Identification of the Staphylococcus aureus etd Pathogenicity Island Which Encodes a Novel Exfoliative Toxin, ETD, and EDIN-B

Bacterial strains and culture conditions. S. aureus TY114 and TY129 were isolated from pus discharged in a cutaneous wound from a patient in the Clinical Laboratory Department in Hiroshima City Hospital. TY67 was a laboratory stock of a clinical isolate from an impetigo patient. Strains positive for edin-B were from a stock of clinical strains in M. Aepfelbacher’s laboratory in Munich, Germany (16). S. aureus RN4220 and RN450 (edin-B-, eta-, and etb-negative strains) were obtained from Richard Novick and Alexander Tomasz, respectively. Coagulase typing was performed by use of coagulase type-specific antisera (Denka-Seiken, Tokyo, Japan). DNA from S. aureus TY114 was cloned into E. coli XL-II Blue. S. aureus was grown in brain heart infusion broth (Difco Laboratories, Detroit, Mich.) or Trypticase soy broth (TSB) (Becton Dickinson). E. coli was grown in Luria-Bertani broth. When necessary, ampicillin (100 μg/ml) or kanamycin (10 μg/ml) was added for selection or plasmid maintenance.

Materials and chemicals.TSKgel HA1000 and TSKgel G3000 SW_XL were purchased from Tosoh, Tokyo, Japan. Lysostaphin was from SIGMA Genosys Japan, Tokyo, Japan. Recombinant ETA with a His tag on the carboxyl-terminal end was expressed in E. coli DH10B and recovered from the soluble fraction in lysis buffer (20 mM Tris-Cl [pH 8.0], 0.2 M NaCl, 0.2% Triton X-100, and protease inhibitor cocktail [Complete Mini, EDTA free; Roche Diagnostics, Mannheim, Germany]). Extracellular domains of mouse or human Dsg1 and Dsg3 with E tags and His tags on the carboxyl termini were expressed in a baculovirus and collected from culture supernatants as previously described (24). These recombinant ETA and Dsg proteins were purified by using TALON affinity resin (Clontech, Palo Alto, Calif.). Primers used in this study are listed in Table 1. Those for sea through sed, edin-B, and tst-1 were designed according to the nucleotide sequences deposited in GenBank: sea, accession number M18970 ; seb, M11118 ; sec, X05815 ; sed, M94872 ; edin-B, AJ277173 ; tst-1, U93688 . Long range PCR was performed with a TaKaRa LA PCR kit (TAKARA Bio Inc., Osaka, Japan) according to the following amplification protocol: an initial denaturation step at 96°C for 1 min, 40 s, followed by 30 cycles of 96°C for 20 s of denaturation, 65°C for 16 min, and 72°C for 20 min.

DNA manipulation.Routine DNA manipulations were performed by standard procedures (39). Preparation of chromosomal DNA from staphylococcal cells and transformation of S. aureus by electroporation were performed as previously described (47). Southern blotting of DNA and hybridization were performed as described previously (46). The nucleotide sequence was determined by using a PRISM Dye Terminator cycle sequencing kit and an automated sequencer (ABI PRISM 310) (both from Perkin-Elmer Japan, Inc., Tokyo, Japan). Pulsed-field gel electrophoresis (PFGE) was performed according to standard protocols by using a CHEF DR-II apparatus (Bio-Rad Japan, Tokyo, Japan) (initial switch time, 5 s; final switch time, 40 s; run time, 22 h; voltage gradient, 200 V in Tris-borate-EDTA at 16°C).

Purification.Recombinant ETD (the orf1 product) expressed in S. aureus was purified from the culture supernatant as follows. S. aureus growing exponentially in TSB was inoculated into 1 liter of the same fresh medium and incubated with continuous agitation by a rotary shaker for 24 h at 37°C. The culture was centrifuged at 10,000 × g for 20 min at 4°C. Concentrated culture filtrate (CCF) was prepared by 80% saturated ammonium sulfate precipitation of the culture supernatant. CCF dialyzed against 10 mM phosphate buffer (pH 6.8) (buffer 1) was applied to a hydroxyapatite column (15 by 95 mm) which was equilibrated with buffer 1. The column was washed with buffer 1 until most of the unbound proteins had passed through. Bound proteins were eluted by stepwise elution with 100, 250, and 500 mM phosphate buffer (pH 6.8). The eluate with 100 mM phosphate buffer (pH 6.8) was dialyzed against buffer 1 and concentrated to 500 μl. The sample was loaded onto a TSKgel SW3000_XL (7.5 by 300 mm; Tosoh) and eluted with buffer 1 at a flow rate of 0.5 ml/min, and the active fraction was collected. The active fraction was further applied to a Bio-Scale CHT2-1 hydroxyapatite column (7 by 52 mm; Bio-Rad) and eluted with a linear gradient from 10 to 500 mM phosphate buffer (pH 6.8). For purification of His₆-tagged ET, primer sets were designed to amplify DNA fragments corresponding to the mature form of ETB or ETD (orf1). Amplified DNA fragments were cloned into pQE70 (Qiagen, Valencia, Calif.) in order to express the fusion protein with a His₆ tag sequence at the C terminus in E. coli. Recombinant proteins were purified by using nickel-nitrilotriacetic acid resin according to the manufacturer's protocol.

