INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Avian-pathogenic Escherichia coli (APEC) comprise a specific subset of pathogenic E. coli that cause extraintestinal diseases of poultry. Of the various forms of E. coli disease in poultry, the most common syndrome starts as a respiratory tract infection in 3- to 12-week-old broiler chickens and turkeys and frequently becomes more generalized. The air sacs are the first organs affected, and systemic spreading may result in pericarditis, perihepatitis, and an often fatal septicemia (15, 29). APEC infections are frequently enhanced or initiated by predisposing factors, which include environmental conditions and viral or Mycoplasma infection (15, 29). O1, O2, and O78 are the most commonly encountered serogroups among APEC (15, 29), and the majority of strains have been shown to belong to a limited number of clonal lineages (69, 70). APEC strains of high virulence are lethal for 1-day-old chicks when administered subcutaneously. Attributes associated with APEC strains include F1 (type 1) and P fimbrial adhesins (16, 21, 53, 66), resistance to serum and phagocytosis (21, 22, 52, 71), the aerobactin siderophore system (21, 41, 65), and colicin V (7, 23, 65, 71) (reviewed in references 15 and 29). Recently the tsh gene, encoding a temperature-sensitive hemagglutinin, first identified by Provence and Curtiss (54), was shown to be associated with APEC but not with E. coli isolated from the feces of healthy chickens (45).

The tsh gene was first identified from APEC O78:K80 strain 7122 and, when cloned into E. coli K-12, was shown to impart mannose-resistant hemagglutination of chicken erythrocytes if bacteria were grown at 26°C on low-osmolarity solid medium (54). The deduced protein encoded by the tsh gene exhibits 50% similarity to immunoglobulin A (IgA) proteases of Neisseria gonorrhoeae and Haemophilus influenzae (54) and, as demonstrated by Stathopoulos et al. (60), the Tsh protein was the first identified member of an expanding subclass of the IgA protease family of autotransporters present in Shigella spp. and numerous pathotypes of E. coli. Autotransporters are a family of autonomously secreted proteins from gram-negative bacteria that are processed as three functional domains, comprising a sec-dependent amino-terminal leader sequence, an extracellular or surface-secreted mature protein (passenger domain), and an outer membrane-associated carboxy-terminal -barrel domain that mediates secretion of the passenger domain (32, 44). These proteins exhibit diverse functions involved in virulence and include adhesins, proteases, cytotoxins, and cell invasion proteins (32, 44). Autotransporters recently identified from pathogenic E. coli or Shigella spp. include EspC (61) and AIDA-I (6) from enteropathogenic E. coli; EspP/PssA from enterohemorrhagic/Shiga toxin-producing E. coli (12, 17); Pet and Pic from enteroaggregative E. coli (24, 31); TibA from enterotoxinogenic E. coli (42); Hbp from E. coli associated with a human wound infection (49); and VirG/IcsA (62), SepA (5), and ShMu (56) from Shigella spp. With the exception of EspC, ShMu/Pic, and TibA, the genes encoding these various autotransporter proteins are located on plasmids.

The Tsh autotransporter is processed as a 106-kDa secreted domain, Tsh_s, exported through a 33-kDa -barrel domain, Tsh, and contains a serine protease motif but does not demonstrate detectable proteolysis of human IgA, chicken IgA, or casein (60). Recently a nearly identical Tsh protein, with only two amino acid differences (Q₂₀₉-K₂₀₉ and A₈₄₂-T₈₄₂), named Hbp, was shown to specifically degrade human hemoglobin and bind heme (49). Although the hemagglutination activity of Tsh occurs predominantly at 26°C, Tsh is also produced at temperatures as high as 42°C and in strain 7122 is increasingly liberated into the extracellular medium at higher temperatures (60). Currently, it is unknown whether Tsh contributes to the pathogenesis of APEC infection. Furthermore, the location of tsh within the genome of strain 7122 and other APEC strains has not been precisely determined.

