BopC Is a Novel Type III Effector Secreted by Bordetella bronchiseptica and Has a Critical Role in Type III-dependent Necrotic Cell Death*

In Bordetella bronchiseptica, the functional type III secretion system (TTSS) is required for the induction of necrotic cell death in infected mammalian cells. To identify the factor(s) involved in necrotic cell death, type III-secreted proteins from B. bronchiseptica were analyzed using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry and electrospray ionization tandem mass spectrometry. We identified a 69-kDa secreted protein designated BopC. The gene encoding BopC is located outside of the TTSS locus and is also highly conserved in both Bordetella parapertussis and Bordetella pertussis. The results of a lactate dehydrogenase release assay and terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end-labeling assay demonstrated that BopC is required for necrotic cell death. It has been reported that tyrosine-phosphorylated proteins (PY) of host cells are dephosphorylated during B. bronchiseptica infection in a TTSS-dependent manner. We found that BopC is also involved in PY dephosphorylation in infected host cells. It appears that the necrotic cell death triggered by BopC occurs prior to the PY reduction in host cells, because Bordetella-induced cell death was not affected even in the presence of a dephosphorylation inhibitor. Furthermore, a translocation assay showed that the signal sequence for both secretion into culture supernatant and translocation into the host cell is located in 48 amino acid residues of the BopC N terminus. This report reveals for the first time that a novel type III effector, BopC, is required for the induction of necrotic cell death during Bordetella infection.

isolated from a large number of four-legged animals (4) and causes kennel cough in dogs (5) and atrophic rhinitis in swine (6), although most cases of B. bronchiseptica infection are asymptomatic (7). These three Bordetella species colonize the host's respiratory tract.
In Bordetella, most virulence factors, such as adhesion factors and toxins, are expressed under the control of a two-component regulatory system composed of BvgA and BvgS (BvgAS system) (8). BvgS is a sensory histidine kinase and is localized in the bacterial membrane (9). When the bacteria are situated in conditions resembling the extracellular environment of the host tissue during colonization, BvgS is autophosphorylated and transfers its phosphate group to BvgA (10). The phosphorylated BvgA binds to the upstream of the promoters of specific genes that encode mainly virulence factors, and acts as the transcriptional activator (11)(12)(13). In this way, the activation of the BvgAS system leads the bacteria to the virulent phase. The BvgAS system also positively regulates the expression of a virulence factor secretion system called the type III secretion system (TTSS) 2 (14).
The TTSS is highly conserved in a number of Gram-negative bacteria and functions as the injector of virulence proteins, so-called effectors, to the host cell (reviewed in Ref. 15). For example, functional TTSS is required for the formation of the pedestal-like structure beneath the bacterial adhesion site in enteropathogenic Escherichia coli (EPEC) (16) and for the bacterial invasion of epithelial cells in Shigella (17,18). If these pathogens are deficient in the TTSS function, their virulence is greatly reduced, indicating that TTSS is one of the major mechanisms for exerting their pathogenicity. In general, the bacteria possessing TTSS exploit several effectors to achieve infection. In EPEC, seven effectors, Tir, EspG, Orf3, EspF, EspH, Map, and Cif, have been identified (19,20), and Shigella secretes VirA, IcsB, IpaH 9.8 , and IpgD (21)(22)(23)(24) via the TTSS as effectors. Thus, effector variations provide different pathological features among bacterial infection.
