Yersinia pseudotuberculosis Produces a Cytotoxic Necrotizing Factor

Abstract

Cell extracts from Yersinia pseudotuberculosis induced multinucleation in HEp-2 cells in a manner similar to the effect caused by Escherichia coli cytotoxic necrotizing factor (CNF). The activity was not dependent on the Yersinia 70-kb virulence plasmid, and the activity was not inhibited by antibodies capable of neutralizing E. coli CNF type 1. The nucleotide sequence of the Yersinia cnf gene was 65.1% identical to the E. coli cnf gene.

Cytotoxic necrotizing factor (CNF) is a protein toxin (110 to 115 kDa) that is prevalent among Escherichia coli isolates from humans and domesticated animals with extraintestinal infections (8; reviewed in references 16 and 27). Caprioli and coworkers (11) identified the prototypical E. coli CNF type 1 (CNF1), and CNF1 is the best-characterized member of an emergent class of dermonecrotic bacterial toxins that constitutively activate eukaryotic small GTP-binding proteins (GTPases) (21, 31, 42, 46). However, there are independent descriptions of a dermonecrotic protein exotoxin in sterile broth culture supernatants from Yersinia pseudotuberculosis (9, 24, 29, 41). This report describes the identification of a CNF from Y. pseudotuberculosis. For clarity, the CNF and the gene encoding the toxin from Yersinia pseudotuberculosis will be referred to, respectively, as CNF_Y and cnf_Y. The gene and gene product from E. coli will be referred to, respectively, as cnf_E and CNF_E.

The following Yersinia species were examined: Y. pseudotuberculosis (33 strains from serogroups IA, IB, II, III, and V), Y. enterocolitica (9 strains from serogroups O3; O4,32; O5,27; O6; O8; O9; O20; O21; and O27), Y. bercovieri (3 strains), Y. frederiksenii (4 strains), Y. intermedia (3 strains), Y. kristensenii (3 strains), Y. mollaretii (2 strains), and Y. rohdei (1 strain). E. coli strain J96 (CNF1⁺) was the source of E. coli CNF_E as previously described (39). E. coli strain DH5α (Life Technologies, Inc., Gaithersburg, Md.) was the host for cloned cnf genes. pHLK102 is a clone of the E. coli cnf_E gene (39), and pHLK602 expresses the cnf_Y gene cloned from Y. pseudotuberculosis strain YPIII as part of the present work. Bacteria were grown for 18 to 24 h at 26 or 37°C in L broth (30) or tryptic soy broth (Difco Laboratories, Inc., Detroit, Mich.), and on L agar (30). Broth cultures were aerated on a shaker (200 rpm) or on a roller drum. Strains containing recombinant plasmids were cultured in media containing ampicillin (100 μg/ml) or chloramphenicol (30 μg/ml).

CNF biological activity and specific activity were assayed with HEp-2 cells as previously described (34, 39). Tryptic soy broth cultures were centrifuged at 4°C to separate the bacteria from the culture broth supernatants, and broth culture supernatants were filter sterilized through 0.2-μm-pore-size low protein-binding membranes (Millipore Corp., Bedford, Mass.). Bacterial cells were concentrated 10-fold in phosphate-buffered saline containing gentamicin (100 μg/ml) and lysed by sonication or by three cycles of freezing and thawing (35). Bacterial lysates were clarified by centrifugation at 4°C and filter sterilized. Cell-free lysates and sterile supernatants were diluted in 0.1 ml of Eagle minimum essential medium with Earle's balanced salt solution (BioWhittaker, Inc., Walkersville, Md.) containing 10% fetal bovine serum (BioWhittaker, Inc.), gentamicin (100 μg/ml), penicillin (100 U/ml), and streptomycin (100 μg/ml) and added to HEp-2 cells cultured in the same medium. The cultures were incubated for 72 h at 37°C in an atmosphere containing 5% CO₂. Culture fluid was then removed, and the cells were fixed and stained with a modified Wright's stain (Hema 3; Fisher Scientific Co.). Stained cells were examined by bright-field microscopy and digitally photographed with a Kodak DC120 camera. The 50% cytopathic dose (CD₅₀) was determined by the dilution of a sample that produced multinucleation in 50% of the cells in a well of the tissue culture plate. Specific activity was defined as the number of CD₅₀s per microgram of protein in a sample. The amount of protein in bacterial cell-free lysates and sterile supernatants was quantitated spectrophotometrically (BCA Protein Assay Kit; Pierce Chemical Co., Rockford, Ill.), and bovine serum albumin was used as a standard. Neutralization of biological activity by specific antibodies was assayed in vitro with HEp-2 cells as previously described (34, 39).