This report determines the prevalence and the location of the tsh gene in APEC strains and investigates the possible role of tsh in the pathogenesis of avian respiratory colibacillosis.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and growth conditions. The APEC strain 7122 and derivatives, avian E. coli clinical isolates, E. coli K-12 strains, and plasmids used are presented in Table 1. Clinical isolates included 300 avian E. coli originating from chickens (117 isolates), turkeys (175 isolates), and ducks (8 isolates). They were collected in France (211 isolates) and Canada (89 isolates), mainly from lesions of colisepticemia and in a few cases from the vitellus of day-old turkey poults, over a period of 10 years. Serogroup was determined by slide agglutination with specific antisera raised against O1, O2, and O78 antigens (Biovac, Angers, France, or the Escherichia coli Laboratory, St. Hyacinthe, Québec, Canada). The number of isolates in each serogroup were 28 O1, 49 O2, 84 O78, and 139 other. Lennox (L) broth and L agar (47) were routinely used for growing E. coli strains and clones, and strain DH5 was routinely used for plasmid cloning and recovery. For infection studies, strain 7122 and derivatives were grown in brain heart infusion broth (BHI; Difco Laboratories, Detroit, Mich.). Ampicillin (100 µg ml¹), kanamycin (25 µg ml¹), chloramphenicol (25 µg ml¹), nalidixic acid (12.5 µg ml¹), and tetracycline (10 µg ml¹) were used as required unless indicated otherwise.

DNA and genetic manipulations. Total bacterial genomic DNA was prepared using a small-scale preparation method (3). Restriction endonucleases and DNA-modifying and ligase enzymes (New England Biolabs and Promega) were used according to the manufacturer's guidelines. Native plasmids from APEC strains were isolated as described by Kado and Liu (37) and separated by electrophoresis. Conjugation and transformation of bacterial cells were performed by standard techniques (47). Counterselection for loss of tetracycline resistance was achieved using a tetracycline-sensitive selective (TSS) agar containing fusaric acid (5 g of tryptone, 5 g of yeast extract, 10 g of NaCl, 15 g of agar, 10 g of NaH₂PO₄ · H₂O, 12 µg of fusaric acid, 50 µg of chlortetracycline HCl, and 13.6 µg of ZnCl₂ per liter of medium) (9).

PCR and DNA hybridization. The primers used for PCR amplification and generation of DNA probes are indicated in Table 2. The locations of primers used for probe generation and PCR analysis are presented in Fig. 1. DNA crude extracts prepared by a rapid boiling method were tested in a 25-µl PCR mixture containing 12.5 pmol (or 6.25 pmol each) of the forward and reverse primers, 5 nmol of each deoxynucleoside triphosphate, and 0.5 U of Taq DNA polymerase in 1× buffer. The PCR conditions were as follows: 94°C for 3 min; annealing as indicated in Table 2 for 1 min, and 72°C for 1 min for 1 cycle; 94°C for 1 min, annealing for 1 min, 72°C for 1 min for 26 cycles; and a final extension at 72°C for 10 min.

View larger version (22K): [in this window] [in a new window]   FIG. 1.   Restriction maps and genetic organization of (a) the tsh gene region and (b) the colicin V (ColV) gene cluster. Organization of the ColV gene cluster is derived from Gilson et al. (27). Genes are indicated below the restriction maps with thick black arrows that point in the direction of transcription. Forward (>) and reverse (<) primers for PCR amplifications of DNA from avian E. coli isolates and synthesis of DNA probes are indicated with numbers. Oligonucleotide primers are described in Table 2. DNA probes Tsh1 (620 bp), Tsh2 (616 bp), and ColV1 (1,203 bp) are indicated with solid lines above the restriction maps. The Tsh1 and Tsh2 probes were generated from plasmid pYA3108 (54), and the ColV1 probe was generated from pHK11 (27). Experimental procedures are detailed in the text.

DNA sequencing and analysis. Nucleotide sequences of the regions flanking the tsh gene from strain 7122 was determined using pYA3108, pYA3104, and derivatives of pYA3107 containing Tn5seq1 or TnphoA insertions (54) as the templates. Primers used for sequencing included pBluescript SK and KS, SP6, and custom-synthesized oligonucleotide primers. Sequencing was performed using ABI prism fluorescent Big Dye Terminators according to the manufacturer's instructions (PE Biosystems, Norwalk, Conn.) and a 480 thermal cycler (PE Biosystems). Sequencing gels were run at the Protein and Nucleic Acid Chemistry Laboratory of Washington University. Comparison of the DNA sequences and predicted open reading frames (ORFs) with sequences in the GenBank genetic sequence database were performed using the BLASTN, BLASTP, and BLASTX programs (2) accessed from the National Center for Biotechnology Information (NCBI). Putative ORFs within the sequence were selected using ORF Finder at NCBI.

