Contribution of the SitABCD, MntH, and FeoB Metal Transporters to the Virulence of Avian Pathogenic Escherichia coli O78 Strain {chi}7122

The roles of SitABCD, MntH, and FeoB metal transporters in thevirulence of avian pathogenic Escherichia coli (APEC) O78 strain ${chi}$ 7122 were assessed using isogenic mutants in chicken infectionmodels. In a single-strain infection model, compared to ${chi}$ 7122,the ${Delta}$ sit strain demonstrated reduced colonization of the lungs,liver, and spleen. Complementation of the ${Delta}$ sit strain restoredvirulence. In a coinfection model, compared to the virulentAPEC strain, the ${Delta}$ sit strain demonstrated mean 50-fold, 126-fold,and 25-fold decreases in colonization of the lungs, liver, andspleen, respectively. A ${Delta}$ mntH ${Delta}$ sit strain was further attenuated,demonstrating reduced persistence in blood and mean 1,400-fold,954-fold, and 83-fold reduced colonization in the lungs, liver,and spleen, respectively. In coinfections, the ${Delta}$ feoB ${Delta}$ sit straindemonstrated reduced persistence in blood but increased colonizationof the liver. The ${Delta}$ mntH, ${Delta}$ feoB, and ${Delta}$ feoB ${Delta}$ mntH strains were asvirulent as the wild type in either of the infection models.Strains were also tested for sensitivity to oxidative stress-generatingagents. The ${Delta}$ mntH ${Delta}$ sit strain was the most sensitive strain andwas significantly more sensitive than the other strains to hydrogenperoxide, plumbagin, and paraquat. sit sequences were highlyassociated with APEC and human extraintestinal pathogenic E.coli compared to commensal isolates and diarrheagenic E. coli.Comparative genomic analyses also demonstrated that sit sequencesare carried on conjugative plasmids or associated with phageelements and were likely acquired by distinct genetic eventsamong pathogenic E. coli and Shigella sp. strains. Overall,the results demonstrate that SitABCD contributes to virulenceand, together with MntH, to increased resistance to oxidativestress.

INTRODUCTION

Top

INTRODUCTION

The Salmonella iron transporter, SitABCD, was first identifiedin Salmonella enterica serovar Typhimurium as a homolog of theYersinia pestis YfeABCD transporter (68) and is a member ofthe periplasmic binding protein-dependent ATP-binding-cassette(ABC) family of metal transporters (12). In Y. pestis and S.enterica serovar Typhimurium, inactivation of the genes encodingthe Yfe and Sit transporters, respectively, resulted in decreasedvirulence in murine experimental infection models (3, 25). SitABCDhomologs are also present in other pathogenic enterobacteria,including Shigella spp. (26, 55, 66), uropathogenic Escherichiacoli (UPEC) (6, 11, 67), and avian pathogenic E. coli (APEC)(28, 50, 56, 59). The genes encoding the Sit system are locatedon either genomic islands or large plasmids, and some strainscarry more than one copy of the sit system (27, 28, 56, 59).

Previously, SitABCD from APEC strain ${chi}$ 7122 was shown to transportiron and manganese, and in the absence of the endogenous manganesetransporter MntH, SitABCD contributed to protection of APECand E. coli K-12 against hydrogen peroxide (56). SitABCD fromS. enterica serovar Typhimurium similarly mediated transportof manganese and iron and, in combination with MntH, contributedto resistance to hydrogen peroxide (4). Furthermore, the lossof Sit in combination with the loss of either MntH or the Feoferrous iron transporter resulted in greater attenuation ofSalmonella virulence in susceptible mice (4). A Shigella flexneri2a sit mutant was less able to grow in iron-restricted medium(38, 55). However, loss of Sit had no appreciable effect onintracellular multiplication or cell-to-cell spread in epithelialcells or on virulence in the guinea pig keratoconjunctivitismodel (38, 55). In contrast, the cumulative loss of Sit andiron transporters such as Feo or the aerobactin siderophoresystem resulted in a reduced capacity to form plaques on epithelialcells (55). In addition, the cumulative loss of Sit and MntHresulted in a reduced capacity of Shigella flexneri 2a to survivein macrophage cell lines (53).

Divalent ion transporters such as SitABCD, MntH, and Feo maycontribute to the virulence of APEC and other bacterial pathogensby facilitating acquisition of iron and/or manganese. Both manganeseand iron are essential cofactors for enzymes required for metabolicprocesses and protection against oxidative stress (1, 23, 33).Thus far, the role of the SitABCD transporter, alone or in combinationwith other divalent metal transporters, such as MntH and Feo,in the virulence of pathogenic E. coli has not been investigated.A sitABCD mutant derivative of APEC strain ${chi}$ 7122 (O78:K80:H9)caused lesions in chickens similar to those induced by the parentalstrain and was also as resistant to hydrogen peroxide as thewild-type parent, suggesting that other metal transporters werepossibly compensating for the absence of the Sit system (56).However, a sitB signature-tagged transposon mutant of APEC strainIMT5155 (O2:K5), which contains two copies of the sit genes,was moderately attenuated compared to its wild-type parent ina coinfection competition model in chickens (37), suggestinga role for the Sit transporter in virulence. Herein we investigatethe individual role of SitABCD in the virulence of APEC O78:K80strain ${chi}$ 7122 in single-strain infection and competitive infectionmodels in chickens. In addition, we determined the roles ofthe MntH and Feo transporters, alone or combined with SitABCD,in the virulence of APEC strain ${chi}$ 7122 in the chicken and in resistanceto oxidative stress agents. The distribution of sit genes amongpathogenic E. coli and E. coli reference strains was also assessed.Finally, the sit sequences available from genomic databaseswere compared and suggest that sit genes were likely inheritedthrough distinct transfer events leading to integration withinconjugative plasmids in APEC or on prophage-associated genomicislands among E. coli and Shigella sp. strains.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Bacterial strains, plasmids, media, and growth conditions. Bacterial strains and plasmids used in this study are listedin Table 1. In addition, clinical and commensal fecal isolatesfrom various sources were used to screen for the presence ofsit sequences. The 72 members of the Escherichia coli reference(ECOR) collection represent a diverse population of E. colistrains that have been grouped phylogenetically by multilocusenzyme electrophoresis (22). The 297 APEC clinical isolateswere described elsewhere (16). APEC strains were previouslyclassified for virulence based on lethality for 1-day-old chicksfollowing subcutaneous inoculation, where LC1 corresponds tothe high-lethality class, LC2 corresponds to the low-lethalityclass, and LC3 corresponds to the nonlethal class (16). Humanextraintestinal pathogenic E. coli (ExPEC) isolates included32 strains from urosepsis and other extraintestinal infectionsfrom the United States obtained from J. R. Johnson (VeteransAdministration Medical Center, Minneapolis, MN). Thirty-twoE. coli fecal isolates from healthy poultry and 23 diarrheagenicE. coli isolates belonging to various pathotypes (12 enterotoxigenicE. coli [ETEC], 6 enteropathogenic E. coli, and 5 enterohemorrhagicE. coli strains) were kindly provided by J. M. Fairbrother (Universityof Montreal, Canada). Strains were grouped phylogeneticallyby either multilocus enzyme electrophoresis or a multiplex PCRmethod (13). Strains were maintained at –80°C in 25%glycerol following overnight culture in Luria-Bertani (LB) broth(10 g yeast extract, 5 g tryptone, and 10 g NaCl per liter).Strains and clones were routinely grown in LB broth or on LBagar plates (15 g agar per liter) at 37°C. E. coli strainDH5 ${alpha}$ was routinely used for plasmid cloning and recovery. Antibioticswere added, as required, at the following concentrations: kanamycin,30 µg/ml; ampicillin, 100 to 200 µg/ml; chloramphenicol,30 µg/ml; nalidixic acid, 15 µg/ml; and tetracycline,10 µg/ml.

