H7 strain EDL 933. InfectImmun 63:1055–1061.

31. Cookson AL, Bennett J, Thomson-Carter F, Attwood GT. 2007. Molec-ular subtyping and genetic analysis of the enterohemolysin gene (ehxA)from Shiga toxin-producing Escherichia coli and atypical enteropatho-genic E. coli. Appl Environ Microbiol 73:6360 – 6369. http://dx.doi.org/10.1128/AEM.00316-07.

32. Lorenz SC, Son I, Maounounen-Laasri A, Lin A, Fischer M, Kase JA.2013. Prevalence of hemolysin genes and comparison of ehxA subtype

Analysis of Virulence Plasmids from STEC

November 2016 Volume 82 Number 21 aem.asm.org 6375Applied and Environmental Microbiology

on June 20, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

patterns in Shiga toxin-producing Escherichia coli (STEC) and non-STECstrains from clinical, food, and animal sources. Appl Environ Microbiol79:6301– 6311. http://dx.doi.org/10.1128/AEM.02200-13.

33. Heilig JS, Elbing KL, Brent R. 2001. Large-scale preparation of plasmidDNA. Curr Protoc Mol Biol Chapter 1:Unit1.7.

34. Hoffmann M, Muruvanda T, Allard MW, Korlach J, Roberts RJ,Timme R, Payne J, McDermott PF, Evans P, Meng J, Brown EW, ZhaoaS. 2013. Complete genome sequence of a multidrug-resistant Salmonellaenterica serovar Typhimurium var. 5-strain isolated from chicken breast.Genome Announc 1:e01068-13.

35. Chin C-S, Alexander DH, Marks P, Klammer AA, Drake J, Heiner C,Clum A, Copeland A, Huddleston J, Eichler EE, Turner SW, Korlach J.2013. Nonhybrid, finished microbial genome assemblies from long-readSMRT sequencing data. Nat Methods 10:563–569. http://dx.doi.org/10.1038/nmeth.2474.

36. Krumsiek J, Arnold R, Rattei T. 2007. Gepard: a rapid and sensitive toolfor creating dotplots on genome scale. Bioinformatics 23:1026 –1028.http://dx.doi.org/10.1093/bioinformatics/btm039.

37. Besemer J, Lomsadze A, Borodovsky M. 2001. GeneMarkS: a self-training method for prediction of gene starts in microbial genomes. Im-plications for finding sequence motifs in regulatory regions. Nucleic AcidsRes 29:2607–2618.

38. Johnson TJ, Siek KE, Johnson SJ, Nolan LK. 2006. DNA sequence of aColV plasmid and prevalence of selected plasmid-encoded virulence genesamong avian Escherichia coli strains. J Bacteriol 188:745–758. http://dx.doi.org/10.1128/JB.188.2.745-758.2006.

39. Johnson TJ, Johnson SJ, Nolan LK. 2006. Complete DNA sequence of aColBM plasmid from avian pathogenic Escherichia coli suggests that itevolved from closely related ColV virulence plasmids. J Bacteriol 188:5975–5983. http://dx.doi.org/10.1128/JB.00204-06.

40. Kelley LA, Sternberg MJ. 2009. Protein structure prediction on the Web:a case study using the Phyre server. Nat Protoc 4:363–371. http://dx.doi.org/10.1038/nprot.2009.2.

41. Lorenz SC, Fischer M, Kase JA. 2014. Improved PCR-RFLP method forthe identification of Escherichia coli enterohemolysin (ehxA) subtypes. JMicrobiol Methods 100:24 –26. http://dx.doi.org/10.1016/j.mimet.2014.02.010.

42. Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. 2013. MEGA6:molecular evolutionary genetics analysis version 6.0. Mol Biol Evol 30:2725–2729. http://dx.doi.org/10.1093/molbev/mst197.