Antisera.The purified protein was emulsified with either Freund complete adjuvant or Freund incomplete adjuvant (Difco Laboratories) (50 μg of protein per ml). For each sample, 2-kg rabbits were immunized on day 1 with the sample emulsified with Freund complete adjuvant and every 2 weeks with the sample emulsified with Freund incomplete adjuvant. At 8 weeks, the rabbits were injected intravenously with 20 μg of the protein sample. Antisera were obtained 15 weeks after the first injection.

In vivo assay.To evaluate the exfoliative activity of ETD, neonatal ICR mice (age, <24 h) or 1-day-old chickens were subcutaneously injected with the toxin dissolved in 100 μl of phosphate-buffered saline (PBS), and the skin was analyzed by eye and microscopically at 1 to 18 h after injection.

Immunofluorescence.Cryosections of nonfixed skin from neonatal mice injected with ETD or PBS alone were used for indirect immunofluorescence to localize Dsg1 and Dsg3, as previously described (3). Anti-Dsg1 sera obtained from patients with pemphigus foliaceus and an anti-Dsg3 mouse monoclonal antibody, AK9, which specifically recognizes the extracellular domain of mouse Dsg3 (K. Tsunoda, T. Ota, M. Aoki, T. Yamada, T. Nagai, T. Nakagawa, S. Koyasu, T. Nishikawa, and M. Amagai, unpublished data) were used. Sections were examined under an Eclipse E800 fluorescent microscope (Nikon Corp., Tokyo, Japan).

In vitro assay.Approximately 1 μg of purified mouse or human Dsg1 or Dsg3 was incubated overnight at 37°C with the indicated amount of ETD in 50 μl of PBS with 1 mM CaCl₂. Digested samples were assessed by immunoblot analysis with an anti-E tag mouse monoclonal antibody (Pharmacia Biotech, Uppsala, Sweden) for detection of the recombinant proteins.

Other procedures.Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting (immunoblotting) were carried out as described previously (49). For detection of ETD and EDIN-B in culture supernatants of clinical strains, S. aureus was grown in TSB (20 ml) with continuous agitation by a rotary shaker for 48 h at 37°C. CCF was prepared as described above and dissolved in 200 μl of PBS. An aliquot (5 μl) of CCF dialyzed against PBS was subjected to SDS-PAGE followed by immunoblotting. Protein was immunodetected by using Renaissance 4CN Plus (Dupont, NEN Research Products, Boston, Mass.). Protein concentrations were determined with the bicinchoninic acid protein assay reagent (Pierce, Rockford, Ill.) with bovine albumin as the standard. Production of enterotoxin and TSST-1 was assayed by using a SET-RPLA enterotoxin detection kit and a TST-RPLA TSST-1 detection kit (both from Denka-Seiken).

Nucleotide sequence accession number.The nucleotide sequence data presented in this report will appear in the DDBJ, EMBL, and GenBank nucleotide sequence databases under accession number AB057421 .