View this table:
[in this window]
[in a new window]

TABLE 1. Bacterial strains and plasmids used in this study

DNA and genetic manipulations. Standard methods were used for isolation of bacterial genomicDNA, DNA manipulation, and cloning (58). Restriction enzymesand DNA ligase used in this study were purchased from New EnglandBiolabs, Invitrogen, or Amersham Pharmacia and used accordingto the suppliers' recommendations. Recombinant plasmids, PCRproducts, and restriction fragments were purified using plasmidminiprep, PCR cleanup, and gel extraction kits (Qiagen or Sigma)as recommended by the supplier. Transformation of E. coli strainswas routinely done by using the calcium/manganese-based or electroshockmethod, as described previously (19).

Presence of sit sequences among E. coli strains. The presence of sit sequences among different E. coli strainswas investigated by PCR amplification of segments of the sitAand sitD genes that span the length of the sitABCD system. ThesitA primers (sitA-F [5'-CGCAGGGGGCACAACTGAT-3'] and sitA-R[5' CCCTGTACCAGCGTACTGG-3']) amplify a 663-bp segment of sitA,and the sitD primers (sitD-F [5'-CTGTGCGCTGCTGTCGGTC-3'] andsitD-R [5'-GCGTTGTGTCAGGAGTAC-3']) amplify a 570-bp segmentof sitD. The specificity of each of the primers and the predictedamplification products were verified by comparative genomic/bioinformaticanalysis against the sit genes from the E. coli and Shigellagenomes or other nucleotide entries in the available databases,including GenBank (http://www.ncbi.nlm.nih.gov/) and coliBASE,an online database for comparative genomics of E. coli and relatedenterobacteria (http://colibase.bham.ac.uk/) (10). Crude DNAextracts of strains were prepared by alkaline lysis (48). Thereactions were carried out using Taq DNA polymerase (New EnglandBiolabs). A 5-µl volume of each bacterial cell lysatewas added to a PCR mixture with a final volume of 25 µlcontaining 6.25 pmol of each primer, 5 nmol of each deoxynucleosidetriphosphate, and 0.5 U of Taq polymerase in 1x buffer. PCRconditions were as follows: 95°C for 1 min, followed by30 cycles of 94°C for 30 s, 54°C for 30 s, and 72°Cfor 1 min and then an extension period of 72°C for 1 min.Specific amplification was confirmed using strains ${chi}$ 7122 andCFT073 as positive controls and E. coli K-12 MG1655, which lacksthe sit genes, as a negative control.

Construction of mutant derivatives of APEC strain ${chi}$ 7122 and single-copy complementation of the ${Delta}$ sit mutation. The ${Delta}$ sitABCD::tetAR (QT205), ${Delta}$ mntH::kan (QT878), and ${Delta}$ mntH ${Delta}$ sitABCD(QT1239) derivatives of strain ${chi}$ 7122 are described elsewhere(56). QT1517 was generated by FLP-mediated excision of the ${Delta}$ mntH::kanallele from strain QT878 by using plasmid pCP20 (14). The ${Delta}$ feoB::kanallele from E. coli K-12 strain JWK3372_1 (2) was used to introduce ${Delta}$ feoB mutations into APEC strains. Briefly, the ${Delta}$ feoB::kan allelewas amplified from genomic DNA of strain JWK3372_1 by usingprimers MfeoB1 (5'-TCTGGTCTCATGTCGCTGTC-3') and MfeoB2 (5'-GGTGGAACTCTGCTTTTTGC-3')and was introduced into strains ${chi}$ 7122, QT205, and QT1517 by homologousrecombination using the ${lambda}$ red recombinase method (14). Successfultransfer of the ${Delta}$ feoB::kan mutation was confirmed by PCR, usingprimers flanking the feoB region. The ${Delta}$ feoB::kan derivativesof ${chi}$ 7122, QT205, and QT1517 were designated QT877, QT1240, andQT1539, respectively.

The ${Delta}$ sitABCD mutant strain QT205 was complemented by single-copyintegration of plasmid pIJ93. pIJ93 was constructed by subcloningthe XbaI-SalI fragment of pIJ28, containing the sitABCD operon,into the same sites of suicide vector pGP704. pIJ93 was conjugatedfrom strain MGN-617 (16) to strain QT205. A strain that wasresistant to ampicillin and found to contain a full-length copyof the sit genes, as confirmed by PCR, was designated QT770.

Sensitivity of E. coli strains to reactive oxygen intermediate (ROI)-generating agents. Sensitivity to oxidative stress was determined by an agar overlaydiffusion method on LB and M9-glucose plates (1.5% agar) asdescribed by Boyer et al. (4), with some modifications. Overnightcultures grown in LB broth were adjusted to an optical densityat 600 nm (OD₆₀₀) of 0.5. For tests on M9 medium, the overnightcultures were washed with M9-glucose prior to OD₆₀₀ adjustment.One hundred microliters of each culture was suspended in moltentop agar (0.5% agar) and poured over the agar plates. Filterpaper disks (6-mm diameter; Becton Dickinson) were added tothe surfaces of the solidified overlays, and 10 µl ofhydrogen peroxide (30%), paraquat (200 mM for LB and 40 mM forM9), plumbagin (53 mM), phenazine methosulfate (PMS) (15 mM),or phenazine ethosulfate (PES) (15 mM) was spotted onto thedisks. The plates were then incubated overnight at 37°C,and following growth, the diameters of inhibition zones weremeasured.