Virulence assay for 1-day-old chicks. APEC isolates were classified for virulence based on lethality for 1-day-old chicks following subcutaneous inoculation as previously described (14). Lethality classes (LC) were defined as follows: LC1, 50% lethal dose (LD₅₀) < 10⁸ CFU; LC2, LD₅₀  10⁸ CFU; LC3, not lethal at 10⁸ CFU.

Construction of a tsh::tetAR(B) mutation in strain 7122 and selection of tetracycline-sensitive derivatives. In order to construct an isogenic tsh mutant of strain 7122, we inserted the tetAR(B) cassette derived from Tn10. To construct a tetAR cassette for insertion into the tsh gene, a 2.2-kb HindIII fragment encoding tetAR(B) from pSBA383 (56) was cloned into the HindIII sites of pBSL86 (1), replacing the nptII kanamycin resistance gene in this vector. The resulting plasmid, pYA3442, contained the tetAR(B) cassette flanked by convenient restriction sites. A PstI fragment bearing the tetAR cassette from pYA3442 was cloned into the compatible NsiI site of the tsh gene on plasmid pYA3418 (60), resulting in plasmid pYA3444. A 6.6-kb BssHII fragment containing the tetAR-interrupted tsh gene from pYA3444 was ligated to the AscI sites of suicide vector pMEG-375. The resulting plasmid, pYA3448, was used for allelic replacement of tsh in APEC strain 7122. The pYA3448 suicide vector containing the tsh::tetAR(B) insert was conjugated from E. coli MGN-617 to 7122 by overnight plate mating on L agar plus 50 µg of diaminopimelic acid per ml. Transconjugants were selected by growth on L agar plates containing tetracycline without diaminopimelic acid. Selection for double-crossover allele replacement was obtained by sacB counterselection on L agar plates without NaCl and containing 5% sucrose (38). A 7122 derivative, strain 7273, was confirmed to contain an insertion of the tsh::tetAR(B) allele resulting from a double crossover, as determined by absence of resistance to ampicillin and chloramphenicol encoded on the suicide vector, PCR amplification using specific oligonucleotide primers 9, 10, and 11 (Table 2), and lack of Tsh protein production as determined by Western blot (data not shown).

3

4

Experimental infection of chickens via the air sacs. Three groups of 10 3-week-old White Leghorn specific-pathogen-free chickens from the Institut National de la Recherche Agronomique experimental farm were reared in separate cages with food and water available ad libitum. Each chicken was inoculated in the right thoracic air sac with 0.1 ml (10⁷ CFU) of a bacterial inoculum consisting of a diluted 24-h BHI culture of E. coli 7122, 7273, or 7274. Blood samples (50 µl) were collected aseptically from each chicken 6, 24, and 48 h following bacterial inoculation and diluted 1:4 in phosphate-buffered saline (PBS, pH 7.4), and 0.1 ml was plated on Drigalski agar (Diagnostics Pasteur, Marnes la Coquette, France) supplemented with nalidixic acid (40 µg ml¹) or with nalidixic acid and tetracycline in the case of E. coli 7273. Another 50-µl volume of blood was incubated in 2 ml of BHI for qualitative detection of E. coli. Positive growth of E. coli in BHI was confirmed by plating enriched cultures on Drigalski agar.

Serum bactericidal assay. Bacterial survival in chicken serum was determined as previously described (21), with an initial bacterial inoculum of approximately 10⁷ CFU ml¹ incubated in fresh 90% normal chicken serum. The serum-bacterium suspensions were incubated at 37°C for 3 h in a 5% CO₂ atmosphere, and counts of viable cells were estimated at the 1- and 3-h time points.