Standard methods were used for DNA purification and cloning (32, 40). Restriction endonucleases (New England Biolabs, Beverly, Mass.) and DNA-modifying enzymes (New England Biolabs and Roche Molecular Biochemicals, Indianapolis, Ind.) were used according to the manufacturers' specifications. Restriction enzyme digestion products and PCR amplification products were transferred from agarose gels to nylon membranes under alkaline conditions (38). Dot blots were prepared by denaturing samples of DNA with 0.5 M NaOH and applying the samples to a nylon membrane filter (Minifold; Schleicher & Schuell, Keene, N.H.). Membranes were hybridized with horseradish peroxidase-labeled probes (described below), washed, and exposed to X-ray film according to the manufacturer's specifications (ECL Direct Labeling and Detection System; Amersham-Pharmacia Biotech, Piscataway, N.J.). Autoradiograms were scanned, and the digital images were annotated with Canvas (Deneba Systems, Inc., Miami, Fla.). The cnf_E probe was a 3.0-kb BamHI-EcoRI gel-purified fragment from pHLK102 (39), and the cnf_Y probe from pHLK602 was a 2.8-kb BsaI-NcoI fragment.

The PCR was done as previously described (39) with commercially synthesized primers (Life Technologies, Inc.) that were designed to amplify cnf sequences from E. coli (19) and Yersinia (37). The following primers were used to amplify E. coli cnf_E: forward, 5′-TATTAATCTTCACAGAGGAG-3′; reverse, 5′-CCGGTTATTTATTAAAGGGCTTAG-3′. The following primers were used to amplify Yersinia cnf_Y: forward, 5′-TGCATCGTCAATAAAAGGAGTGTT-3′; reverse, 5′-CAATTTGGTTTTACTGGTGGTTCA-3′. The sequences of the Yersinia cnf_Y primers correspond to nucleotides 800 to 823 and 2918 to 2941 of the sequence that was determined in this study.

DNA sequencing was performed with dichlororhodamine terminator cycle-sequencing reagents (Perkin-Elmer, Foster City, Calif.) according to the manufacturer's specifications, and automated sequencing was performed on an ABI Prism 377 instrument. Universal primers were used to sequence the ends of cloned DNAs, and custom primers were designed from the Y. pseudotuberculosis strain YPIII sequences that were obtained. These custom primers were used to complete the sequencing of the cloned DNA, and they were also used in direct sequencing of genomic DNA. The data reported here were obtained by sequencing both strands of the template. BLAST (3) alignments were performed with data at the National Center for Biotechnology Information (NCBI) (at the website http://www.ncbi.nlm.nih.gov/blast/) and from the Sanger Center (at the website http://www.sanger.ac.uk/Projects/Y_pestis/) (37). Additional analyses of DNA sequences were performed with the software packages Lasergene, version 4.0.3 (DNAStar, Inc., Madison, Wis.) and the Wisconsin Package, version 10.1-Unix (Genetics Computer Group, Madison, Wis.).