Experimental infection of chickens via the air sacs. Two different infection models, a comparative single-straininfection model and a competitive coinfection model, were usedto investigate the importance of different metal transportersfor the virulence of APEC. Chickens used in these studies wereWhite Leghorn specific-pathogen-free chickens obtained fromeither Charles River Spafas (now Charles River Laboratories[Franklin, CT]) or the Canadian Food Inspection Agency (Ottawa,Canada). For the single-strain infection model, groups of 3-week-oldWhite Leghorn chickens were inoculated in the right thoracicair sac with 0.1 ml (10⁷ CFU) of a bacterial inoculum consistingof a diluted 24-h beef heart infusion broth culture of APECstrain ${chi}$ 7122 or an isogenic mutant derivative. For the coinfectionexperiments, strains were prepared as for the single-straininfections, and equal quantities (5 x 10⁶ CFU) of each mutantstrain and a virulent ${Delta}$ lacZYA derivative of strain ${chi}$ 7122 (QT51)were used as the inoculum. Use of the ${Delta}$ lacZYA derivative of strain ${chi}$ 7122 (QT51) in coinfections permitted a direct evaluation ofthe number of colonies of QT51 (Lac^– colonies) comparedto the metal transporter mutant (Lac⁺ colonies) on each plate.For the coinfection studies, blood samples were collected asepticallyfrom each chicken 6, 24, and 48 h following bacterial inoculationand were plated directly or diluted and then plated on MacConkey-lactoseagar plates (Difco) supplemented with nalidixic acid (40 µg/ml).All birds were euthanized at 48 h postinfection and then necropsied.For the single-strain infection experiments, gross mean lesionscores for the air sacs and combined lesion scores for the pericardiumand liver were determined as described by Lamarche et al. (36).Organs were removed aseptically. The left lung, liver, and spleenof each animal were weighed, suspended in phosphate-bufferedsaline, and homogenized with an Omnimixer homogenizer. Dilutionsof homogenates were plated onto MacConkey-lactose agar plateswith appropriate antibiotics for bacterial quantification. Severalrandomly selected colonies per organ were verified by serotypingusing O78-specific antiserum.

Bioinformatic analysis and comparison of sit sequences from different E. coli and Shigella strains. The sequences and locations of sit gene clusters within thegenomes of various E. coli and Shigella strains were identifiedfrom the available databases, including GenBank (http://www.ncbi.nlm.nih.gov/)and coliBASE. Sequence analyses, multiple sequence alignmentby CLUSTALW, and generation of a phylogenetic tree were doneusing the MEGA3 software package (35; http://www.megasoftware.net).The phylogenetic tree of the sitABCD gene clusters was constructedby using the neighbor-joining method (57) of pairwise comparisonby maximum likelihood analysis using the Jukes-Cantor estimate(30) to calculate nucleotide substitution rates.

Statistical analyses. Statistical analyses were performed using the Prism 4.0b softwarepackage (GraphPad Software).

RESULTS

Top

RESULTS

The sit genes are associated with prophage elements and conjugative plasmids. A previous report indicated that depending on the E. coli strain,sitABCD genes were located either on pColV-type plasmids oron the bacterial chromosome in ExPEC or APEC, and certain strainscontained both genomic and plasmid-carried copies of the sitgenes (56). Analysis of the genome and nucleotide sequence databasesindicated that sit sequences are present in ExPEC-UPEC strainsCFT073, 536, and UTI89, APEC strain O1, enteroaggregative E.coli (EAEC) O42 (Fig. 1), and all of the currently availableShigella genomes but are absent from E. coli K-12 strains andthe genomes of other pathogenic E. coli strains. sit gene clustersare also present on the plasmids of APEC O1(pAPEC-O1-ColBM),APEC O2(pAPEC-O2-ColV), and APEC ${chi}$ 7122(pAPEC-1) (Fig. 1), andgenomic and plasmid-carried copies of sit sequences are presentin APEC O2:K5 strain IMT5155 (GenBank accession no. AM072350and AM072351).

View larger version (12K):
[in this window]
[in a new window]

FIG. 1. (A) Map location of sit genes in E. coli and Shigella strains. The large circle represents the E. coli K-12 MG1655 genome as a comparative reference, with 0 indicating the start of the sequence. The insertion locations corresponding to E. coli K-12 are indicated on the outer perimeter. The inner black arrows represent the genomic islands and the orientation of the sit genes in different pathogenic strains. The outer gray arrows indicate corresponding e14 and Rac prophage elements located in the same regions in strain MG1655. The asterisks denote the E. coli K-12 corresponding genes flanking either side of SI-19, the genomic island encompassing the sit genes in S. flexneri 2a strain 301 (26). (B) Phylogenetic tree of the sitABCD genes from E. coli and Shigella spp., constructed using the neighbor-joining method with MEGA3 software (35). Ec, Escherichia coli; Sflex, Shigella flexneri; Sson, Shigella sonnei 53G; Sdys, Shigella dysenteriae M131. The full-length sequences spanning from the sitA start codon to the sitD stop codon were analyzed, and sitABCD genes from Salmonella enterica serovar Typhimurium LT2 were used as the outlier sequence. Distances were calculated using the Jukes and Cantor estimate, which assumes a uniform rate of nucleotide substitutions. Branch lengths are indicated on the scale or above branches that are off the scale. Bootstrap confidence levels are indicated adjacent to the nodes. sit sequence data were obtained from NCBI with GenBank accession numbers AY598030 (pAPEC-1 of E. coli ${chi}$ 7122) (56), AY545598 (E. coli APEC O2/pAPEC-O2-ColV) (29), DQ381420 (E. coli APEC O1/pAPEC-O1-ColBM) (27), NC_004431 (E. coli CFT073) (67), NC_008563 (E. coli APEC O1) (28), NC_007946 (E. coli UTI89) (11), NC_008253 (E. coli 536) (6), NC_004337 (S. flexneri 2a 301) (26), NC_004741 (S. flexneri 2a 2457T) (66), and NC_003197 (S. enterica LT2) (42) or from coliBASE (10; http://colibase.bham.ac.uk/) for E. coli O42, Shigella dysenteriae M131, and Shigella sonnei 53G. Data available from coliBASE were produced by the sequencing group at the Sanger Institute (http://www.sanger.ac.uk/Projects/Escherichia_Shigella/).

