The Shiga Toxin 1-Converting Bacteriophage BP-4795 Encodes an NleA-Like Type III Effector Protein

Abstract

In this study, the complete DNA sequence of Shiga toxin 1-converting bacteriophage BP-4795 was determined. The genome of BP-4795 consists of 85 open reading frames, including two complete IS629 elements and three morons at the end of its late regulatory region. One of these morons encodes a type III effector that is translocated by the locus of enterocyte effacement-encoded type III secretion system into HeLa cells, where it localizes with the Golgi apparatus.

Enterohemorrhagic Escherichia coli (EHEC) and enteropathogenic E. coli are able to colonize the gastrointestinal tract of their host by the formation of attaching and effacing lesions (8). These lesions are mediated by a type III secretion system (T3SS), which is encoded in a chromosomal pathogenicity island, termed the “locus of enterocyte effacement” (LEE). The LEE genes are organized into five polycistronic operons (LEE1 to -5) (19). LEE1, LEE2, and LEE3 encode the structural components of the T3SS, whereas LEE4 encodes particular translocator and effector proteins. The proteins intimin and Tir, necessary for intimate attachment of the bacteria to host cells, are encoded by LEE5. The T3SS spans both bacterial membranes as a macromolecular complex and builds a needle structure upon contact with the host cell to transfer effector proteins directly into eukaryotic cells (8). The Tir protein is inserted in the host cell membrane and functions as receptor for the outer membrane protein intimin (16). A number of different intimin types have been identified to date, which basically have been classified as intimin-α, -β, -γ, -δ, -ɛ, and -ζ (summarized in references 1 and 34). Tir and other T3SS-translocated effectors subvert the host cell metabolism and are involved in signal transduction and rearrangement of the eukaryotic cytoskeleton to facilitate the survival of the pathogen (9). Some recently described effector proteins, i.e., Cif (18), NleA/EspI (11, 22), EspF_U/TccP (3, 10), and EspJ (5), are encoded outside the LEE and linked with prophage sequences. Moreover, a number of non-LEE-encoded type III effectors, such as NleB, -C, -D, -E, -F, and -G, have been identified in Citrobacter rodentium (7).

Besides causing attaching and effacing lesions, EHEC strains produce Shiga toxin (Stx), which is considered their major virulence factor. Stx is encoded in the late region of lambdoid prophages close to the 3′ end of the antiterminator gene Q (24, 31, 32). The ability of these bacteriophages to transmit virulence genes by horizontal gene transfer may play a key role in the emergence of new EHEC pathotypes and for the distribution of virulence genes to E. coli strains (29).

Genetic structure of BP-4795.

In a previous study, E. coli O84:H4 strain 4795/97, isolated from the stool of a patient with diarrhea, attracted attention because it encodes the rare intimin variant ζ (34). Due to our interest in characterizing the genetic diversity of Stx phages, we decided to determine the complete nucleotide sequence of Stx1-encoding bacteriophage BP-4795, isolated from this strain. Phage transduction, preparation of phage DNA, cloning, hybridization, sequencing, and sequence analysis were basically performed as previously described (23, 27, 30). By assembly of all sequence contigs obtained, a phage genome with a circular permutation was revealed. Analysis of the 57,930-bp sequence of BP-4795 revealed an overall G+C content of 50.6%, containing 85 open reading frames (ORFs). The genetic organization of the sequenced region is depicted in Fig. 1.

FIG. 1.

Genetic map of the prophage BP-4795 (57,930 bp). Bars above the black line indicate ORFs in a 5′ to 3′ transcription direction and bars below the line depict ORFs transcribed in 3′ to 5′ direction. The shading of the bars indicates highest sequence identity to genes of related phages. Black bars, 933W; dark gray bars, 933V and H19B; middle gray bars, CP-933K, -M, -N, -P, -R, and -U; light gray bars, Stx2 phage I, Stx1 phage, and Stx1 phages phi-O153 and CP-1639, as well as some less-characterized phage genes of E. coli; and hatched bars, genes without homology in the NCBI database or with identity to gram-negative bacteria other than E. coli. Gene designations are taken from AJ556162. IS629 elements are labeled.