43. Hale TL. 1991. Genetic basis of virulence in Shigella species. MicrobiolRev 55:206 –224.

44. Lan R, Lumb B, Ryan D, Reeves PR. 2001. Molecular evolution of largevirulence plasmid in Shigella clones and enteroinvasive Escherichia coli.Infect Immun 69:6303– 6309. http://dx.doi.org/10.1128/IAI.69.10.6303-6309.2001.

45. Weiglmeier PR, Rösch P, Berkner H. 2010. Cure and curse: E. coliheat-stable enterotoxin and its receptor guanylyl cyclase C. Toxins (Basel)2:2213–2229. http://dx.doi.org/10.3390/toxins2092213.

46. Yamamoto T, Echeverria P. 1996. Detection of the enteroaggregativeEscherichia coli heat-stable enterotoxin 1 gene sequences in enterotoxi-genic E. coli strains pathogenic for humans. Infect Immun 64:1441–1445.

47. Savarino SJ, Fasano A, Watson J, Martin BM, Levine MM, GuandaliniS, Guerry P. 1993. Enteroaggregative Escherichia coli heat-stable entero-toxin 1 represents another subfamily of E. coli heat-stable toxin. Proc NatlAcad Sci U S A 90:3093–3097. http://dx.doi.org/10.1073/pnas.90.7.3093.

48. Kostakioti M, Stathopoulos C. 2004. Functional analysis of the Tsh auto-transporter from an avian pathogenic Escherichia coli strain. Infect Immun72:5548 –5554. http://dx.doi.org/10.1128/IAI.72.10.5548-5554.2004.

49. Dozois CM, Dho-Moulin M, Brée A, Fairbrother JM, Desautels C,Curtiss R. 2000. Relationship between the Tsh autotransporter andpathogenicity of avian Escherichia coli and localization and analysis of theTsh genetic region. Infect Immun 68:4145– 4154. http://dx.doi.org/10.1128/IAI.68.7.4145-4154.2000.

50. Dautin N. 2010. Serine protease autotransporters of Enterobacteriaceae(SPATEs): Biogenesis and function. Toxins 2:1179 –1206. http://dx.doi.org/10.3390/toxins2061179.

51. Weiss A, Kortemeier D, Brockmeyer J. 2014. Biochemical characterizationof the SPATE members EspP� and EspI. Toxins (Basel) 6:2719–2731. http://dx.doi.org/10.3390/toxins6092719.

52. Dutta PR, Cappello R, Navarro-García F, Nataro JP. 2002. Functionalcomparison of serine protease autotransporters of Enterobacteriaceae.

Infect Immun 70:7105–7113. http://dx.doi.org/10.1128/IAI.70.12.7105-7113.2002.

53. Wells TJ, Totsika M, Schembri MA. 2010. Autotransporters of Esche-richia coli: a sequence-based characterization. Microbiology 156:2459 –2469. http://dx.doi.org/10.1099/mic.0.039024-0.

54. Pukatzki S, Ma AT, Sturtevant D, Krastins B, Sarracino D, Nelson WC,Heidelberg JF, Mekalanos JJ. 2006. Identification of a conserved bacterialprotein secretion system in Vibrio cholerae using the Dictyostelium hostmodel system. Proc Natl Acad Sci U S A 103:1528 –1533. http://dx.doi.org/10.1073/pnas.0510322103.

55. Pukatzki S, Ma AT, Revel AT, Sturtevant D, Mekalanos JJ. 2007. TypeVI secretion system translocates a phage tail spike-like protein into targetcells where it cross-links actin. Proc Natl Acad Sci U S A 104:15508 –15513.http://dx.doi.org/10.1073/pnas.0706532104.

56. Shrivastava S, Mande SS. 2008. Identification and functional characteriza-tion of gene components of type VI secretion system in bacterial genomes.PLoS One 3:e2955. http://dx.doi.org/10.1371/journal.pone.0002955.