The sit gene clusters are present at four distinct chromosomallocations in the E. coli-Shigella genomes (Fig. 1A). Among thesequenced ExPEC-UPEC strains and APEC O1, the sitABCD genesare all part of genomic regions containing prophage sequencesthat are inserted at the icd-ycgX intergenic region, which isthe attB attachment site for certain lambdoid bacteriophages.These prophage element-containing regions have identical junctionborders to those in the E. coli K-12 genome and vary in sizefrom 55 kb (CFT073) (67) to 47.3 kb (APEC O1) (28). In E. coliK-12 MG1655, this region harbors the e14 element, a vestigeof a lambdoid prophage, which encompasses 15.4 kb (43). In EAECstrain O42, Shigella sonnei, and Shigella dysenteriae, the sitgenes are part of 48.4-kb, 25.6-kb, and 19.9-kb regions, respectively,comprised mainly of phage elements, and are inserted at theydaO (b1344)-ynaF (b1376) region of E. coli K-12 MG1655 (10).This region is the attachment site for Rac prophage and relatedprophage elements in E. coli K-12 and O157:H7 strains (8). InS. flexneri 2a strain 301, sit genes are in the middle of a42-kb genomic island, Shigella island 19 (SI-19), located ina region that has undergone rearrangements compared to E. coliMG1655. SI-19 contains genes that are homologous to phage sequencesand insertion sequence elements and is flanked by sequencescorresponding to ycjI and yebU in E. coli K-12 MG1655 (26).In S. flexneri 2a strain 2457T, sit genes are in the middleof a 37-kb genomic island that is adjacent to aspS. Despitedifferences in the locations of the sit genes within the genomesof S. flexneri 2a strains 301 and 2457T, the lambdoid prophagesequences immediately flanking the sit genes in these two genomesare similar. Sequences flanking the sit genes on plasmids aresimilar to each other but are distinct from those adjacent tothe sit genes present on the genomic islands of any of the E.coli or Shigella strains.

Comparison of the nucleotide sequences encoding SitABCD fromE. coli and Shigella strains in a neighbor-joining tree demonstratedthat sit gene clusters form three main groups, comprised ofthe ExPEC-APEC genomic sit sequences, the Shigella sit sequences,and the plasmid-carried APEC sit sequences (Fig. 1B). The E.coli O42 sit sequence branched separately but was most closelyrelated to Shigella sequences. Phylogenetic analyses of eachof the sit genes generated the same distinct groupings (datanot shown). Overall, the sit sequences demonstrated nucleotidevariability at 156 sites, which corresponded to a variabilityof 4.5% of the sit gene cluster. Sequences within each of thethree main cluster groups demonstrated low variability (from0.4% to 0.6% overall) within their respective clusters. In contrast,between cluster groups, variability increased two- to sixfold(from 1.2% to 2.5%). Taken together, these differences in genomiclocations and nucleotide sequence variability among the sitgenes from different E. coli and Shigella strains support thelikelihood that sit genes were acquired by these two closelyrelated species via a number of distinct genetic events involvinghorizontal gene transfer mediated by phages, plasmids, or othermobile genetic elements.

Distribution of sit genes among ECOR collection and E. coli clinical and fecal commensal isolates. The presence and distinct localization of sit sequences amongdifferent E. coli and Shigella sp. strains prompted us to furtherinvestigate the distribution of sit sequences among a diversepopulation of strains, including ECOR collection, ExPEC, diarrheagenicE. coli, APEC, and avian fecal commensal E. coli strains. PCRanalysis was performed using sitA- and sitD-specific primerswhich correspond to conserved regions in the sitA and sitD genescurrently available from the sequence databases. For all strainstested, there was no discrepancy between results when usingeither the sitA or sitD primers. Among the 72 members of theECOR collection, which have been arranged into four phylogeneticgroups (A, B1, B2, and D) on the basis of multilocus enzymeelectrophoresis (22), the presence of sit sequences was significantlyassociated with phylogenetic group B2 compared to the othergroups (P < 0.0001) (Table 2). All B2 group strains fromthe ECOR collection were sit positive, whereas in the otherphylogenetic groups 50% or fewer were sit positive. In addition,all 32 human ExPEC strains tested were sit positive, whereasonly 2 of 23 diarrheagenic E. coli isolates contained sit sequences(Table 2). The presence of sit sequences was also investigatedin APEC and avian fecal commensal isolates (Table 3). Among297 APEC strains classified according to their lethality for1-day-old chicks, sit sequences were more common in the highlyvirulent APEC (class 1 lethality) strains (99%) than in theless virulent strains (81% for class 2 and 68% for class 3)(P < 0.001). In addition, sit sequences were highly associatedwith APEC compared to avian fecal commensal isolates (P <0.001). For APEC isolates, 275/297 (93%) isolates were sit positive,whereas only 14/32 (44%) avian fecal isolates were sit positive.Overall, the results demonstrate that sit sequences are highlyprevalent in E. coli strains associated with extraintestinalinfections compared to diarrheagenic E. coli and commensal fecalisolates.

View this table:
[in this window]
[in a new window]

TABLE 2. Distribution of sit-positive strains within phylogenetic groups among ECOR collection and Escherichia coli clinical isolates

View this table:
[in this window]
[in a new window]

TABLE 3. Distribution of sit genes among APEC and avian commensal fecal isolates according to lethality class

SitABCD contributes to virulence of APEC during infection of chickens. Since sit sequences were highly associated with E. coli isolatesfrom extraintestinal infections in poultry, we investigatedthe importance of sit genes for the virulence of APEC O78 strain ${chi}$ 7122 in chickens. We first compared the virulence potentialof the ${Delta}$ sitABCD mutant strain QT205 with that of its isogenicwild-type parent strain, ${chi}$ 7122, in a single-strain infectionmodel. Compared to wild-type parent strain ${chi}$ 7122, strain QT205( ${Delta}$ sitABCD) was attenuated and demonstrated significantly reducedbacterial numbers in the lungs and livers of infected chickens(Fig. 2a and b). QT205 demonstrated a mean 1-log reduction incolonization of the spleens; however, the difference was notsignificant (P = 0.0545) (Fig. 2c). The sit-complemented derivativeof QT205 (QT770) regained virulence to levels comparable tothat of the wild-type parent strain, ${chi}$ 7122 (Fig. 2). Despitethe decreased persistence of QT205 in chickens, this straincaused gross lesions of airsacculitis and pericarditis/perihepatitisthat were similar to those caused by the wild-type parent (datanot shown).

View larger version (11K):
[in this window]
[in a new window]

FIG. 2. Colonization of extraintestinal organs of chickens by APEC strain ${chi}$ 7122 and derivatives in the single-strain infection model. Data are presented as log CFU/gram of tissue. Each data point represents a tissue sample from an individual infected chicken at 48 h postinfection. Organs sampled were the lungs (a), liver (b), and spleen (c). Strains tested were wild-type APEC strain ${chi}$ 7122 (•), QT205 ( ${Delta}$ sitABCD) ( ${blacktriangleup}$ ), and QT770 (QT205::sitABCD) ( ${blacksquare}$ ). P values for comparative differences in colonization by QT205 and the wild-type strain, as determined by the Mann-Whitney test, are indicated below the x axes.