Identification of a new cluster of virulence-related genes.In a systematic screening study for the ET gene and the edin gene in S. aureus clinical isolates from Hiroshima City Hospital, we found two strains showing PFGE bands faintly hybridizing with eta and etb probes. Interestingly, the same PFGE band could hybridize with the edin-B probe. Therefore, we prepared chromosomal DNAs from these strains and tested them by PCR amplification using mixed primers designed on the basis of the homology observed in the ETA, ETB, and ShETB (Staphylococcus hyicus ETB) genes. A single product of approximately 430 bp was amplified. Since this was close to the expected size of 434 bp, the PCR product was subjected to direct nucleotide sequencing. The nucleotide sequence obtained indicated that the relevant portion of the ORF similar to the ET gene was amplified. Since etb and edin-C are located close together on the ETB plasmid (59), we hypothesized that the ORF and edin-B were closely associated. Therefore, we designed specific primer sets for the ORF and edin-B, and we tested our hypothesis by PCR amplification using different combinations of the primers. Among these, a combination of ET-14 and ednB-2 generated a PCR product of ca. 2 kb. Sequencing of the DNA fragment revealed three ORFs (orf1, orf2, and orf3). The orf1 product exhibited homology to the C-terminal regions of various ETs, and the orf3 product exactly matched the N-terminal 224 amino acids of EDIN-B, suggesting that a potential ET gene and edin-B coexist at the same locus. An inverse PCR strategy was used to identify the flanking region of the 2-kb DNA fragment. Use of primers ET-15 and ednB-1 on EcoRI digests of chromosomal DNA allowed amplification of 1.7 kb of DNA. Sequencing and assembly of the DNA fragments resulted in a contiguous sequence of 2,943 bp which covers the full length of orf1, orf2, and orf3 (edin-B). Comparison of the initial sequence data to genome sequences of S. aureus Mu50 and N315 indicated that the DNA flanking orf1 and edin-B was not present in Mu50 and N315. The identification of possible virulence-related ORFs in such a small DNA fragment led us to search for flanking regions of the 3.3-kb DNA for relevant ORFs. Use of primers ET28 and ET29 on XbaI digests and of primers ET15 and ET40 on HindIII digests amplified 6.1- and 6.2-kb DNA fragments, respectively. Sequencing analysis of these DNA fragments and the concatenation of sequences obtained resulted in a contiguous sequence of 14,849 kb containing 9,054 kb of DNA unique to strain TY114 (Fig. 1). The G+C content of the 9,054 kb of unique DNA was 30.67%.

• Open in new tab
• Download powerpoint

FIG. 1.

Schematic representation of the etd pathogenicity island. The 9,054 bp of DNA unique to the strain TY114 chromosome, including orf1, edin-B, and identified ORFs (shaded arrows), is presented. The genetic organization of the putative insertion region of the island on strain N315 chromosomal DNA is shown at the top. Solid arrows, ORFs found in N315.

In the unique sequence obtained, we identified seven protein-coding regions, including orf1 and edin-B (tentatively named orf1 to orf7) (Table 2). The orf1 product was found to display sequence similarity to ETA and ETB; it showed 40% sequence identity to ETA and 59% sequence identity to ETB (Table 3; Fig. 2). Recently, Sato et al. (40) identified a novel ET from S. aureus Horse-1, isolated from a horse with phlegmon, and cloned the gene. They designated it ETC. The orf1 product showed only 13% sequence identity to ETC. On the other hand, S. hyicus has been shown to produce several ETs (5, 41). Of these, genes for two serologically distinct toxins, ShETA and ShETB, have been cloned (54). The orf1 product showed 16% sequence identity to ShETA and 63% sequence identity to ShETB. Phylogenetic comparison of the orf1 product with known ETs revealed that the orf1 product is most similar to ShETB in its primary structure (Fig. 3). The orf2 product showed a predicted amino acid similarity with glutamyl endopeptidase of Bacillus intermedius (37). The orf7 product exhibited high sequence similarities to phage resistance proteins of Lactococcus lactis, such as AbiK (20), and it probably functions as an abortive infection protein against certain phages of S. aureus. In the 5′ extremity, a single copy of IS256, which is 1,173 bp long and has terminal inverted repeats of 26 bp (positions 988 to 1013 and 2286 to 2311), is present. The presence of direct repeats of 8 bp in both boundaries of IS256 (positions 980 to 987 and 2312 to 2319) suggests that it was inserted as a single insertion element into the chromosomal DNA. ORFs just downstream of IS256 showed predicted amino acid similarities to the hsdS and hsdM products, the sequence specificity and modification functions of the type-IC restriction-modification system (26).

• Open in new tab
• Download powerpoint

FIG. 2.

Alignment of ETA, ETB, ETC, ORF1, ShETA, and ShETB. Conserved and identical amino acids are shaded. The multiple alignment was constructed by MacVector software (Genetics Computer Group, Oxford, England) using Clustal W. Amino acids corresponding to the catalytic triad of the S2 family of serine proteases (36) are boxed. Accession numbers are P09331 (ETA), AAA26628 (ETB), BAA99412 (ETC), BAB08178 (ShETA), and BAA99411 (ShETB).

• Open in new tab
• Download powerpoint

FIG. 3.

Phylogenetic tree showing relationships among ETA, ETB, ETC, ORF1, ShETA, and ShETB. The multiple alignment was constructed by MacVector software (Genetics Computer Group) using Clustal W.