Statistical analyses. The prevalence of the tsh gene among the different E. coli isolates relative to virulence was analyzed using the chi-square test. In experimental-infection assays, lesion scores and bacterial counts were compared by analysis of variance between groups of chickens; the chi-square test was used to compare the number of contaminated chickens in the case of qualitative detection of E. coli.

Nucleotide sequence accession number. The tsh DNA region of pAPEC-1 of E. coli strain 7122 has been entered as GenBank nucleotide accession number AF218073.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Association of tsh with lethality of avian E. coli isolates. To determine the prevalence of tsh among avian E. coli, 300 clinical isolates from poultry were examined by PCR amplification using the tsh gene-specific primer set Tsh1 (Fig. 1 and Table 2). The presence of the tsh gene was detected in about half of the E. coli isolates tested (49.7%). Its occurrence among isolates from either diseased chickens or turkeys, irrespective of their geographic origin (France or Québec), was significantly associated (P < 0.001) with high lethality for chicks. Among isolates belonging to the high-lethality class (LC1), 61.6% were tsh positive, whereas 30.0 and 9.8% tsh-positive isolates were found in the low lethality class (LC2) and in the nonlethal class (LC3), respectively. When considering tsh-positive isolates, 90.6% belonged to LC1, whereas only 6.0 and 3.4% were LC2 and LC3, respectively.

Experimental infection of chickens with strain 7122 and derivatives. To determine whether tsh contributes to the pathogenesis of respiratory colibacillosis, wild-type APEC strain 7122 and tsh mutant derivatives 7273 and 7274 were compared in a chicken experimental infection model. In contrast to chickens inoculated with the wild-type strain 7122, chickens inoculated with the tsh::tetAR mutant 7273 exhibited fewer and less pronounced lesions (P < 0.01) in the air sacs (Table 4). However, compared to strain 7122, mutant 7273 caused similar lesions of pericarditis and perihepatitis and persisted in organs and blood to a similar degree (Table 4). Birds infected with mutant 7273 complemented with tsh on plasmid pYA3108 exhibited air sac lesions similar to those seen in chickens infected with wild-type strain 7122 (data not shown). As with mutant 7273, mutant 7274 caused reduced lesions in the air sacs of birds, but, interestingly, it was much more attenuated. Strain 7274 caused few lesions of pericarditis and perihepatitis, did not persist in the blood, and poorly colonized the lung, spleen, and liver (Table 4). These results suggest that other genes linked to tsh that contribute to APEC pathogenesis were concurrently lost following loss of the tsh::tetAR allele from strain 7274 by fusaric acid selection.

View this table: [in this window] [in a new window]   TABLE 4.   Comparison of ability of APEC strain 7122 and derivatives to induce colibacillosis lesions and persist in internal organs and blood of chickens experimentally inoculated via the air sacsa

Association of the tsh gene with ColV plasmids in APEC isolates. In a recent report by Otto and coworkers (49), a second tsh allele encoding a protein termed Hbp was shown to be located on a ColV-type plasmid of an E. coli O8 strain isolated from a human wound abscess. To determine whether tsh is located on ColV plasmids in APEC isolates, plasmid extracts from 14 highly lethal tsh-positive APEC isolates from Québec, Canada, belonging to different serogroups (eight O78, two O2, and one each O1, O35, O22, and O45) and from strains 7122 and tsh mutants 7273 and 7274 were tested by Southern hybridization using the Tsh1 and ColV1 probes (Fig. 1). Plasmid extracts of the APEC isolates demonstrated one or more plasmids of high molecular weight (Fig. 2A). Wild-type APEC strain 7122 and tsh insertion mutant 7273 contained three large plasmids, whereas its fusaric acid-selected Tc^s revertant 7274 had lost the largest of these three plasmids, which we termed pAPEC-1 (Fig. 2A). With the exception of strain 7274, the Tsh1 probe hybridized to one plasmid of various sizes from each of the APEC isolates and also hybridized to plasmid pColV-K30, which was used as a reference. Furthermore, for all but four of the APEC isolates, the same plasmid that hybridized to the Tsh1 probe also hybridized to the ColV1 probe. For the other four isolates, plasmid extracts did not hybridize to the ColV1 probe. PCR amplification using the ColV1 primer set (Fig. 1) demonstrated the same results as the Southern blots (Fig. 2A). All except two of the APEC isolates analyzed, including strain 7122, were positive for the aerobactin system (data not shown). Unlike its wild-type parent, strain 7274 no longer contained tsh-, colicin V-, or aerobactin gene-specific sequences, indicating that these genes are encoded on plasmid pAPEC-1 in strain 7122.