We also investigated the role of the Feo ferrous iron transporterand the MntH manganese transporter in the virulence of APECstrain ${chi}$ 7122. In single-strain infection experiments, in contrastto strain QT205 ( ${Delta}$ sitABCD), strains QT877 ( ${Delta}$ feoB) and QT878 ( ${Delta}$ mntH)persisted in all tissues at levels similar to that of wild-typeAPEC strain ${chi}$ 7122. Thus, in the presence of a functional SitABCDsystem, FeoB or MntH did not appear to be of major importancefor the virulence of APEC ${chi}$ 7122 in chickens. In addition, inthe single-strain infection model, mutant strains QT1239 ( ${Delta}$ sitABCD ${Delta}$ mntH) and QT1240 ( ${Delta}$ sitABCD ${Delta}$ feoB) were not any more attenuatedthan the ${Delta}$ sitABCD single mutant QT205 (data not shown). The mutantderivatives lacking functional Sit, MntH, and/or Feo transportersdemonstrated no appreciable difference in growth rate comparedto the wild-type parent when cultured in either rich (Luriabroth) or minimal (M9-glucose) medium (data not shown). Therefore,the reduced colonization of tissues by the sit and other attenuatedmutants during infection of chickens was most likely due toa decreased capacity to survive in vivo, not a global reductionin fitness or growth.

Compared to single-strain infection models, competitive coinfectionmodels using virulent strains and isogenic mutants can demonstratemore sensitivity to differences in colonization or virulence.Because single-strain infections in chickens did not show anyattenuation for strains QT877 ( ${Delta}$ feoB) and QT878 ( ${Delta}$ mntH), and sinceno differences in attenuation between strain QT205 ( ${Delta}$ sitABCD)and the double mutants QT1239 ( ${Delta}$ sitABCD ${Delta}$ mntH) and QT1240 ( ${Delta}$ sitABCD ${Delta}$ feoB) were observed in single-strain infections, we tested thesestrains in a competitive coinfection model. For this model,we used APEC strain QT51, a ${Delta}$ lacZYA derivative of strain ${chi}$ 7122,as the competitor strain. QT51 was shown to be as virulent asthe wild-type parent strain ${chi}$ 7122 in both single-strain infectionand coinfection experiments (data not shown). The use of a virulentLac-negative wild-type derivative permitted us to directly comparethe levels of the Lac-positive mutant derivative and the virulentLac-negative competitor strain from the same biological samplesby differential counts on MacConkey-lactose agar plates.

In coinfection experiments, strain QT205 ( ${Delta}$ sitABCD) was clearlyattenuated and showed a significantly reduced competitive index(CI) compared to that of competitor strain QT51, with a mean50-fold decrease in the lung (Fig. 3a) and a mean 126-fold decreasein the liver (Fig. 3b). These results were consistent with theresults observed for the single-strain infection experiments.In addition, QT205 was also significantly reduced, by a meanof 25-fold, in the spleen (Fig. 3c). During the course of theinfection, no significant differences between QT51 and QT205were apparent in the blood at 6 h and 24 h postinfection (Fig.4). However, QT205 was significantly reduced, by a mean of 3.6-fold,in the blood by 48 h postinfection. Unlike strain QT205, strainsQT877 ( ${Delta}$ feoB) and QT878 ( ${Delta}$ mntH) were present at similar levelsto those of strain QT51 in all tissues and blood (Fig. 3 and4). These results were consistent with the lack of attenuationof these strains in the single-strain infection model.

View larger version (7K):
[in this window]
[in a new window]

FIG. 3. Comparative colonization of chicken tissues by APEC ${chi}$ 7122 derivatives lacking the Sit, MntH, or Feo metal transporter and virulent ${chi}$ 7122 ${Delta}$ lac derivative strain QT51. A competitive coinfection model was used in which QT51 and different metal transporter mutants were inoculated simultaneously. At 48 h postinfection, tissues were sampled, and results are presented as the log₁₀ CI. The CI represents the relative numbers of the two test strains from the tissues sampled (the output ratio) compared to the initial numbers of the strains in the inoculum (input ratio). Negative CI values indicate a decreased capacity for the mutant to compete with the virulent test strain (QT51). Horizontal bars indicate the mean log₁₀ CI values. Each data point represents a sample from an individual chicken. Ten to 12 chickens were used for each infection group. Organs sampled were the lungs (a), spleen (b), and liver (c). Strains tested were QT205 ( ${Delta}$ sitABCD) (•), QT878 ( ${Delta}$ mntH) ( ${triangleup}$ ), QT877 ( ${Delta}$ feoB) ( ${square}$ ), QT1239 ( ${Delta}$ sitABCD ${Delta}$ mntH) ( ${blacktriangleup}$ ), and QT1240 ( ${Delta}$ sitABCD ${Delta}$ feoB) ( ${blacksquare}$ ). Statistically significant decreases in CI values are indicated with asterisks (*, P < 0.005; **, P < 0.001), as determined by the Wilcoxon matched-pair test.

View larger version (7K):
[in this window]
[in a new window]

FIG. 4. Comparative persistence of APEC ${chi}$ 7122 derivatives lacking the Sit, MntH, or Feo metal transporter and virulent ${chi}$ 7122 ${Delta}$ lac derivative strain QT51 in the blood of infected chickens. Experiments were conducted using a coinfection model as described in the legend to Fig. 3. Blood was sampled at 6 h (a), 24 h (b), and 48 h (c) postinfection. Strains tested were QT205 ( ${Delta}$ sitABCD) (•), QT878 ( ${Delta}$ mntH) ( ${triangleup}$ ), QT877 ( ${Delta}$ feoB) ( ${square}$ ), QT1239 ( ${Delta}$ sitABCD ${Delta}$ mntH) ( ${blacktriangleup}$ ), and QT1240 ( ${Delta}$ sitABCD ${Delta}$ feoB) ( ${blacksquare}$ ). Statistically significant decreases in CI values, as determined by the Wilcoxon matched-pair test, are indicated with asterisks (*, P < 0.005; **, P < 0.001).

In coinfections using the double mutants, strain QT1239 ( ${Delta}$ sitABCD ${Delta}$ mntH) was the most attenuated strain (Fig. 3 and 4). QT1239demonstrated a significantly reduced CI in the blood at alltimes, with mean decreases of 2.2-fold, 16.2-fold, and 10-foldcompared to the competitor strain at 6 h, 24 h, and 48 h, respectively.QT1239 also demonstrated a mean 1,400-fold reduction in thelungs, 954-fold reduction in the liver, and 83-fold reductionin the spleen compared to strain QT51. In contrast, strain QT1240( ${Delta}$ sitABCD ${Delta}$ feoB) demonstrated significant, 2.2-fold and 5.6-foldreductions in the blood at 24 h and 48 h, respectively. QT1240was also reduced 26.8-fold in the lungs and 6.8-fold in thespleen but was not significantly reduced in the liver (Fig.3). Strain QT1539 ( ${Delta}$ feoB ${Delta}$ mntH) did not demonstrate any significantdecrease compared to strain QT51 in either the blood or tissues(data not shown). Comparison of the mean CIs for strains QT205( ${Delta}$ sitABCD), QT1239 ( ${Delta}$ sitABCD ${Delta}$ mntH), and QT1240 ( ${Delta}$ sitABCD ${Delta}$ feoB)demonstrated no significant differences between the groups inthe blood and spleen. In the lung, QT1239 ( ${Delta}$ sitABCD ${Delta}$ mntH) wassignificantly more attenuated in the coinfection model thanwas QT205 ( ${Delta}$ sitABCD) (P = 0.0178). In contrast, QT205 ( ${Delta}$ sitABCD)was significantly more attenuated than QT1240 ( ${Delta}$ sitABCD ${Delta}$ feoB)(P = 0.0061) in the liver. Taken together, these results demonstratethat, individually, the Sit system is most important for competitivesurvival and persistence compared to the Feo and MntH transporters.In addition, a cumulative loss of Sit and MntH transportersglobally resulted in more marked attenuation than that withthe loss of only the Sit transporter. In contrast, a cumulativeloss of the Sit and Feo systems compared to the individual lossof Sit actually resulted in reduced persistence in the bloodbut increased colonization of the liver.