In B. bronchiseptica, the functional TTSS is required for the induction of cytotoxicity, including necrotic cell death, in cultured mammalian cells during infection (25). Moreover, B. bronchiseptica with TTSS defi-* This work was supported by a Grant-in-aid for Young Scientists (Area B, Grant ciency failed at long term colonization of trachea of rats (14) and mice (26), suggesting that TTSS also has important roles in Bordetella pathogenicity. Four type III-secreted proteins, BopB, BopD, BopN, and Bsp22, have been identified in Bordetella (26,27). We have demonstrated that BopB and BopD make a complex and form pores on the host plasma membrane as a conduit of effectors (28). However, no proteins have been characterized as definite type III effectors in Bordetella. Here, we identified and characterized a novel type III effector, BopC, in Bordetella.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Cell Culture, and Media-The wild-type strain used in this study was B. bronchiseptica S798 (27). The type III secretion mutant (type III Ϫ ) and BopB mutant (⌬BopB) were derived from S798 (27). Bordetella strains were cultured in Stainer and Scholte (SS) liquid medium with a starting A 600 of 0.2, and the inoculum was prepared from fresh colonies grown on Bordet and Gengou (BG) agar as described previously (29 -31). For the infection assay, B. bronchiseptica strains cultured for 18 h at 37°C with vigorous shaking were used. E. coli DH10B, MC1061, and SM10pir were used as hosts for the construction of various plasmids. L2 cells (ATCC CCL-149) were maintained in F-12K (Invitrogen) with 10% fetal calf serum, and HeLa cells (ATCC CCL-2) were maintained in minimum essential medium Eagle (Sigma) with 10% fetal calf serum. COS-7 (ATCC CRL-1651) and 293T (32) were maintained in Dulbecco's modified Eagle's medium (Sigma).

MALDI-TOF MS and ESI-MS/MS Analyses-
The protocols for preparing samples for MALDI-TOF MS or ESI-MS/MS analyses were described elsewhere (33). Briefly, after SDS-PAGE, protein bands were stained with Coomassie Brilliant Blue (CBB) and excised from the gels. The resulting gel pieces were treated with a reducing agent such as dithiothreitol. After alkylation by iodoacetamide, the proteins were digested with trypsin (sequencing grade, Roche Molecular Biochemicals), and the resulting samples were analyzed by the Voyager-DE PRO MALDI mass spectrometer (Applied Biosystems) or Q-Tof 2 (Micromass). To retrieve the amino acid sequence of proteins analyzed by mass spectrometry, all speculative open reading frames (ORFs) were picked up from the genomic sequence obtained from the B. bronchiseptica Sequencing Group at the Sanger Institute (ftp.sanger.ac.uk/pub/pathogens/bb/) (34). These ORFs were converted into amino acid sequences, and then the molecular mass data base of trypsinized peptides was constructed.
Construction of Plasmid for Gene Expression in Bordetella and Mammalian Cells-A vector used for the gene expression in Bordetella, pRK415 R4-R3-F, was constructed as follows. A 1.8-kbp fragment containing attR4, ccdB, the chloramphenicol acetyl transferase gene, and attR3 was amplified by PCR with the primers HindIII-R4-R3-F (5Ј-CCCAAGCTTCAGGAAACAGCTATGACCATG-3Ј) and HindIII-R4-R3-R (5Ј-CCCAAGCTTGTTTTCCCAGTCACGACGTT-3Ј) using pDEST R4-R3 (Invitrogen) as the template. The underlined portions indicate the HindIII sites. This fragment was digested with HindIII and then inserted into the HindIII site in pRK415 (37). The resulting plasmid, in which the ccdB is oriented in the opposite direction from the lac promoter in pRK415, was designated pRK415 R4-R3-F. To insert the fha promoter region into pDONR P4-P1R (Invitrogen), a 0.5-kbp fragment was amplified by PCR with the primers B4F-fhaP (5Ј-ATAGAA-AAGTTGTCCTTTTCCATCAGGACCCG-3Ј) and B1R-fhaP (5Ј-TG-TACAAACTTGATTCCGACCAGCGAAGTGAA-3Ј) using B. bronchiseptica S798 genomic DNA as the template. This PCR product was cloned into the vector by means of adaptor PCR and site-specific recombination techniques using the Gateway cloning system. The resulting plasmid was designated pDONR-fhaP. Likewise, to insert the rrnB terminator region into pDONR P2R-P3 (Invitrogen), a 0.4-kbp fragment was amplified by PCR with the primers B2F-rrnB (5Ј-TGTA-CAAAGTGGGGCTGTTTTGGCGGATG-3Ј) and B3R-rrnB (5Ј-A-TAATAAAGTTGAAACAAAAAGAGTTTGTAGAAACG-3Ј) using pTrc99A (Amersham Biosciences) as the template. This PCR product was cloned into the vector, and the resulting plasmid was designated pDONR-rrnB.