Broth supernatants and cell-free lysates from cultures of different species of Yersinia were tested for cytopathic activity. A total of 58 strains of Yersinia were examined, including 33 strains of Y. pseudotuberculosis. HEp-2 cells treated with extracts from Y. pseudotuberculosis strains YPIII (7, 22) and IP2666c (cured of the 70-kb virulence plasmid [45]) developed into giant, multinucleated cells that remained viable and attached to the culture substrate (Fig. 1). Y. pseudotuberculosis strain YPIIIc, a derivative of strain YPIII that was cured of the virulence plasmid (7, 22), also expressed CNF_Y activity when grown in vitro at 26 or 37°C. CNF_Y activity was not detected in other strains of Y. pseudotuberculosis (Table 1 and data not shown) or from additional Yersinia species (Fig. 1B and C). Additional preliminary experiments indicated that CNF_Y activity was constitutively expressed throughout the growth cycle of Y. pseudotuberculosis and expression was not regulated by temperature or [Ca²⁺] (not shown). The morphology of cells treated with extracts from CNF_Y⁺ Y. pseudotuberculosis was indistinguishable from that caused by E. coli CNF_E (compare Fig. 1D with Fig. 1E and F). The specific activity in cell-free lysates of Y. pseudotuberculosis strain YPIII was 38-fold higher than that of the CNF_E⁺ E. coli strain J96, and a substantial amount of CNF_Y activity (10³ CD₅₀s/μg of protein) was found in the supernatants from broth cultures of CNF_Y⁺ Y. pseudotuberculosis strains (Fig. 1E). CNF_E activity from E. coli strain J96 was cell associated, and no biological activity was detected in filter-sterilized broth-culture supernatants, an observation that was consistent with previous studies of CNF_E (11, 23). Y. pseudotuberculosis strains YPIII and IP2666c belong to serogroup III (7, 22, 45), whose strains have been previously reported to produce a dermonecrotic protein exotoxin (9, 24, 26, 29, 41, 47). In the present work, however, additional serogroup III strains did not express CNF_Y biological activity and these strains contained deletions within the cnf_Y gene (described below). Thus, no relationship was found between serogroup and toxin production.

FIG. 1.

HEp-2 cell tissue culture assay for CNF biological activity. (A) Untreated control; (B) broth culture supernatant from the CNF_Y⁻ Y. enterocolitica strain 8081v; (C) cell-free lysate from CNF_Y⁻ Y. enterocolitica strain 8081v; (D) cell-free lysate from the CNF_E⁺ E. coli strain J96; (E) broth culture supernatant from the CNF_Y⁺ strain Y. pseudotuberculosis strain YPIII; (F) cell-free lysate from the CNF_Y⁺ strain Y. pseudotuberculosis strain YPIII. All photomicrographs (panels A through F) were taken at the same magnification (10× lens objective) to clearly depict the giant, multinucleated phenotype of cells in panels D to F. Cells were digitally photographed by light microscopy with a Kodak DC120 camera, and the image file was annotated with Canvas.

TABLE 1.

Distribution of cnf_Y among Y. pseudotuberculosis strains

Straina	CNFY biological activityb	Approx size (kb) of cnfY allelec	HPId	Serotype	Source (reference)
YPIII	+	3.0	−	III	V. L. Miller (7, 22)
YPIIIc	+	3.0	−	III	V. L. Miller (7, 22)
713425	−	2.1	−	IA	V. L. Miller
722080	−	2.1	−	IB	V. L. Miller
730317	−	2.1	−	I	V. L. Miller
730440-1	−	1.8	+	IB	V. L. Miller
ATCC 29833	−	1.8	NDe	ND	ATCCf
PTB1	−	1.8	+	ND	K. A. McDonough (25)
PTB4	−	1.8	+	IA	K. A. McDonough (25)
PTB6	−	1.8	+	III	K. A. McDonough (25)
PTB7	−	1.8	+	ND	K. A. McDonough (25)
PTB8	−	1.8	+	III	K. A. McDonough (25)
PTB9	−	1.8	+	IA	K. A. McDonough (25)
PTB13	−	2.1	−	I	K. A. McDonough (25)
PTB16	−	1.8	−	III	K. A. McDonough (25)
PTB20	−	2.1	−	IA	K. A. McDonough (25)
PTB22	−	1.8	−	IA	K. A. McDonough (25)
IP2515c	−	2.1	ND	II	J. B. Bliska (45)
IP2666c	+	3.0	ND	III	J. B. Bliska (45)
IP2775c	−	1.8	ND	I	J. B. Bliska (45)
IP2777c	−	1.8	ND	I	J. B. Bliska (45)
IP2790c	−	1.8	ND	I	J. B. Bliska (45)
IP32953	−	1.8	+	I	E. Carniel (10)

^aStrains with names ending in ‘c' are cured of the Yersinia virulence plasmid.