57. Cascales E. 2008. The type VI secretion toolkit. EMBO Rep 9:735–741.http://dx.doi.org/10.1038/embor.2008.131.

58. Shneider MM, Buth SA, Ho BT, Basler M, Mekalanos JJ, Leiman PG. 2013.PAAR-repeat proteins sharpen and diversify the type VI secretion systemspike. Nature 500:350–353. http://dx.doi.org/10.1038/nature12453.

59. Ma J, Sun M, Bao Y, Pan Z, Zhang W, Lu C, Yao H. 2013. Geneticdiversity and features analysis of type VI secretion systems loci in avianpathogenic Escherichia coli by wide genomic scanning. Infect Genet Evol20:454 – 464. http://dx.doi.org/10.1016/j.meegid.2013.09.031.

60. Lorenz SC, Kotewicz ML, Hoffmann M, Gonzalez-Escalona N, FischerM, Kase JA. Complete Genome Sequences of Four Enterohemolysin-Positive (ehxA), Enterocyte Effacement-Negative (LEE) Shiga Toxin-Producing Escherichia coli. Genome Announc, in press.

61. Sarris PF, Scoulica EV. 2011. Pseudomonas entomophila and Pseudomo-nas mendocina: potential models for studying the bacterial type VI secre-tion system. Infect Genet Evol 11:1352–1360. http://dx.doi.org/10.1016/j.meegid.2011.04.029.

62. Fritsch MJ, Trunk K, Diniz JA, Guo M, Trost M, Coulthurst SJ. 2013.Proteomic identification of novel secreted antibacterial toxins of the Ser-ratia marcescens type VI secretion system. Mol Cell Proteomics 12:2735–2749. http://dx.doi.org/10.1074/mcp.M113.030502.

63. Sana TG, Hachani A, Bucior I, Soscia C, Garvis S, Termine E, Engel J,Filloux A, Bleves S. 2012. The second type VI secretion system of Pseu-domonas aeruginosa strain PAO1 is regulated by quorum sensing and Furand modulates internalization in epithelial cells. J Biol Chem 287:27095–27105. http://dx.doi.org/10.1074/jbc.M112.376368.

64. Doughty S, Sloan J, Bennett-Wood V, Robertson M, Robins-BrowneRM, Hartland EL. 2002. Identification of a novel fimbrial gene clusterrelated to long polar fimbriae in locus of enterocyte effacement-negativestrains of enterohemorrhagic Escherichia coli. Infect Immun 70:6761–6769. http://dx.doi.org/10.1128/IAI.70.12.6761-6769.2002.

65. Lathem WW, Grys TE, Witowski SE, Torres AG, Kaper JB, Tarr PI,Welch RA. 2002. StcE, a metalloprotease secreted by Escherichia coli O157:H7, specifically cleaves C1 esterase inhibitor. Mol Microbiol 45:277–288.http://dx.doi.org/10.1046/j.1365-2958.2002.02997.x.

66. Grys TE, Siegel MB, Lathem WW, Welch RA. 2005. The StcE proteasecontributes to intimate adherence of enterohemorrhagic Escherichia coliO157:H7 to host cells. Infect Immun 73:1295–1303. http://dx.doi.org/10.1128/IAI.73.3.1295-1303.2005.

67. Leyton DL, Sloan J, Hill RE, Doughty S, Hartland EL. 2003. Transferregion of pO113 from enterohemorrhagic Escherichia coli: similarity withR64 and identification of a novel plasmid-encoded autotransporter, EpeA.Infect Immun 71:6307– 6319. http://dx.doi.org/10.1128/IAI.71.11.6307-6319.2003.

68. Bakker D, Willemsen P, Willems R, Huisman TT, Mooi F, Oudega B,Stegehuis F, de Graaf F. 1992. Identification of minor fimbrial subunitsinvolved in biosynthesis of K88 fimbriae. J Bacteriol 174:6350 – 6358.