Junctions of unique DNA sequence.The junctions of the unique DNA sequences were defined by BLAST searches and PCR. BLAST searches determined that the unique DNA fragment of 9,054 bp (between positions 851 and 9904) was inserted into the chromosome between the ORFs SA2004 and SA2005 on the left and right sides, respectively, and a stretch of 98 bp between positions 2276975 and 2277072 on the N315 chromosome is missing in the unique DNA region (Fig. 4). At both ends of the unique DNA sequence, direct repeats of the pentamer AATTC were present, suggesting that integration of the DNA occurred through a recombination event. We next performed PCR using primers TY85 (on the left of the left junction) and TY86 (on the right of the right junction) based on the known sequences of N315 and Mu50. With a sample of chromosomal DNA from strain TY114 as a template, a PCR product with the expected fragment size of 9.4 kbp was generated by using the LA PCR kit (TAKARA Bio Inc.) (data not shown). In a control PCR using chromosomal DNA from RN450 as a template, a nucleotide product with the expected size of 475 bp appeared (data not shown). These results further support the finding that the unique DNA fragment is present in the chromosome at the specific site between ORFs SA2004 and SA2005 on the left and right sides, respectively, in TY114.

• Open in new tab
• Download powerpoint

FIG. 4.

Left and right junction sequences. The etd pathogenicity island starts at the first arrow (position 2276975 in the N315 chromosome and position 851 in the deduced sequence) and ends at the second arrow (position 2277072 in the N315 chromosome and position 9904 in the deduced sequence). Direct repeats of the pentamer AATTC are boxed.

The orf1 product is a new member of the ETs.To confirm that the orf1 product actually possesses toxin activity, orf1 was cloned into the S. aureus-E. coli shuttle vector pCL8 to generate pTY179, and the gene product was purified from the culture supernatant of TY2182, a derivative of S. aureus RN4220 which was transformed with pTY179. The molecular mass of the purified protein is 27.2 kDa, and the N-terminal sequence, NTYEE, exactly matches the sequence starting from the amino-terminal residue at position 33 of the orf1 product. The sequence starting from the ATG codon with an extension of 32 amino acids is assumed to be a signal sequence, since the protein is extracellularly secreted. We injected 20 μg of the purified protein (mORF1) subcutaneously into neonatal mice. The mice started to show gross blisters around the injection site within 1 h after injection (Fig. 5). Consequently, we constructed a DNA fragment corresponding to the processed form of orf1 by PCR and placed it in frame into a plasmid to express a His₆-tagged fusion protein; then this His₆-tagged protein was purified. Similar results were obtained with injection of the purified His₆-tagged protein into the mice (data not shown). Previous studies demonstrated that ShETs exhibited exfoliative activity in a 1-day-old chicken but not in a neonatal mouse (42). Since the orf1 product was most similar to ShETB in primary structure, we inoculated mORF1 into the back skins of 1-day-old chickens to see whether it showed exfoliative activity toward chickens. However, no exfoliation was observed upon injection of as much as 50 μg of the protein, even after 24 h postinoculation. We prepared a mORF1-specific antibody by using the mORF1 purified from the culture supernatant of TY2182 as described in Materials and Methods, and we compared the serological properties of mORF1 with those of ETA and ETB. As shown in Fig. 6, anti-mORF1 reacted specifically with mORF1 and reacted very weakly with ETB but not at all with ETA. Taken together, these results clearly demonstrated that mORF1 possesses exfoliative activity toward neonatal mice and that it is serologically distinct from ETA and ETB. We therefore designated this protein ETD, as a new member of the ETs produced by S. aureus.

• Open in new tab
• Download powerpoint

FIG. 5.

Exfoliative activity of mORF1 in neonatal mice. Neonatal mice injected with mORF1 show extensive blisters within 1 h after injection (b), while neonatal mice injected with saline alone do not show blisters (a). Histological examination of the mouse injected with mORF1 shows the characteristic splitting at the granular layer (d), while that of the mouse injected with saline shows intact skin (c).

• Open in new tab
• Download powerpoint

FIG. 6.

Immunodetection of ET by the anti-mORF1 antibody. Purified recombinant ETA, ETB, and mORF1 were resolved on an SDS-12% PAGE gel and either stained with Coomassie brilliant blue (a) or transferred to a nitrocellulose membrane (b through d). The membrane was subjected to immunodetection by an anti-ETA antibody (b), anti-ETB antibody (c), or anti-mORF1 antibody (d). Lanes 1, ETA (0.7 μg); lanes 2, ETB (0.7 μg); lanes 3, mORF1 (0.6 μg).