^bSterile cell extracts induced the formation of giant, multinucleated HEp-2 cells.

^cThree cnf_Y alleles were identified in this study: 3.0 is the size of a full-length gene shown in Fig. 2 and 3; 2.1 is the size of a 3′-deleted gene also found in Y. pestis strain CO92 (37) and shown in Fig. 3; 1.8 is the size of a variant with 5′ and 3′ deletions.

^dHPI, Yersinia high-pathogenicity island.

^eND, not determined.

^fATCC, American Type Culture Collection, Manassas, Va.

A DNA probe prepared from the cloned E. coli cnf_E gene hybridized to a single restriction endonuclease fragment in a Southern blot of total DNA from Y. pseudotuberculosis strain YPIII (not shown). However, PCR primers that were designed to amplify cnf_E from E. coli (39) did not amplify sequences from Y. pseudotuberculosis strain YPIII. A large open reading frame (ORF), with a high level of similarity to the amino acid sequence of E. coli CNF_E, was identified by a TBLASTN search of the Y. pestis genome sequence (37), and PCR primers designed from the cnf_Y sequence in Y. pestis were used to amplify a 2.1-kb product from Y. pseudotuberculosis strain YPIII. This PCR product was cloned and used as a probe to identify a 5.4-kb chromosomal fragment from Y. pseudotuberculosis strain YPIII DNA that was digested with BamHI and SalI (not shown). This fragment was cloned into pBRKS (44) to create pHLK602, which expressed CNF_Y biological activity in E. coli K-12 strain DH5α. Additional evidence that the Y. pseudotuberculosis cnf_Y sequence was responsible for the observed biological activity was obtained by inserting an Ω-Cm^r fragment (20) into the chromosomal cnf_Y gene of Y. pseudotuberculosis strain YPIII by allelic exchange (17). The derivative with the cnf_Y:: Ω-Cm^r mutation, named strain WEX5000, was confirmed by PCR and Southern blotting and expressed no CNF_Y activity in the HEp-2 tissue culture assay. The mutation in strain WEX5000 was complemented by pHLK602.

Analysis of the sequence of the Yersinia DNA cloned in pHLK602 identified a large ORF with a predicted translation product that was highly similar to CNF_E. There are two alleles of E. coli cnf (cnf₁ and cnf₂) whose DNA and predicted amino acid sequences are ≥84% identical (19, 36). When the DNA or predicted amino acid sequences of cnf_Y from Y. pseudotuberculosis were compared to either of the two E. coli sequences, there was ≤1% difference in the results (not shown). Thus, the sequence of cnf₁ (19) was used for the comparisons of cnf_Y and cnf_E reported here. The cnf_Y ORF was 3,045 bp in size and had 65.1% sequence identity with cnf_E. The predicted amino acid sequence of CNF_Y was 60.8% identical and 68.5% similar to that of CNF_E (Fig. 2). The homology between CNF_Y and CNF_E was evenly distributed throughout the entire sequence. The two largest areas of dissimilarity were 5 amino acids (aa) in length, there was one region containing four nonconservative amino acid changes, and all of the remaining dissimilarities were 3 aa in length, or smaller. Cysteine and histidine residues that were essential for the biological activity of CNF_E (43) were conserved in the predicted sequence of CNF_Y (Fig. 2). The DNA sequences flanking cnf_Y had significant similarities to transposases (BLASTX E value, 5e-36) and oxidoreductases (BLASTX E value, 7e-40) (Fig. 3). The overall content of guanosine and cytosine (percent G+C) in the DNA that was sequenced in this study was 39% (Fig. 3), but there was a noticeable demarcation. The percent G+C for cnf_Y and the DNA 5′ to the gene was 34%, while 3′ to the cnf_Y gene the sequence contained 51% G+C. Taken together, the sequence data suggested that the cnf_Y gene was introduced into Y. pseudotuberculosis from another bacterium.