Sensitivity of E. coli strains to ROI-generating compounds. The sensitivities of APEC strain ${chi}$ 7122 and its mutant derivativesto ROI-generating compounds was assessed on both rich (LB) andminimal (M9-glucose) media (Fig. 5). On LB medium, only the ${Delta}$ mntH ${Delta}$ sitABCD (QT1239) strain was more sensitive than the wild-typeparent strain to the ROI-generating compounds H₂O₂ (P = 0.002)and plumbagin (P = 0.027). On minimal medium, QT1239 was moresensitive to H₂O₂ (P = 0.015), plumbagin (P = 0.004), and paraquat(P = 0.004). On minimal medium, the ${Delta}$ mntH mutant QT878 was moresensitive to H₂O₂ (P = 0.012) and paraquat (P < 0.001) thanthe APEC wild-type parent but was less sensitive to these productsthan the ${Delta}$ mntH ${Delta}$ sitABCD derivative. In addition, on minimal medium,the ${Delta}$ feoB ${Delta}$ sitABCD strain (QT1539) was also somewhat more sensitiveto H₂O₂ (P = 0.007) than the wild-type parent. In contrast,the ${Delta}$ sitABCD mutant QT205 did not demonstrate any increased sensitivityto ROI-generating products compared to the wild-type parent(data not shown).

View larger version (32K):
[in this window]
[in a new window]

FIG. 5. Sensitivities of wild-type APEC strain ${chi}$ 7122 and isogenic metal transport mutants to ROI-generating compounds. Strains were grown on either LB agar or M9 agar, and tests were performed as described in Materials and Methods. ROI-generating compounds tested were 30% hydrogen peroxide (H₂O₂), 53 mM plumbagin, 15 mM PMS or PES, and 40 mM paraquat. Paraquat was ineffective against all strains grown on LB medium, even at higher concentrations (up to 200 mM). Sensitivity to PES was similar to sensitivity to PMS (not shown). The results represent the means of replicate experiments for a minimum of three samples. Vertical bars represent the standard errors of the means. Only wild-type (WT) strain ${chi}$ 7122 and its mutant derivatives that demonstrated a significant increase (P < 0.05) in sensitivity to ROI-generating products are indicated. APEC strains indicated in the legend are ${chi}$ 7122 (7122 WT; •), QT878 (7122 mntH; ${blacktriangleup}$ ), QT1239 (7122 mntH sit; ${blacklozenge}$ ), and QT1240 (7122 feoB sit; ${blacksquare}$ ). E. coli K-12 strains BW25113 (K-12 ${blacksquare}$ ), JW3933 (K-12 oxyR; ${blacklozenge}$ ), and JW4023 (K-12 soxS; •) were used as comparative controls. For all conditions tested, K-12 strain BW25113 was significantly more sensitive (P < 0.05) to ROI-generating compounds than was APEC strain ${chi}$ 7122.

APEC strain ${chi}$ 7122 was intrinsically more resistant to all ofthe ROI-generating compounds tested than the K-12 control strainfollowing growth on either LB or minimal medium (P < 0.05)(Fig. 5). In addition, the APEC metal transporter mutants thatwere more sensitive to certain ROI-generating compounds werealways resistant to PMS and PES. Compared to the other ROI-generatingcompounds, PMS and PES specifically generated a superoxide stressresponse, as the ${Delta}$ soxS K-12 strain, but not the ${Delta}$ oxyR derivative,demonstrated sensitivity to these compounds (Fig. 5). Takentogether, these results demonstrate that the MntH and SitABCDdivalent metal transporters synergistically contributed to increasedresistance to ROIs that generate either a H₂O₂-specific or amixed H₂O₂ and superoxide stress response. In contrast to thein vivo studies with chickens, in which the Sit transporterwas the only system studied that individually contributed tovirulence, in vitro only the MntH transporter was shown to individuallycontribute to resistance to oxidative stress.

DISCUSSION

Top

DISCUSSION

In this report, we investigated the individual and combinedroles of the SitABCD, MntH, and FeoB transporters in the virulenceof APEC strain ${chi}$ 7122. These transport systems are mainly involvedin the import of manganese and/or ferrous iron. MntH is a proton-dependentNRAMP-related transporter that is highly selective for manganese(33, 41), FeoB is a GTPase that functions in ferrous iron transport(20, 31), and SitABCD (YfeABCD in Yersinia spp.) is an ABC transporterthat mediates uptake of manganese and iron (3, 33, 56).

Iron plays a number of vital functions in bacterial cells andis a cofactor of numerous enzymes (1). Importantly, iron playsa role in protection against oxidative stress, as it is a componentof the SodB superoxide dismutase (SOD) and catalase enzymes,which eliminate superoxide (O₂^–) and H₂O₂, respectively.However, excess iron can also contribute to oxidative damagethrough the generation of free radicals (1). Due to the limitedavailability of iron in the host, pathogenic bacteria have acquirednumerous iron transport systems, some of which are major virulencefactors (5). Bacteria obtain ferric iron from the host via eithersiderophore systems, low-molecular-weight chelators that solubilizeiron, or host iron transport protein receptors, whereas ferrousiron is transported by FeoB and SitABCD-like systems (5). Manganeseis also a cofactor for a number of bacterial enzymes, contributesto protection against oxidative stress (33), and can contributedirectly to the detoxification of ROIs (23). In E. coli andother enterobacteria, two main systems, MntH and SitABCD, havebeen identified as being involved in manganese transport (33,41, 56). Since iron and manganese are critical for bacterialtolerance to oxidative stress as well as other metabolic functions,SitABCD, MntH, and FeoB could potentially promote bacterialsurvival during infection by facilitating the transport of thesetrace metals and by countering oxidative stress during infectionof the host.