View larger version (68K): [in this window] [in a new window]   FIG. 2.   (A) Localization of tsh to ColV-type plasmids in APEC isolates and pColV-K30. Analysis of plasmids of tsh-positive APEC isolates, 7122, and derivatives by ethidium bromide staining (upper panels) and Southern hybridization of plasmid extracts (lower panels). Asterisks indicate samples from ColV-negative isolates obtained by PCR. Hybridizations depicted are with the Tsh1 probe. Hybridization with the ColV1 probe is marked below the lower panels as positive (+) or negative (). In positive samples, the ColV1 probe hybridized with the same plasmid as the Tsh1 probe. Arrows indicate the pAPEC-1 plasmid containing tsh in APEC strain 7122. Lane 1, 7273; lane 2, 7274; lanes 3 and 11, pColV-K30; lanes 4 and 12, 7122; lane 5, TK27; lane 6, CN30; lane 7, TK40; lane 8, TK60; lane 9, CN137; lane 10, CN139; lane 13, CN144; lane 14, CN151; lane 15, CN163; lane 16, CN165; lane 17, CN69; lane 18, TK49; lane 19, CN14; lane 20, CN71. CN (isolated from chicken) and TK (isolated from turkey) samples are clinical isolates from Québec, Canada. (B) Replacement of plasmid pAPEC-1 from strain 7122 with a Tn10-tagged pColV-K30 plasmid; plasmid profiles of donor, recipient, and transconjugant strains. Lane 1, donor strain MEG-617(pColV-K30::lacZ); lane 2, recipient 7122 native plasmid profile before conjugation; lane 3, strain 7275, a transconjugant of 7122 mated with MEG-617(pColV-K30::lacZ) following selection on medium containing tetracycline and nalidixic acid.

Bactericidal effect of serum. To determine whether tsh or the pAPEC-1 plasmid contributes to serum resistance, APEC strain 7122 and derivatives as well as an E. coli K-12 strain, 7276, and a K-12 transconjugant containing pAPEC-1 (tsh::tetAR), 7277, were tested for survival in normal chicken serum (Fig. 3). Wild-type APEC strain 7122 was the most serum-resistant strain and exhibited a 10-fold growth increase after 3 h of incubation (Fig. 3). In addition, the tsh insertion mutant 7273 and pAPEC-1 (fusaric acid)-cured derivative 7274 were also serum resistant and exhibited a two- to threefold increase in growth after 3 h of incubation (Fig. 3). In contrast, the K-12 E. coli strain 7276 was serum sensitive and decreased in viability nearly 10⁵-fold after 3 h of incubation. The same K-12 strain bearing pAPEC-1 (tsh::tetAR), 7277, was also serum sensitive, but to a lesser extent than strain 7276, and demonstrated a 10³-fold decrease in viability after 3 h of incubation (Fig. 3).

View larger version (20K): [in this window] [in a new window]   FIG. 3.   Effect of 90% normal chicken serum on survival of APEC strain 7122 and derivatives and E. coli K-12 with and without the pAPEC-1 plasmid. Strains: 7122, APEC wild-type strain; 7273, 7122 tsh::tetAR; 7274, 7273 pAPEC-1; 7276, E. coli K-12; and 7277, 7276(pAPEC-1). Results are from a representative experiment from three independent assays.