Preparation of Proteins from Culture Supernatant and Whole Bacterial Cells-The proteins released into bacterial culture supernatants and whole bacterial cell lysates were prepared by trichloroacetic acid precipitation. The culture supernatants were filtered, and the bacterial pellets were resuspended in distilled water. Trichloroacetic acid was then added to each sample at a final concentration of 10%. After incubation on ice for 15 min, the samples were centrifuged for 5 min. The resulting precipitated proteins were dissolved in the SDS-PAGE sample buffer.
Preparation of Antibodies-To prepare the anti-BopC antibodies, the peptides corresponding to the amino acid residues 62-80 (CKHDVR-QFQASGDRSLQQLR) or 569 -586 (CSQLDHDAKTSGNESQLNR) of BopC (a cysteine residue was added to each peptide at the N terminus for cross-link with the carrier protein) were conjugated separately with hemocyanin from keyhole limpets (Sigma) by using 3-maleimido-benzoic acid N-hydroxysuccinimide ester (Sigma). These cross-linked peptides were used to immunize rabbits, and the resulting antisera were incubated with each peptide immobilized on epoxy-activated Sepharose 6B (Amersham Biosciences) to obtain specific immunoglobulin fractions. The mixture of these purified anti-BopC antibodies was used for immunoblot assays.
Infection Assays-L2 cells seeded on coverslips were infected with bacteria at a multiplicity of infection (m.o.i.) of 200 and then were centrifuged for 5 min and incubated for 20 min at 37°C under an atmosphere of 5% CO 2 . The cells were then washed with phosphate-buffered saline (PBS) and fixed in methanol. Fixed cells were stained with Giemsa solution. In the case of the terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end-labeling (TUNEL) assay, 4% paraformaldehyde was used as the reagent for fixation. The TUNEL assay was performed by using an In Situ Cell Death Detection Kit (Roche Applied Science). To examine the release of lactate dehydrogenase (LDH) from B. bronchiseptica-infected cells, 7.5 ϫ 10 4 HeLa cells seeded on 24-well plates were infected at an m.o.i. of 100. They were then centrifuged for 5 min and incubated at 37°C under an atmosphere of 5% CO 2 . The amounts of LDH were measured spectrophotometrically using a Cyto-Tox 96 Non-Radioactive Cytotoxicity Assay kit (Promega). To calculate the percentage of non-viable cells, 3 ϫ 10 5 HeLa cells seeded on 6-well plates were infected as in the LDH assay described above. After 60 min of infection, cells were washed with cold PBS twice, and then PBS containing 1 mM EDTA was added to the wells. After 15 min of incubation at 4°C, the cells were detached from substrata using a Pasteur pipette, and then the cell suspensions were centrifuged. The cells were resuspended with PBS containing 0.04% trypan blue and the resulting stained cells were counted under a microscope using a hemocytometer. Staurosporine (Sigma) was used for inducing apoptosis at a final concentration of 2.5 mM.
Hemolytic Assay-The measurement of type III-dependent hemolytic activity was carried out as described previously (35). Briefly, bacterial pellets from overnight cultures and rabbit red blood cells were washed with PBS and adjusted to 5 ϫ 10 10 bacteria/ml and 3 ϫ 10 9 cells/ml with PBS, respectively. The suspensions were mixed together (50-l aliquots per suspension) on a 96-well plate and centrifuged for 5 min for close contact; the combined suspension was then incubated at 37°C for 30 min in a CO 2 incubator. The bacteria-red blood cell suspensions were gently resuspended with an additional 100 l of PBS, and then the plate was centrifuged. The supernatants were transferred to a fresh plate, where the optical density at 492 nm was measured.