The genomic structure of BP-4795 resembles those of lambdoid phages and contains genes characteristic for regulatory components such as N, Q, CI, CII, and CIII and contains a lysis cassette consisting of S, R, and Rz (Fig. 1). Interestingly, the genome of BP-4795 is more related to Stx2-converting phages than to Stx1-converting phages present in the NCBI databases (Fig. 1). A number of genes, especially in the recombination and replication region and the area flanking stx₁, are highly similar to genes of Stx2 phage BP-933W (26) (Fig. 1). Other genes are related to genes of Stx1 phage H19B, genes of defective Stx1 prophages CP-933V of EDL933 and CP-1639 of E. coli O111:H⁻ strain 1639/77, and genes of cryptic phages CP-933K, -M, -N, -P, -R, and -U of E. coli strain EDL933 (Fig. 1).

The first gene in the prophage sequence is int, putatively encoding the phage integrase. It demonstrates 98% sequence identity to intV of Stx1-converting phage CP-933V of E. coli O157:H7 strain EDL933 (25). This similarity of Int to the integrase of CP-933V prompted us to assume that the integration site of BP-4795 is located also in the yehV gene. By computer analysis, an attP core region with the sequence 5′-CCTGTCACGTTACGCGCGTG-3′ was suggested. This was confirmed by PCR and sequence analysis of the junction between prophage and chromosome. Thereby, the integration of BP-4795 into yehV could be confirmed (data not shown).

BP-4795 harbors two intact IS629 elements (Fig. 1). ORF79 and ORF80 are located between these elements. The ORF79 product is in part identical to an unknown protein encoded by a putative ISEc8 element. The ORF80 product resembles to some extent a hypothetical protein encoded in a LEE-flanking region of E. coli O103:H2 strain RW1374 (15). The second IS629 element is followed by ORF83, putatively encoding a protein with 86% identity to protein Z6024, encoded by phage CP-933P in E. coli O157:H7 strain EDL933 (25), as well as 77% and 71%, respectively, identity to NleA/EspI of C. rodentium strain DBS100 (11, 22) and the EspI-like protein of E. coli O103:H2 strain E22 (18). These proteins are type III effectors, which are encoded outside the LEE. The ORF84 product is partially similar to the avirulence A protein of Pseudomonas syringae. The last gene of the prophage genome ORF85 is related to dinI of Shigella flexneri (see AJ556162).

These data let us assume that ORF83 might encode a T3SS effector. Therefore, we examined this ORF further, and due to sequence similarities and in order to follow the proposed nomenclature, we defined the putative protein as an NleA variant and designated it NleA₄₇₉₅ for non-LEE-encoded effector A of BP-4795.

NleA₄₇₉₅ is secreted by the LEE-T3SS.

A 1,744-bp fragment containing ORF83 and the upstream sequence including its promoter was amplified by PCR using primers ORF83-for2 (5′-CCC GAA TTC CAG AAG GGC ATA AAG CTG CCA AGC-3′) and ORF83-HA-rev (5′-CCC TCT AGA TTA AGC GTA ATC TGG AAC ATC GTA TGG GTA GAC TCT TGT TTC TCG GAT TAT ATC AAC-3′). Primer ORF83-HA-rev was used to introduce nine codons specifying an epitope of the human influenza virus hemagglutinin (HA). Moreover, restriction sites for EcoRI and XbaI were inserted with these primers. The PCR product was cloned into the high-copy-number vector pBluescript II KS(+) (Stratagene) and the low-copy-number vector pWSK29 (33), giving pLB15-1 and pLW1-1, respectively. The resulting recombinant plasmids were transformed into wild-type E. coli strain 4795/97 as well as into the escN deletion mutant 4795ΔescN, the latter of which was generated as described previously (6). For this technique, plasmids pKD46, pKD4, and pCP20 (kindly provided by B. L. Wanner [West Lafayette, Ind.]) were employed. A kanamycin resistance gene, amplified from plasmid pKD4 by PCR with primer pair d-escN-for (5′-ATA GGC TTT CAA TCG TTT TTT CGT AAC TAC TGA TAT CTT TGG CAT GGT CCA TAT GAA TAT CCT CC-3′) and d-escN-rev (5′-TGT GCC GGC ATT ATC ATT AAA TTC GTC TAC CAG ATA TGA AGG GCG ATT GTG TAG GCT GGA GC-3′) was transformed into E. coli 4795/97 containing temperature-sensitive plasmid pKD46. For preparation of proteins for immunoblot analysis, strains were grown overnight in Luria-Bertani medium containing 100 μg/ml ampicillin. Cultures were then diluted 1:100 in M9 minimal medium supplemented with 44 mM NaHCO₃ and 0.1% Casamino Acids (28) and grown without shaking at 37°C in 5% CO₂ to an optical density at 600 nm of 0.6 to 0.8. One to three milliliters of the culture was harvested by centrifugation at 16,000 × g for 3 min. The pellet was resuspended in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer (50 mM Tris-HCl [pH 6.8], 4% SDS, 2% β-mercaptoethanol, 12.5% glycerol, and 0.001% bromophenol blue), boiled for 10 min, and centrifuged at 16,000 × g for 5 min.