Proximity of tsh to the ColV gene cluster in APEC isolates. Initial Southern hybridizations of HindIII- or EcoRI-digested 7122 DNA with the Tsh1 or ColV1 DNA probes demonstrated that both probes hybridized to either a 20-kb HindIII fragment or a 15-kb EcoRI fragment (data not shown), suggesting that tsh and the ColV gene cluster are in close proximity on plasmid pAPEC-1. Furthermore, restriction maps of pYA3107 (54) and of the ColV gene cluster of pColV-K30 (26) (Fig. 1) suggested that the ColV gene cluster was situated downstream of tsh, with the cvi gene proximal to tsh. As single KpnI sites are present within the tsh gene and at the end of the ColV gene cluster (Fig. 1), KpnI digests of genomic DNA from eight tsh-positive and ColV-positive APEC isolates (7122, TK27, CN30, TK40, TK60, CN144, CN163, and CN165) were hybridized with DNA probes ColV1 and Tsh2 (Fig. 1) to determine if these genes are closely linked in other APEC isolates. DNA digested with KpnI from all the APEC isolates except TK60 exhibited hybridization of Tsh2 and ColV1 probes to a single KpnI fragment that varied in size from 11 to 8.5 kb. Based on the locations of the KpnI sites within tsh and the ColV gene cluster (Fig. 1), in these isolates the suggested distance from the end of the tsh gene to the start of the ColV gene cluster is between 3.2 and 5.7 kb. In contrast, KpnI digests of DNA from strain TK60 exhibited multiple fragments that hybridized with the Tsh2 or ColV1 probe. These results indicate that, with the exception of isolate TK60, the tsh and ColV genes are closely located on the ColV plasmids of the APEC isolates examined.

Sequence analysis of the tsh region of strain 7122. Including the previously sequenced 4,699-bp region containing tsh (54), a total of 10,587 bp of the tsh region of pAPEC-1 was sequenced (Fig. 4). With a G+C content of 50.4%, the region is similar to the E. coli K-12 mean G+C content (50.8%) (8). Regions flanking the tsh gene exhibit identities to insertion sequence (IS) elements IS91, IS911, IS100, IS30, and IS1294, bacteriophage N15, and prophage 933L (50) DNA, and the def (fms) gene of E. coli K-12 (Fig. 4). In addition, the sequenced tsh region of plasmid pAPEC-1 starting at position 2760 and including the tsh gene is 99.8% identical to the sequenced portion of a ColV plasmid that encodes the Hbp autotransporter from E. coli strain EB1 (49) (Fig. 4), which is identical to Tsh of strain 7122 except for two residue substitutions.

View larger version (12K): [in this window] [in a new window]   FIG. 4.   Organization and analysis of the tsh region of pAPEC-1 from strain 7122. The scale is given in kilobase pairs. ORFs are represented by arrows above the scale. Solid arrows represent complete coding (black arrows) or interrupted (grey arrow) sequences showing high identity to known genes or IS elements. Putative ORFs are represented by hatched arrows. Positions and predicted lengths of the ORF products are presented in Table 5. Horizontal lines below the scale represent sequences exhibiting nucleotide identity to the specified region. The percent nucleotide identity and accession numbers of sequences exhibiting nucleotide identity to the represented regions are as follows: IS91, 100% (K04543); IS100, 97% (Z32853); IS30, 100% (X00792); "pColV-K30" of E. coli EB-1, 99.8% (AJ223631); IS911, 100% (X17613); bacteriophage N15, 79% (AF064539); IS1294, 92% (X82430); def (fms), 97% (X63666); prophage 933L, 99% (AF097644); she PAI, 97% (U97493); pic, 100% (AF097644); and pssA, 98% (Y13614).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The virulence of APEC is associated with the presence of unique DNA regions in the chromosome that are absent from E. coli K-12 strains (11). The tsh gene, which is absent from E. coli K-12, was identified in APEC strain 7122 and encodes a protein that shows homology to the IgA proteases of Haemophilus and Neisseria spp. (54) and was more recently shown to be processed and secreted as an autotransporter (60). The autotransporters are a family of secreted proteins from gram-negative bacterial pathogens, and many of these proteins have been implicated as actual or probable virulence factors (32, 44). Maurer et al. (45) noted that tsh was present in APEC strains but not in E. coli isolated from the feces of healthy birds, suggesting that tsh could be associated with the virulence of E. coli in chickens.