FIG. 2.

Alignment of the predicted amino acid sequences of CNF_Y (Y) from strain YPIII and CNF_E (E) by the method of Lipman and Pearson. Conserved amino acids that were shown to be essential for the activity of CNF_E (Cys866 and His881) (43) are bolded and marked by asterisks under those positions, and identical (|) or similar (:, .) residues are indicated.

FIG. 3.

Comparison of cnf_Y DNA sequences from Y. pseudotuberculosis strain YPIII and Y. pestis strain CO92. Boxes indicate sequences that are ≥99% identical in the two species, and dotted lines indicate deletions. Nucleotide positions are numbered according to the sequence present in each species. The BamHI and SalI sites used in the cloning of cnf_Y from strain YPIII also are shown. ▨, putative transposase; □, intervening regions; ▪, CNF_Y; ░⃞, putative oxidoreductase.

Despite the considerable identity between the DNA and predicted amino acid sequences of cnf_Y and cnf_E, the CNF_Y biological activity in extracts from Y. pseudotuberculosis strain YPIII and E. coli K-12 strain DH5α(pHLK602) was not neutralized by goat polyclonal antibodies prepared against purified E. coli CNF_E (34). The antibodies completely prevented the multinucleating activity of E. coli CNF_E when they were used at a final dilution of up to 1:80, but there was no inhibition of the activity from Y. pseudotuberculosis strain YPIII with final antibody dilutions as low as 1:10 (not shown). Preimmune sera had no effect on the activity of either CNF. Identical results were obtained with a mouse monoclonal antibody (33).

The DNA sequence that was determined from Y. pseudotuberculosis strain YPIII in this study was compared to a similar sequence in the genome database of Y. pestis strain CO92 (37) (Fig. 3). There was ≥99% identity between the DNA sequences that were present in both species. However, there were deletions unique to each species. The sequence from Y. pestis contained the first 2,114 bp of the cnf_Y ORF, but there was a 931-bp deletion at the 3′ end of the gene. Distal to the deletion, Y. pestis retained the 3′ flanking sequence with similarity to oxidoreductases. The sequences preceding cnf_Y in both species also contained deletions in the intervening region between the putative transposase and the start of the cnf_Y gene (Fig. 3).

Chromosomal DNAs from pathogenic and nonpathogenic species of Yersinia were examined for the presence of the cnf_Y gene. All of the isolates of Y. pseudotuberculosis (n = 33) that were examined contained cnf_Y DNA that hybridized to the cnf_Y probe (representative results from seven strains of Y. pseudotuberculosis are shown in Fig. 4) or amplified in a PCR with cnf_Y-specific primers (not shown). The cnf_Y gene was detected in Y. pseudotuberculosis strain YPIIIc and seven other Y. pseudotuberculosis strains cured of the 70-kb virulence plasmid. cnf_Y sequences were not detected in Y. enterocolitica or in nonpathogenic Yersinia species (Fig. 4). A Southern blot of Y. pseudotuberculosis DNAs showed polymorphisms at the cnf_Y locus (Fig. 5), and further analysis revealed that Y. pseudotuberculosis strains YPIII, YPIIIc, and IP2666c (all CNF_Yp⁺) possessed the complete cnf_Y gene shown in Fig. 3. All of the remaining strains of Y. pseudotuberculosis carried a single deletion like that in Y. pestis strain CO92 (Fig. 3), or they had the Y. pestis deletion and a second deletion between nucleotides 1745 and 2065 with respect to the cnf_Yp sequence from Y. pseudotuberculosis strain YPIII (not shown). The single cnf_Y deletion in Y. pestis strain CO92 (Fig. 3) was confirmed in additional independent Y. pestis isolates (H. A. Lockman and P. L. Worsham, unpublished observation). The discovery of identical cnf_Y deletions in Y. pseudotuberculosis and Y. pestis is consistent with the evidence that the plague bacillus is a recently evolved clone of Y. pseudotuberculosis (1). However, in an examination of a subset of Y. pseudotuberculosis strains, none of the strains that contained a plague-like cnf_Y allele (n = 6) possessed the Yersinia high-pathogenicity island (HPI) (12) (Table 1), a feature that distinguished them from Y. pestis. In contrast, 8 of 10 Y. pseudotuberculosis strains that were HPI⁺ also possessed the second cnf_Y deletion (described above) (Table 1) that was not found in Y. pestis. It was also evident from this analysis that only those strains of Y. pseudotuberculosis with a full-length cnf_Y gene expressed biological activity, and the strains that did not express CNF_Y activity had significant deletions of the cnf_Y ORF.