69. Van den Broeck W, Cox E, Oudega B, Goddeeris B. 2000. The F4fimbrial antigen of Escherichia coli and its receptors. Vet Microbiol 71:223–244. http://dx.doi.org/10.1016/S0378-1135(99)00174-1.

70. Bakker D, Willemsen P, Simons L, van Zijderveld F, de Graaf F. 1992.Characterization of the antigenic and adhesive properties of FaeG, themajor subunit of K88 fimbriae. Mol Microbiol 6:247–255. http://dx.doi.org/10.1111/j.1365-2958.1992.tb02006.x.

71. Deneke CF, McGowan K, Larson A, Gorbach SL. 1984. Attachment of

Lorenz et al.

6376 aem.asm.org November 2016 Volume 82 Number 21Applied and Environmental Microbiology

on June 20, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

human and pig (K88) enterotoxigenic Escherichia coli strains to eitherhuman or porcine small intestinal cells. Infect Immun 45:522–524.

72. Oshima K, Toh H, Ogura Y, Sasamoto H, Morita H, Park S-H, Ooka T,Iyoda S, Taylor TD, Hayashi T, Itoh K, Hattori M. 2008. Completegenome sequence and comparative analysis of the wild-type commensalEscherichia coli strain SE11 isolated from a healthy adult. DNA Res 15:375–386. http://dx.doi.org/10.1093/dnares/dsn026.

73. Anderson DG, Moseley SL. 1988. Escherichia coli F41 adhesin: geneticorganization, nucleotide sequence, and homology with the K88 determi-nant. J Bacteriol 170:4890 – 4896.

74. Korth MJ, Apostol J, Moseley SL. 1992. Functional expression of heter-ologous fimbrial subunits mediated by the F41, K88, and CS31A determi-nants of Escherichia coli. Infect Immun 60:2500 –2505.

75. Kotloff KL, Nataro JP, Blackwelder WC, Nasrin D, Farag TH, Pan-chalingam S, Wu Y, Sow SO, Sur D, Breiman RF, Faruque AS, ZaidiAK, Saha D, Alonso PL, Tamboura B, Sanogo D, Onwuchekwa U,Manna B, Ramamurthy T, Kanungo S, Ochieng JB, Omore R, OundoJO, Hossain A, Das SK, Ahmed S, Qureshi S, Quadri F, Adegbola RA,Antonio M, Hossain MJ, Akinsola A, Mandomando I, Nhampossa T,Acácio S, Biswas K, O’Reilly CE, Mintz ED, Berkeley LY, Muhsen K,Sommerfelt H, Robins-Browne RM, Levine MM. 2013. Burden andaetiology of diarrhoeal disease in infants and young children in developingcountries (the Global Enteric Multicenter Study, GEMS): a prospective,case-control study. Lancet 382:209 –222. http://dx.doi.org/10.1016/S0140-6736(13)60844-2.

76. Jores J, Rumer L, Wieler LH. 2004. Impact of the locus of enterocyteeffacement pathogenicity island on the evolution of pathogenic Esche-richia coli. Int J Med Microbiol 294:103–113. http://dx.doi.org/10.1016/j.ijmm.2004.06.024.

77. Frost LS, Ippen-Ihler K, Skurray RA. 1994. Analysis of the sequence andgene products of the transfer region of the F sex factor. Microbiol Rev58:162–210.

78. Schmidt H, Zhang W-L, Hemmrich U, Jelacic S, Brunder W, Tarr P,Dobrindt U, Hacker J, Karch H. 2001. Identification and characterization ofa novel genomic island integrated at selC in locus of enterocyte effacement-negative, Shiga toxin-producing Escherichia coli. Infect Immun 69:6863–6873. http://dx.doi.org/10.1128/IAI.69.11.6863-6873.2001.