Detection of Tyrosine-phosphorylated Proteins (PY) in Host Cells-The protocol of immunofluorescence staining has been described elsewhere (40). HeLa cells were infected with bacteria as mentioned above. When sodium orthovanadate (Na 3 VO 4 ) was used to inhibit of phosphatase activity, this reagent was added to the extracellular medium at a final concentration of 1 mM during infection. After the fixation, PY were stained with anti-PY monoclonal antibody 4G10 (Upstate Biotechnology). As a secondary antibody, Alexa Fluor 488 or 594 goat anti-mouse IgG (Molecular Probes) was used. Filamentous actin (F-actin) was stained with rhodamine phalloidin (Molecular Probes). Bacteria were stained with anti-B. bronchiseptica antisera (Denka Seiken). To detect PY in HeLa cells infected with bacteria by immunoblot assay, infected cells were washed with ice-cold PBS three times, and then the cells were treated with lysis buffer (41). The cell lysates were sonicated for 30 s and clarified by centrifugation at 15,000 ϫ g for 15 min. The supernatants were then separated by SDS-PAGE with 10% gel and immunoblotted with anti-PY antibody RC20 (BD Biosciences) or anti-actin monoclonal antibody (Chemicon).
Effector Translocation Assay-HeLa cells were seeded at 5 ϫ 10 4 cells/well in a 6-well plate, and then the plate was incubated at 37°C overnight in a CO 2 incubator. HeLa cells were exposed to B. bronchiseptica for 60 min at an m.o.i. of 1000. The infected cells were washed with Hanks' balanced salt solution (Sigma), and then cells were stained with CCF2-AM solution according to the manufacturer's protocol for the GeneBLAzer In Vivo Detection Kit (Invitrogen). The stained cells were analyzed under a fluorescence microscope using a filter set for 4Ј,6-diamidino-2-phenylindole (Zeiss).
Transfection Assay-HeLa, COS-7, or 293T cells were seeded at 2.5 ϫ 10 4 cells/well in a 24-well plate, and then the plate was incubated at 37°C overnight in a CO 2 incubator. Mock vector or pcDNA-BopC was introduced into the cells using FuGENE6 (Roche Applied Science) according to the manufacturer's protocol. After incubation for 18 h in a CO 2 incubator, the amounts of LDH released into the extracellular media were measured as described above.

BopC Is a Novel Secreted Protein via the Bordetella TTSS-The B.
bronchiseptica wild-type and type III Ϫ strains were cultured in SS medium for 18 h, and the supernatant proteins were recovered by trichloroacetic acid precipitation. These samples were separated by SDS-PAGE with 6% or 12% gel, then stained with silver or CBB, respectively. As expected (26,27), BopB, BopN, BopD, and Bsp22 were detected specifically in the lanes where the wild-type supernatants were loaded. Additionally, ten bands were observed specifically in the lanes where the wild-type supernatants were loaded (Fig. 1A, arrows). Each band was excised from CBB-stained gels and treated with trypsin. The resulting peptides were analyzed by MALDI-TOF MS or ESI-MS/MS. Three bands were assigned as degraded products of Bsp22 or BopD (Fig. 1A, small arrows in right panel). The ESI-MS/MS analysis indicated that the deduced amino acid sequence of a trypsinized peptide derived from band a (Fig. 1A) was LLEPNNDEFVR (Fig. 1B). This sequence corresponds to amino acid residues 34 -44 of a 69-kDa hypothetical protein that is encoded in an ORF located at nucleotides 4,502,054 -4,504,030 of the B. bronchiseptica genome that Parkhill et al. published (34). By the MALDI-TOF MS analysis of trypsinized peptides from band b, five molecular masses, 961.4953, 1345.7384, 1541.8305, 2089.0730, and 2342.3590, were obtained. These masses corresponded to the following amino acid residues of the 69-kDa protein described above, respectively, 161-168, 34 -44, 149 -160, 169 -186, and 214 -235 (Table 1). Although the other five protein bands indicated by small arrows in Fig. 1A (left panel) were also analyzed by both MALDI-TOF MS and ESI-MS/MS, clear results were not obtained. These results strongly suggest that bands a and b were derived from the presumed 69-kDa protein, designated BopC in this study.