FIG. 4.

Dot blot of total DNAs from 33 strains of Yersinia. The blot was hybridized at high stringency with a cnf_Y gene probe. Individual strains of Y. pseudotuberculosis are listed by strain name or number next to the position of the DNA from that strain; +, positive control, the cloned cnf_Y from Y. pseudotuberculosis strain YPIII; Ye, Y. enterocolitica; Yb, Y. bercovieri; Yf, Y. frederiksenii; Yi, Y. intermedia; Yk, Y. kristensenii; Ym, Y. mollaretii; Yr, Y. rohdei. All samples were applied to the blot in duplicate. The autoradiogram was scanned, and the digital image file was annotated with Canvas.

FIG. 5.

cnf_Y probe hybridization to a Southern blot of chromosomal DNAs from Y. pseudotuberculosis strains. Lane 1, YPIII; lane 2, YPIIIc; lane 3, 713425; lane 4, 722080; lane 5, 730317; lane 6, 730440-1; lane 7, ATCC 29833. One microgram of each DNA was digested with NsiI. The numbers in the left margin indicate molecular size standards in kilobases. The autoradiogram was scanned, and the digital image file was annotated with Canvas.

The discovery of a chromosomally encoded CNF_Yp in Y. pseudotuberculosis adds a potential virulence factor to the variety of potent cytotoxins that are produced by these bacteria. Altogether, the results of this study indicated that some Y. pseudotuberculosis strains express a CNF toxin that may, like the E. coli CNF, target eukaryotic small GTPases. Despite the remarkable similarities between the biological activities and the nucleotide and predicted amino acid sequences of the cnf genes from Y. pseudotuberculosis and E. coli, the absence of neutralization of CNF_Y by anti-CNF_E antibodies indicated that there are significant antigenic differences between the proteins. CNF_Y activity was found in the supernatants of broth cultures of Y. pseudotuberculosis, a result suggesting that CNF_Y is secreted, unlike the cell-associated E. coli CNF_E (11, 23). Pathogenic strains of Yersinia also produce plasmid-encoded cytotoxins (Yops) that are substrates of a type III secretion system (2, 13-15). Like CNFs, some Yop cytotoxins profoundly alter the cytoskeletal architecture of mammalian cells by affecting the activity of GTPases. YopE is a GTPase-activating protein that negatively regulates Rho activity (6, 49), and YopT inactivates Rho via a covalent modification (50). YopH is a potent tyrosine phosphatase that acts on p130^Cas and FAK, thereby inhibiting formation of focal adhesion complexes (5). YpkA (YopO) is a serine/threonine kinase that is activated by actin and subsequently inhibits Rho activity (18, 28). YpkA also binds directly to Rho and Rac (4). Understanding the combined roles of all of these toxins in the pathogenic lifestyle of Y. pseudotuberculosis awaits additional studies, but it is relevant to note that YopT is not expressed by the CNF_Y⁺ Y. pseudotuberculosis strain YPIII (48). Thus, different strains of this species may express virulence through alternate mechanisms.

Nucleotide sequence accession number.

The cnf_Y DNA sequence reported here, from Y. pseudotuberculosis strain YPIII, was deposited in GenBank (NCBI) under accession number AF324349.