79. Tatsuno I, Horie M, Abe H, Miki T, Makino K, Shinagawa H, TaguchiH, Kamiya S, Hayashi T, Sasakawa C. 2001. toxB Gene on pO157 ofenterohemorrhagic Escherichia coli O157: H7 is required for full epithelialcell adherence phenotype. Infect Immun 69:6660 – 6669. http://dx.doi.org/10.1128/IAI.69.11.6660-6669.2001.

80. Brockmeyer J, Bielaszewska M, Fruth A, Bonn ML, Mellmann A,Humpf H-U, Karch H. 2007. Subtypes of the plasmid-encoded serineprotease EspP in Shiga toxin-producing Escherichia coli: distribution, se-cretion, and proteolytic activity. Appl Environ Microbiol 73:6351– 6359.http://dx.doi.org/10.1128/AEM.00920-07.

81. Christenson JK, Gordon DM. 2009. Evolution of colicin BM plasmids:the loss of the colicin B activity gene. Microbiology 155:1645–1655. http://dx.doi.org/10.1099/mic.0.026666-0.

82. Brunder W, Khan AS, Hacker J, Karch H. 2001. Novel type of fimbriaeencoded by the large plasmid of sorbitol-fermenting enterohemorrhagicEscherichia coli O157: H�. Infect Immun 69:4447– 4457. http://dx.doi.org/10.1128/IAI.69.7.4447-4457.2001.

83. Müsken A, Bielaszewska M, Greune L, Schweppe CH, Müthing J,Schmidt H, Schmidt MA, Karch H, Zhang W. 2008. Anaerobic condi-tions promote expression of Sfp fimbriae and adherence of sorbitol-fermenting enterohemorrhagic Escherichia coli O157:NM to human intes-tinal epithelial cells. Appl Environ Microbiol 74:1087–1093. http://dx.doi.org/10.1128/AEM.02496-07.

84. Bielaszewska M, Prager R, Vandivinit L, Müsken A, Mellmann A, HoltNJ, Tarr PI, Karch H, Zhang W. 2009. Detection and characterization ofthe fimbrial sfp cluster in enterohemorrhagic Escherichia coli O165:H25/NM isolates from humans and cattle. Appl Environ Microbiol 75:64 –71. http://dx.doi.org/10.1128/AEM.01815-08.

85. Yan X, Fratamico PM, Bono JL, Baranzoni GM, Chen C-Y. 2015.Genome sequencing and comparative genomics provides insights on theevolutionary dynamics and pathogenic potential of different H-serotypesof Shiga toxin-producing Escherichia coli O104. BMC Microbiol 15:83.http://dx.doi.org/10.1186/s12866-015-0413-9.

86. Makino K, Ishii K, Yasunaga T, Hattori M, Yokoyama K, Yutsudo CH,Kubota Y, Yamaichi Y, Iida T, Yamamoto K, Honda T, Han CG,Ohtsubo E, Kasamatsu M, Hayashi T, Kuhara S, Shinagawa H. 1998.Complete nucleotide sequences of 93-kb and 3.3-kb plasmids of an en-terohemorrhagic Escherichia coli O157:H7 derived from Sakai outbreak.DNA Res 5:1–9. http://dx.doi.org/10.1093/dnares/5.1.1.

87. Hayashi T, Makino K, Ohnishi M, Kurokawa K, Ishii K, Yokoyama K,Han C-G, Ohtsubo E, Nakayama K, Murata T, Tanaka M, Tobe T, IidaT, Takami H, Honda T, Sasakawa C, Ogasawara N, Yasunaga T, KuharaS, Shiba T, Hattori M, Shinagawa H. 2001. Complete genome sequenceof enterohemorrhagic Escherichia coli O157:H7 and genomic comparisonwith a laboratory strain K-12. DNA Res 8:11–22. http://dx.doi.org/10.1093/dnares/8.1.11.

Analysis of Virulence Plasmids from STEC

November 2016 Volume 82 Number 21 aem.asm.org 6377Applied and Environmental Microbiology

on June 20, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from