Construction and Secretion Profile of the bopC-deficient Mutant-To examine whether or not bands a and b shown in Fig. 1 are really derived from 69-kDa BopC, a mutant of bopC was constructed, as documented under "Experimental Procedures." The culture supernatant samples prepared from the wild-type strain, BopC mutant (⌬BopC), BopC mutant harboring pBopC (⌬BopC/pBopC), and type III Ϫ were subjected to SDS-PAGE, and the resulting gels were stained with CBB or silver (Fig. 2, A and B). In the ⌬BopC culture supernatant, both bands corresponding to a and b shown in Fig. 1A disappeared. In contrast, both bands were detected in the ⌬BopC/ pBopC-complemented strain (Fig. 2, A and B). The band intensities of other type III-secreted proteins, such as BopB, BopN, BopD, and Bsp22, were slightly higher in the supernatant sample prepared from ⌬BopC than in those from the wild-type strain ( Fig. 2A), and the band intensities of those proteins from the ⌬BopC/pBopC strain were similar to those from the wild-type strain ( Fig. 2A). Interestingly, five bands also disappeared in the ⌬BopC supernatant sample (Fig. 2B, small arrows). The results of immunoblot analysis using anti-BopC antibodies indicated that the specific signals were detected in band a (Fig. 2C, large arrow), band b (data not shown), and in the other five bands (Fig. 2C, small arrows). Four of these bands were each larger than 200 kDa. These results clearly demon-

TABLE 1 Tryptic peptides obtained from band b in Fig. 1A by MALDI-TOF MS analyses
The identities of the peptides were assigned by matching the measured monoisotopic mass values of single-charged ions (͓MϩH͔ ϩ ) to those calculated for the predicted p69 tryptic fragments.

No.
Mass strated that bands a and b were derived from BopC. This protein seemed to form multimeric complexes during SDS-PAGE.
BopC Is Required to Induce Cytotoxicity in Mammalian Cells-Infection of cultured mammalian cells with wild-type B. bronchiseptica induces morphological changes and injury to the host cell membrane, and these events depend on the TTSS (14,25,27). To examine whether or not the type III-secreted protein BopC is involved in these cytotoxic phenotypes, L2 rat lung epithelial cells were exposed to B. bronchiseptica. The cells infected with bacteria for 20 min were stained with Giemsa solution and examined under a light microscope (Fig. 3A). Almost 95% of the L2 cells infected with the wild-type or ⌬BopC/pBopC strain were detached from substrata, and the remainder of the adherent cells had shrunken cytoplasm and condensed nuclei. In contrast, the cells infected with ⌬BopC or type III Ϫ had the same normal morphology (Fig. 3A) as uninfected cells (data not shown). Next, HeLa cells were infected with B. bronchiseptica. The release of LDH into the extracellular medium was measured, and the released amount was compared with that of a positive control (Triton X-100 treatment) (Fig. 3B). When the cells were infected with the wild-type strain or ⌬BopC/pBopC for 120 min, the amount of LDH release reached ϳ75% or 50%, respectively. In contrast, neither ⌬BopC nor type III Ϫ showed any ability to elicit LDH release in the infected cells, even when the infection time was prolonged to 360 min (Fig. 3B). The amount of LDH release from HeLa cells treated with staurosporine for 5 h was very similar to that of uninfected cells (data not shown). To calculate the percentage of non-viable cells, HeLa cells were infected with the various strains of B. bronchiseptica for 60 min, and the number of dead HeLa cells was counted by trypan blue staining (Fig. 3, C and D). About 57% or 44% of cells infected with the wild-type strain or ⌬BopC/pBopC, respectively, were stained by trypan blue (Fig. 3D). In contrast, cells treated with staurosporine were scarcely stained at this time point (Fig. 3D).
As reported previously (25), although B. bronchiseptica induces necrotic cell death, in which caspases are not activated, there is a population of infected cells having TUNEL-positive signals in their nuclei (26 -28). In this population, the nuclei were stained diffusely by TUNEL reagent (26,27). In contrast, the nuclei of typical apoptotic cells, such as cells treated with staurosporine for 5 h (Fig. 3E), showed condensed pattern in TUNEL staining (42). To analyze the involvement of BopC in cell death, HeLa cells infected with B. bronchiseptica were stained with TUNEL reagent (Fig. 3E). The 15% or 10% of cells infected with, respectively, the wild-type strain or ⌬BopC/pBopC for 120 min had TUNELpositive nuclei (Fig. 3F). However, their nuclei were diffusely stained, and typical apoptotic cells, including chromatin condensation, were not observed. No TUNEL-positive nuclei were observed in cells infected with ⌬BopC or type III Ϫ (Fig. 3, E and F). These results indicate that BopC is required for the induction of mammalian cell death.