FeoB and MntH orthologs are widespread among a diverse populationof bacteria (20, 49), and these systems are conserved amongE. coli strains, as they are present in all the currently sequencedE. coli and Shigella genomes. Unlike feoB and mntH, sit genesare present in only a subset of E. coli strains. sit genes werepredominant among ExPEC and APEC isolates and E. coli strainsbelonging to phylogenetic group B2 but were absent from E. coliK-12 and most diarrheagenic E. coli strains, except for enteroaggregativeE. coli strain O42 (Fig. 1A; Tables 2 and 3). Among avian E.coli strains, we determined that sit genes were highly associatedwith the virulence of APEC for 1-day-old chicks (Table 3). Thisis the first report describing an association of sit genes withthe virulence of APEC. In addition, sit genes were significantlyassociated with APEC (93%) compared to avian fecal commensalisolates (44%) (Table 3). Similarly, Rodriguez-Seik et al. (51)detected sit in 86.4% of APEC strains, compared to 42.7% offecal commensal isolates from poultry. Overall, sit sequencesare clearly prevalent among ExPEC and APEC strains comparedto intestinal commensal E. coli strains.

In some E. coli strains, two or more copies of the sit genesare present, with one copy being carried on a plasmid and anothercopy being chromosomally carried (17, 27-29). Recently, Ewerset al. (17) also detected the specific presence of episomaland chromosomal copies of sit by PCR using specific primers.They detected 31.6% prevalence of sit among APEC strains byusing chromosome-specific primers and 73.2% prevalence of sitamong APEC strains by using episome-specific primers. Theseresults suggest that APEC strains more commonly contain a plasmid-carriedcopy of sit than chromosomally carried sit genes. Chromosomalcopies of sit genes are associated with different phage elements(Fig. 1A). Hence, in certain strains, sitABCD represents anothervirulence factor that may have been acquired via bacteriophages,which are common contributors to bacterial diversification andadaptability (8, 46). It is tempting to speculate that recombinationor integration events within a single strain may have led tothe duplication of sit genes. However, comparative sequenceanalysis demonstrates higher identity among sit sequences locatedin genomic islands from different strains than between genomicisland- and plasmid-carried sit sequences from the same strain(Fig. 1B). These findings favor the likelihood that sit genesmay have been acquired independently from different sourcesby distinct events resulting in the incorporation of these genesinto either plasmids or prophage-associated genomic islandsin the genomes of Shigella and ExPEC.

The increased association of sit genes with ExPEC and APEC comparedto other E. coli strains suggested a possible role for thistransporter in virulence during extraintestinal infection. InUPEC strain CFT073, sit genes were upregulated in the urinesof mice during urinary tract infection or in vitro followinggrowth in human urine (63). Also, in Shigella flexneri 2a, theexpression of sit genes was upregulated during infection ofcells (39, 54). This further suggests that SitABCD may contributeto metal transport in vivo and to virulence during extraintestinalinfections. In both the single-strain infection and competitiveinfection models, the APEC ${Delta}$ sitABCD mutant was clearly attenuatedand less able to colonize the tissues and persist in the bloodof infected chickens (Fig. 2 to 4). In contrast, the single-straininfection and coinfection results demonstrated that in the presenceof a functional SitABCD transporter, the inactivation of FeoBor MntH did not reduce the virulence of APEC strain ${chi}$ 7122. Inaddition, the combined loss of feoB and mntH did not reducevirulence. Hence, in contrast to the MntH and FeoB transporters,the Sit transporter contributed significantly to the survivalof APEC in extraintestinal tissues. These results indicate thatthe divalent-ion transporter SitABCD imparts a selective advantagecompared to the MntH or FeoB transporter in vivo. The infectionmodel used for these studies was the air sac inoculation model(15, 16), which has been shown to result in a more uniform rateof infection of extraintestinal tissues and blood than thatobtained with the more natural aerosol route of infection (18,64). However, by using the air sac route in the single-straininfection model, despite the significantly decreased survivalof the ${Delta}$ sitABCD mutant in chickens, this strain caused grosslesions of airsacculitis and pericarditis/perihepatitis thatwere similar to those caused by the wild-type parent. This islikely because the attenuated strains remained in the tissuesat levels of 3 log or above at 48 h postinfection in most birds.Such bacterial levels are likely to be sufficient to still elicita strong inflammatory response and the generation of gross lesions.It is possible that the use of the aerosol route in single-straininfection experiments could potentially demonstrate differencesin pathology of colibacillosis between strains that were notobserved via the air sac route, as the aerosol route requiresAPEC to initially colonize and proliferate in the upper respiratorytract and then subsequently invade deeper tissues.

The use of the competitive infection model provided increasedsensitivity compared to that of the single-strain infections.This was most evident in analyses of mutants that had lost bothSitABCD and MntH divalent manganese transport functions or SitABCDand FeoB divalent iron transport functions (Fig. 3 and 4). Nosignificant difference at any site was observed in the single-straininfections by comparing these double mutants to the mutantswhich lacked the Sit transporter alone. This is likely due tothe combination of competitive pressure between strains, inaddition to pressures due to immune defenses and nutritionallimitations within the host, in the competitive infection model(52, 65).

The contribution of Sit to the virulence of APEC may be duein large part to its function as an efficient manganese transporter(56) and to the importance of manganese for coping with oxidativestress. Cellular accumulation of manganese contributes to protectionagainst oxidative stress by direct catalytic scavenging of ROIsand can also lead to increased activity of bacterial manganese-dependentSOD (MnSOD) (23). In E. coli, MnSOD, encoded by sodA, is aninducible SOD that responds to increases in oxidative stress,and MnSOD more effectively protected E. coli against paraquatand H₂O₂ than did FeSOD, encoded by sodB (7). An efficient capacityto accumulate Mn by APEC via the Sit and MntH transporters istherefore likely to contribute to survival in host extraintestinalsites. The infection results demonstrated that loss of MntHalone had no discernible effect on virulence in the chicken,whereas loss of Sit alone did reduce virulence. Conversely,MntH alone contributed to protection of APEC against ROIs, whereasSit alone had no significant contribution to resistance to oxidativestress in vitro (Fig. 5). This may be explained by differencesin environmental conditions, which may affect the activity ofeach of these transport systems. MntH transports metals viaa proton-dependent gradient and, as such, functions best underacidic conditions (32). In contrast, Sit-mediated transportis most efficient under alkaline conditions (32, 56). The pHin chicken blood and tissues is maintained at approximately7.4, a pH at which Sit efficiently mediates Mn transport (4,56). It is also possible that the ATPase-mediated Sit ABC transportermay be more effective at tolerating stresses within the hostthat could impede the proton gradient required for proper functioningof MntH-mediated transport. Although Sit alone appeared to bemore important than MntH in vivo, it is clear that the combinedloss of the Sit and MntH transporters resulted in the greatestattenuation, particularly in the lung, and in greater sensitivityto a number of ROI-generating compounds. Taken together, theseresults support a synergistic role for Sit and MntH to transportmanganese and contribute to a greater overall resistance tooxidative stress and persistence during infection.