Herein, we have confirmed that tsh is associated with the virulence of avian E. coli isolates. Of the 300 avian E. coli isolates examined in the current report, half of the isolates were tsh positive, and tsh was specifically more frequent (P < 0.001) in high-lethality isolates compared to low-lethality isolates. In addition, of the tsh-positive isolates identified, most (90.6%) were from the high-lethality group (Table 3). The presence of tsh was similar among isolates from diseased chickens and turkeys and was not any more associated with serogroups O1, O2, and O78, commonly incriminated in avian colibacillosis, than with isolates from other serogroups (Table 3). In the tsh-positive strains examined, tsh was always plasmid encoded and was linked to colicin V genes, when they were present, on the same plasmid. As we have demonstrated that tsh is encoded on a ColV-type transmissible plasmid related to pColV-K30 in strain 7122, it is likely that tsh is also encoded on transmissible plasmids in other APEC isolates. As such, the presence of tsh among diverse serogroups of virulent APEC isolates is not surprising. At present it is unknown whether tsh may also occur on the chromosome of certain strains.

The association of tsh with lethal APEC isolates suggested that tsh may be a virulence factor and/or could be physically linked to some independent virulence determinant(s). Experimental infection of chickens with an isogenic tsh knockout derivative of APEC strain 7122 demonstrated that Tsh contributes to the development of lesions and fibrin deposition in the air sacs. In preliminary studies on the dynamics of air sac infection by E. coli, we have determined that the onset of airsacculitis in chickens is more rapid for APEC strain 7122 than for the tsh mutant 7273, suggesting that Tsh increases the rate of colonization at this site and development of airsacculitis. Tsh was first identified as a temperature-sensitive hemagglutinin for chicken erythrocytes (54), suggesting that it may act as an adhesin, particularly in the initial stages of colonization of the avian respiratory tract. The in vivo results from the infection studies further support the likelihood that Tsh plays a role in colonization in the air sacs. Other autotransporters such as Pic, TibA, and Hap also act as adhesins or hemagglutinins (31, 33, 42). In the case of Pic, which is closely related to Tsh, it has recently been suggested that it is involved in the early stages of pathogenesis and most probably promotes intestinal colonization by enteroaggregative E. coli (31).

In addition to its potential role as an adhesin, Tsh may act as a protease on a specific substrate in the air sacs. Tsh belongs to a group of autotransporter proteins that contain serine protease sites (32). A number of these proteins have been shown to exhibit protease activity against substrates such as casein (PssA) (17), pepsin A and coagulation factor V (EspP) (12), and gelatin (Pic) (31). In our hands, Tsh did not cleave human or chicken IgA, casein, or pepsin A (60). However, Otto et al. demonstrated that the gene product of the tsh allele (hbp) from E. coli strain EB1 degrades hemoglobin (49).

The hemoglobin-degrading and heme-binding properties of Tsh do not appear to be required for APEC 7122 infection of deeper tissues. Although it has not been demonstrated that heme bound to Tsh (Hbp) can be utilized by extraintestinal E. coli strains (49), the presence of the aerobactin siderophore on the pColV plasmid in strain 7122 (unpublished data), as well as on most other APEC strains (18, 65), likely compensates for inactivation of Tsh or obviates any such role for Tsh in the survival of APEC in blood and systemic tissues. Whether Tsh-mediated proteolysis of a substrate within the air sacs results in lesion formation and fibrin deposition remains to be demonstrated.