Secreted proteins were harvested by centrifugation at 5,400 × g for 40 min at 4°C. Supernatants were filtered through 0.45-μm-pore-size filters and precipitated with 0.1 volume of 100% ice-cold trichloroacetic acid on ice overnight. Proteins were collected by centrifugation at 20,000 × g for 40 min at 4°C. The pellets were washed with 100% ice-cold acetone. The air-dried pellets were dissolved in SDS-PAGE sample buffer, boiled for 10 min, and centrifuged at 16,000 × g for 5 min. Samples were analyzed by Western blotting according to standard procedures (28) and probed with the primary anti-HA antibody from a hybridoma supernatant (clone 12CA5; Roche), kindly provided by D. Lindemann (Dresden, Germany), and a second peroxidase-conjugated goat anti-mouse antibody (Pierce). Antigen-antibody complexes were visualized by adding chlornaphthol reagent (6 mg chlornaphthol was solved in 2 ml ethanol and then mixed with 10 ml phosphate-buffered saline containing 5 μl 30% [vol/vol] H₂O₂).

To examine whether NleA₄₇₉₅ is secreted by the LEE-T3SS, bacteria were grown under T3SS-inducing conditions as described above. A protein with the expected size of approximately 50 kDa was detectable in the cell lysates of all strains expressing HA epitope-tagged NleA₄₇₉₅ (Fig. 2, lanes 2, 3, 5, and 6). However, the amount of NleA₄₇₉₅ obtained in the cell lysates of the 4795ΔescN deletion mutant was small (lanes 5 and 6). Moreover, the protein was visible only in the supernatants of strains 4795/pLW1-1 and 4795/pLB15-1 (Fig. 2, lanes 2 and 3) but not in the deletion mutants carrying the same plasmids or the negative controls (Fig. 2, lanes 1, 4, 5, and 6). This indicates that NleA₄₇₉₅ needs a functional T3SS for its secretion in the culture supernatant. Furthermore, it was apparent that NleA₄₇₉₅ is much more strongly induced in the wild type than in the deletion mutant 4795ΔescN.

FIG. 2.

Western blot analysis of HA epitope-tagged NleA₄₇₉₅ in bacterial cell lysates and concentrated supernatants. Control strains, 4795/pBluescript II KS(+) (lane 1) and 4795ΔescN/pWSK29 (lane 4). Lane 2, strains 4795/pLW1-1; lane 3, 4795/pLB15-1; lane 5, 4795ΔescN/pLW1-1; and lane 6, 4795ΔescN/pLB15-1.

NleA₄₇₉₅ localizes with the Golgi apparatus in HeLa cells.