Loss of Sit and MntH rendered the APEC strain more susceptibleto a number of ROI-generating compounds, particularly on minimalmedium (Fig. 5). Manganese transporters in a number of bacteriahave been shown to contribute to resistance to H₂O₂ and superoxide(23). As observed for Salmonella enterica serovar Typhimuriumand Shigella flexneri 2a mntH sit mutants (4, 53), the APEC ${Delta}$ mntH ${Delta}$ sitABCD mutant was more sensitive to H₂O₂ (Fig. 5) (56).In addition, the APEC ${Delta}$ mntH ${Delta}$ sitABCD mutant was also more sensitiveto certain redox cycling agents, including plumbagin and paraquat.These compounds generate superoxide radicals, but superoxidecan be converted readily into H₂O₂ by SODs within bacterialcells. The use of ${Delta}$ oxyR and ${Delta}$ soxS K-12 control strains, whichwere both more sensitive to plumbagin and paraquat, also suggestedthat subsequent generation of H₂O₂ occurred following exposureto these compounds. In contrast, neither APEC ${Delta}$ mntH ${Delta}$ sitABCD northe ${Delta}$ oxyR control was more sensitive to phenazines (PMS and PES)than the relevant parent strain, although these compounds moreeffectively killed the ${Delta}$ soxS control. Differences in sensitivityto redox cycling agents may therefore be due to their mechanismsof action for generation of oxidative stress or possibly othertoxic effects. In E. coli, the response to oxidative stressgenerated by PMS differs considerably from that generated byparaquat or plumbagin, with PMS generating a high level of catalaseactivity and little increase in SOD activity, whereas paraquatand plumbagin induce increases in both SOD and catalase activities(21, 60).

Although the Sit transporter from APEC strain ${chi}$ 7122 was shownto mediate uptake of iron in an iron transport-deficient E.coli K-12 mutant (56), it is unlikely that iron transport bySitABCD is a key feature of APEC virulence. Strain ${chi}$ 7122 usesthree confirmed siderophore systems (aerobactin, salmochelins,and enterobactin) in addition to the Feo and SitABCD transporters.Elimination of the aerobactin and salmochelin siderophore systemsresulted in a nearly complete loss of persistence of APEC strain ${chi}$ 7122 in chickens (15), suggesting that the remaining iron transportsystems (FeoB, SitABCD, and enterobactin) were unable to meetthe iron transport requirements for survival of APEC duringextraintestinal infection. This is not surprising, since FeoBand SitABCD function mainly as ferrous iron transporters (20,56), and iron present in extraintestinal tissues is predominantlyferric iron associated with host ferro-proteins. Furthermore,although enterobactin is a highly efficient siderophore in vitro,it competes poorly as a siderophore in host tissues (1, 5).The finding that the cumulative loss of the Sit and Feo ferrousiron transporters did not result in greater attenuation thanthat of the ${Delta}$ sitABCD mutant is in further support of a limitedrequirement of these systems for iron acquisition in extraintestinaltissues (Fig. 3). However, the combined loss of Sit and Feoresulted in a decreased capacity to persist in the blood (Fig.4). It is possible that in the blood, where iron availabilityis minimal, the ferrous transport functions of Feo and Sit incombination with efficient ferric siderophore systems couldprovide a modest competitive edge. Furthermore, the combinedloss of Feo and Sit also increased the sensitivity to hydrogenperoxide (Fig. 5), suggesting the cumulative importance of thesetwo transporters under certain conditions. In contrast, combinedloss of Sit and Feo actually resulted in a significant regainin colonization of the liver compared to that of the ${Delta}$ sitABCDmutant in the competitive coinfection model (Fig. 3c). The liverfunctions in iron storage and recycling. During infection, increasedcell death and tissue necrosis in the liver may lead to localizedtissue anoxia and increased availability of ferrous iron atlesion sites. Under anaerobic conditions and with increasedavailability of ferrous iron, the Feo transporter may thereforebe deleterious. In support of this, compared to its wild-typeparent, an E. coli K-12 feo mutant grown anaerobically was moreresistant to H₂O₂-mediated killing than its isogenic parentstrain (34).

In APEC, the sit genes are commonly located on large conjugative(ColV or ColBM) plasmids (17, 27, 29, 56, 59). The roles ofsuch plasmids in various APEC or pathogenic E. coli strainshave been established in a number of reports (16, 18, 24, 61,62, 64), although specific plasmid-carried genes contributingto APEC virulence are less well characterized. The Sit transporter,along with the aerobactin and salmochelin siderophores (15)and Tsh (16), represents another plasmid-encoded virulence factorof APEC strain ${chi}$ 7122. Compared with the divalent manganese oriron transporter MntH or FeoB, the importance of SitABCD ironand manganese transport for virulence of APEC is preponderant.Hence, in addition to providing an advantage for intracellularsurvival for Salmonella enterica and Shigella flexneri 2a, inAPEC Sit appears to provide an adaptive advantage during extraintestinalsurvival. The reduction in virulence together with the increasedsensitivity to ROI-generating compounds observed for some ofthe APEC metal transport mutants suggests that reduced survivalduring infection could possibly be due to increased killingby phagocytes or the products they may liberate into the extracellularenvironment. Although APEC strain ${chi}$ 7122 is refractory to internalizationby avian phagocytes, APEC strains belonging to other serogroups,such as O2 or O1, are more readily internalized and can survivewithin phagocytes (44). In future studies, it will be interestingto determine if Sit contributes to increased survival of certainAPEC strains within host cells and its potential role in thevirulence of ExPEC in human extraintestinal infections.

ACKNOWLEDGMENTS

We acknowledge Geneviève Garriss for construction ofstrain QT770.

Funding for research was provided by the Natural Sciences andEngineering Research Council (NSERC), Canada, the Canadian Foundationfor Innovation (CFI), and a Canada Research Chair (CRC) to C.M.D.and by USDA grant 00-35204-9224 to R.C. and C.M.D. M.S., M.C.,and M.H.L. were the recipients of Fondation Armand-Frappierscholarships. J.P. was funded by a scholarship from Fonds dela Recherche en Santé Québec (FRSQ).

FOOTNOTES

* Corresponding author. Mailing address: INRS-Institut Armand-Frappier, 531 boul. des Prairies, Laval, Québec, Canada H7V 1B7. Phone: (450) 687-5010, ext. 4221. Fax: (450) 686-5501. E-mail: charles.dozois{at}iaf.inrs.ca

Published ahead of print on 19 November 2007.

Editor: J. B. Bliska

REFERENCES

Top