DNA sequencing of the region flanking the tsh gene on plasmid pAPEC-1 of strain 7122 identified IS elements and phage-related DNA (Fig. 4). As has been observed for other plasmids and regions flanking pathogenicity islands, recombination has occurred in the sequenced regions flanking tsh in strain 7122, as demonstrated by sequential insertion of different IS elements (Fig. 4). However, the region between the tsh gene and ColV gene cluster in the APEC isolates examined appears to be quite conserved, based on the presence and similar distance of phage 933L-related DNA 3' of tsh and linkage analysis between tsh and ColV sequences. Sequences similar to those flanking tsh have been identified adjacent to pathogenicity islands or on virulence plasmids in other enterobacteria. In particular, IS100 sequences are present on plasmids of Yersinia spp. (35, 43), the EAF plasmid of enteropathogenic E. coli strain B171 (64), adjacent to certain high-pathogenicity islands of Y. pestis and Yersinia pseudotuberculosis (30), and within pathogenicity island 5 (PAI-5) of uropathogenic E. coli J96 (63). P4-like cryptic prophage-related sequences similar to prophage 933L of enterohemorrhagic strain EDL933 (50) are frequently associated with genes that were probably acquired through horizontal transfer, including Shigella flexneri pathogenicity island 2 (48, 67), PAI-6 of uropathogenic E. coli strain CFT073 (20, 39), the 3' junction of PAI-1 from uropathogenic E. coli 536 (40), and the EAF plasmid of certain enteropathogenic E. coli strains (10) (accession no. AF119170). The presence of P4-related phage genes adjacent to pathogenicity islands and virulence genes suggests that intact phages or portions of these sequences may have mediated horizontal acquisition of these virulence genes through recombination (20). Interestingly, the genes encoding the ShMu, Pic, and PssA autotransporters, which are closely related to Tsh, are flanked by the same phage 933L DNA adjacent to tsh in strain 7122 (Fig. 4) and most other APEC isolates in this study. Furthermore, as with tsh from strain 7122, pic is also flanked by IS911-related sequences at its 5' end (31). The presence of these common flanking sequences suggests that in addition to sharing protein similarities with these autotransporters, tsh, she, pic, and pssA may have been acquired by different E. coli pathotypes through similar prophage- or recombination-mediated mechanisms.

The demonstration that tsh is encoded on large plasmids, usually containing the colicin V gene cluster, in APEC isolates (Fig. 2) and the decreased pathogenicity of strain 7274, which has lost the ColV-type plasmid bearing tsh (Table 4), strongly suggest that tsh is linked to other genes that contribute to APEC pathogenesis. It is well established that ColV plasmids may enhance the virulence of extraintestinal E. coli (58, 59, 68), and virulent APEC isolates are more often ColV positive than less-virulent clinical isolates or E. coli from healthy birds (7, 18). Ike et al. (36) demonstrated that curing of plasmid pKI100 from an APEC strain of serogroup O2 resulted in loss of lethality for chicks, a decrease in serum resistance, and loss of aerobactin hydroxamate siderophore expression. Recently, a conjugative plasmid encoding a hydroxamate but no colicin activity was shown to contribute to respiratory tract colonization and virulence of an APEC strain following aerosol infection of chickens (28). The traits and virulence determinants associated with ColV plasmids include the aerobactin iron-sequestering siderophore system, resistance to killing by serum complement and phagocytosis, motility, and adherence to intestinal cells (68). In APEC strains, the most clearly established of these traits is the presence of the aerobactin operon, which is often located on ColV plasmids in these strains (18, 65). In addition, in APEC strains, ColV production is associated with serum resistance (71), which may be encoded by genes such as iss or traT (68).

Loss of the ColV-type plasmid pAPEC-1 from strain 7122 was achieved by fusaric acid counterselection for loss of tetracycline resistance using the tsh::tetAR(B) derivative strain 7273. Infection experiments with the pAPEC-1-cured strain 7274 and the tsh insertion mutant 7273 clearly demonstrated that pAPEC-1 contains genes in addition to tsh that are involved in the pathogenicity of the wild-type strain 7122 in the lower respiratory tract and extrarespiratory tissues of experimentally infected chickens (Table 4). Although the presence of pAPEC-1 did increase the ability of a K-12 strain to survive in serum by about 100-fold, pAPEC-1 appears to play a limited role in the serum resistance of strain 7122, as the pAPEC-1-cured derivative 7274 was also resistant to the bactericidal effects of 90% chicken serum (Fig. 3). However, unlike strain 7122, 7274 has lost the aerobactin system encoded on pAPEC-1. Moreover, it is possible that attenuation of strain 7274 is partly due to a reduced ability to obtain iron from the iron-restrictive environment of the avian host following loss of the aerobactin siderophore system. Construction of aerobactin and aerobactin-Tsh double-knockout mutants would further elucidate the role of iron acquisition by aerobactin and possibly of the Tsh heme-binding hemoglobin protease in the pathogenesis of E. coli infection in poultry.