E. coli strains 4795/97 and 4795ΔescN, both carrying vectors pLW1-1 and pLB15-1, as well as E. coli 4795/97 harboring pBluescript II KS(+) as control were used for infection of HeLa cells (Fig. 3). Bacterial strains were grown overnight in Luria-Bertani medium containing 50 μg/ml carbenicillin in glass test tubes in a roller drum. Subsequently, bacterial cultures were diluted with phosphate-buffered saline and directly added to HeLa cells for infection. The human epithelial cell line HeLa (ATCC CCL-2) was grown in Dulbecco's modified Eagle medium (PAA Laboratories) containing 10% fetal calf serum and glutamine in a humidified atmosphere of 5% CO₂. For infection studies, cells were used between passage number 13 and 20 and the bacterial cell counts were adjusted to give a final multiplicity of infection of 10. At 6 h after infection, immunofluorescence analyses were performed basically as described previously (17). Cells were stained with DAPI (4′,6′-diamidino-2-phenylindole) (blue) to visualize bacteria and nuclei of the HeLa cells. Immunostaining was performed with a HA antibody (Roche) and Cy3-conjugated secondary antibody to label NleA₄₇₉₅-HA (red). Furthermore, the trans-Golgi network was labeled using a monoclonal antibody against human Golgin-97 (Molecular Probes) and a Cy2-conjugated secondary antibody (green).

FIG. 3.

Infection of HeLa cells with recombinant E. coli strains 4795/97 and 4795ΔescN, both carrying vectors pLW1-1 and pLB15-1. The wild-type strain 4795/97, transformed with pBluescript II KS(+) (pBS) was used as a control. See the text for details. Scale bars represent 10 μm.

We detected a signal inside the HeLa cells infected with wild-type strains expressing the HA-epitope-tagged NleA₄₇₉₅ from both the low- and the high-copy-number vectors (Fig. 3). In contrast, the samples with the 4795ΔescN deletion mutant and the control strain displayed no signals (Fig. 3). So it can be concluded that NleA₄₇₉₅ is translocated into host cells via the LEE-T3SS. The merged staining patterns (Fig. 3) indicate that NleA₄₇₉₅ localizes with the Golgi apparatus. Gruenheid et al. (11) observed a similar subcellular localization for NleA, which shares sequence similarity to NleA₄₇₉₅.

Concluding remarks.

Stx1-converting phage BP-4795 belongs to the group of lambdoid phages, which exhibits a modular genetic structure and a large extent of mosaicism (14, 27). The presence of two IS629 elements and of genes associated with type III secretion at one end of the prophage let us suggest that these so-called morons were acquired by horizontal gene transfer. Morons are additional genes with an autonomous promoter and a factor-independent transcriptional terminator, which can be transcribed autonomously, even from a repressed phage (13). Typically, their nucleotide composition is different from that of adjacent genes. An example for such a moron is sopE of the Salmonella phage SopEφ (20). In fact, the T3SS-associated morons of BP-4795 exhibit a G+C content of approximately 40% in comparison to about 50% for the rest of BP-4795 (data not shown). The origin of most of the morons is unknown (13), as is the case for the T3SS-associated genes of BP-4795. We assume that these morons were acquired as a result of a random nonhomologous recombination event (13) or due to a transduction when the phage picks up bacterial genes as a cause of incorrect excision from the chromosome.

The ORF84 protein exhibits an amino acid identity of 37% to AvrA, an avirulence protein of P. syringae, but only over half of the length. Avirulence denotes the inability of a pathogen to cause disease on a resistant host plant (2). AvrA belongs to a group of secreted effector proteins that might serve related functions in the cross talk between bacterial pathogens and their plant and animal hosts. Other effectors of this group are AvrA of Salmonella enterica serovar Typhimurium and its relative YopJ of pathogenic Yersinia, which inhibit the NF-κB pathway at separate points and increases apoptosis in human epithelial cells (4, 12).

NleA₄₇₉₅ is the first type III effector encoded by a gene colocated with stx₁ on a single prophage genome. With a length of 459 amino acid (aa) residues, NleA₄₇₉₅ is the largest member within this group of related proteins consisting of Z6024 (441 aa), NleA/EspI (430 aa), and an EspI-like protein (412 aa). The amino acid sequence of these effectors is conserved over large regions (data not shown). The functionality of NleA/EspI has been shown in animal experiments (11, 22), indicating that the amino acids missing are not necessary for their function. Mundy et al. (21) could not detect espI in intimin-negative EHEC strains. However, they could not establish a correlation between the presence of espI and a specific intimin type. NleA₄₇₉₅ can be considered a further member of the NleA family, the intracellular function of which has yet to be established.

Nucleotide sequence accession number.

The sequence of BP-4795 has been deposited in GenBank as a 57,930-bp linear prophage sequence under the accession number AJ556162.