ETD selectively digests Dsg1 but not Dsg3.ETA has been shown to specifically target the mouse and human desmosomal cadherin Dsg1, but not Dsg3, and to cleave Dsg1 in vitro and vivo (3). To determine whether ETD affects Dsg1 as does ETA, we examined skin from neonatal mice injected with purified ETD by immunofluorescence with antibodies to Dsg1 and Dsg3. When skin was examined 1 h after injection of 2 μg of the protein, the immunofluorescence of Dsg1 was markedly diminished, whereas the immunofluorescence of Dsg3 was not affected (Fig. 7). These results clearly indicated that Dsg1 was selectively affected by ETD in vivo. To demonstrate the direct proteolysis of the extracellular domain of Dsg1 by ETD, we incubated the recombinant protein representing the entire extracellular domain of Dsg1 and Dsg3 with purified ETD in vitro. It cleaved the 81-kDa recombinant mouse Dsg1 down to a 29-kDa peptide in a dose-dependent fashion, but it did not cleave mouse Dsg3 at all (Fig. 8A). It also cleaved recombinant human Dsg1 in the same manner, but not human Dsg3 (Fig. 8B). These findings clearly indicate that ETD selectively recognizes and cleaves the extracellular domain of mouse and human Dsg1 as does ETA.

• Open in new tab
• Download powerpoint

FIG. 7.

Immunofluorescence for desmogleins in the epidermis of mice injected with ETD. Neonatal mice were injected with PBS (A and C) or ETD (B and D) and stained for Dsg1 (A and B) or Dsg3 (C and D). Cell surface staining of Dsg1 in mice injected with ETD is much weaker than that in mice injected with saline alone, whereas the staining of Dsg3 is not affected by injection with ETD. Arrowheads indicate epidermal basement membrane.

• Open in new tab
• Download powerpoint

FIG. 8.

In vitro treatment of baculovirus-expressed recombinant desmogleins by ETD. Approximately 1 μg of the purified extracellular domain of mouse (left) or human (right) Dsg1 or Dsg3 produced by a baculovirus was incubated with various concentrations of ETD. The extracellular domain of Dsg1, but not that of Dsg3, was cleaved by ETD in a dose-dependent fashion. Arrows point to the 29-kDa degraded band of Dsg1. Bars on the left side indicate molecular size standards of 150, 100, 50, and 25 kDa (from top to bottom).

edin-B-positive strains are positive for etd.We analyzed the genomic fingerprints and genetic and phenotypic characteristics of edin-B-positive strains from Germany and Japan, and we screened edin-B-positive strains for the presence of etd. As shown in Fig. 9, we confirmed that their DNA patterns by PFGE analysis were very similar, as reported previously (16). A DNA fragment of ca. 140 kb hybridized with both the edin-B and etd probes, indicating that all edin-B-positive strains were positive for etd. By performing immunoblotting using antibodies against ETD, we found that the 12 edin-B- and etd-positive strains produced ETD (Fig. 9). Of 18 strains, 16 strains belonged to coagulase type 2 and 2 strains were nontypeable. Southern blot analysis demonstrated that six strains were sec positive and two strains were seb positive (Fig. 9). This is in contrast with the previous observation that strains positive for eta or etb were negative for enterotoxin genes (sea through sed) (60).

• Open in new tab
• Download powerpoint

FIG. 9.

PFGE and genotypic and phenotypic characterizations of edin-B-positive S. aureus strains. SmaI-digested DNA of edin-B-positive strains was subjected to PFGE and was run on agarose gels. Gels were subjected to Southern hybridization with probes for detecting genes of interest. DNA fragments that hybridized with the probe for detecting the gene of interest are boxed in green (etd and edin-B), yellow (seb), or red (sec). The hybridization signals for edin-B and etd corresponded to the same ca. 140-kb DNA fragment, and that for an enterotoxin gene (seb or sec) corresponded to a 320- to 350-kb DNA fragment. At the bottom, results of coagulase serotyping and data on production of ETD, EDIN-B, enterotoxin, and TSST-1 are summarized. Production of ETD and EDIN-B was assessed by immunoblotting as described in Materials and Methods. Production and serotyping of enterotoxins and TSST-1 were assayed by using the SET-RPLA and TST-RPLA detection kits, respectively. NT, nontypeable. Lanes: 1, TY114; 2, TY129; 3, TY67; 4, TY130; 5, TY134; 6, TY136; 7, TY137; 8, TY140; 9, TY132; 10, TY133; 11, TY135; 12, TY138; 13, TY139; 14, TY141; 15, TY142; 16, TY143; 17, TY144; 18, TY131. Lanes 1 to 3 show results for Japanese strains; lanes 4 to 18 show results for German strains.