BopC Is Not a Pore-forming Factor for Effector Translocation-It is well known that TTSS-possessing bacteria, including B. bronchiseptica, induce hemolysis in which hemoglobins leak through the pores formed by type III-secreted proteins, so-called translocators. Although B. bronchiseptica has a cyclic adenylase toxin/hemolysin, we have measured the TTSS-dependent hemolytic activity and demonstrated that type IIIsecreted proteins BopB and BopD are translocators in B. bronchiseptica (27,28). To determine whether BopC is included in the translocator or not, ⌬BopC was exposed to rabbit red blood cells for 30 min, and hemolysis was measured as described under "Experimental Procedures" (Fig.  3G). The hemolytic activity of ⌬BopC (29.3% against positive control in which red blood cells were treated with 1% Triton X-100) was at almost the same level as that of the wild-type strain (33.4%), indicating that BopC is not a translocator.
BopC Induces Dephosphorylation of PY in Host Cells-A previous study showed that the wild-type B. bronchiseptica induces dephosphorylation of infected L2 cells and that this phenotype is dependent on the TTSS (14). To examine whether or not BopC is involved in the dephosphorylation of PY in host cells, immunofluorescence and immunoblot assays were performed using anti-PY antibodies (Fig. 4A). HeLa cells infected with the wild-type strain, ⌬BopC, ⌬BopC/pBopC, and type III Ϫ were stained with not only anti-PY antibodies but also rhodamine-phalloidin (Fig. 4A) or anti-B. bronchiseptica antisera (supplemental Fig. S1) to visualize F-actin or bacterial adherence to host cells, respectively. The results of fluorescence microscopy showed that fluorescent signals from adherent bacteria had almost the same intensity regardless of whether the cells were infected for 60 min with strains used here (supplemental Fig. S1). This suggests that BopC is not involved in bacterial adherence Supernatant samples were prepared from the wild-type strain, BopC mutant (⌬BopC), BopC mutant harboring pBopC (⌬BopC/pBopC), and type III Ϫ . The resulting samples were separated by SDS-PAGE with 12% (A) or 6% (B and C) gels. The gels were stained with CBB (A) or silver (B), and a 6% gel was subjected to immunoblot assay using anti-BopC antibodies (C). Large arrows indicate the bands a or b shown in Fig. 1A. The six arrows in panel C indicate BopC-specific bands. to infected cells. In the case of HeLa cell infection with the wild-type strain, all cells infected for 0 min still had specific PY signals that seemed to localize mainly in focal adhesions. However, the cells infected for 20 or 60 min had a certain proportion of cells with no PY signals (Fig. 4A,  arrowheads). In contrast, all cells infected with ⌬BopC or type III Ϫ for 60 min still had PY signals to the same extent as did uninfected cells (Fig.  4A). Next, lysates were prepared from HeLa cells infected with B. bronchiseptica and were subjected to immunoblot assay using anti-PY or anti-actin antibodies. The amount of each loaded sample was confirmed by the actin signal intensity (Fig. 4B). As reported previously (14), the PY signal profiles were changed and the signal intensity of signals was significantly lower in HeLa cells infected with the wild-type strain or ⌬BopC/pBopC for 60 min than that with type III Ϫ (Fig. 4B). In contrast, HeLa cells infected with ⌬BopC had almost the same profiles as those from the cells infected with type III Ϫ or from uninfected cells (Fig. 4B). These results indicate that BopC induces dephosphorylation of PY in infected mammalian cells.
Dephosphorylation of PY Occurs after the BopC-dependent Cell Death-Upon infection with wild-type B. bronchiseptica, BopC induces in both cell death (estimated by LDH release and TUNEL assays) and dephosphorylation of PY in host cells. To determine which phenomenon, cell death or dephosphorylation, is induced by BopC more directly, HeLa cells were infected with the wild-type B. bronchiseptica in the presence or absence of a dephosphorylation inhibitor, sodium orthovanadate (Na 3 VO 4 ). First, the degree of inhibition was examined. Cells exposed for 60 min were immunostained as described above, and then the PY signal-lacking cells were counted under fluorescent microscope. Although the bacterial adherence was not affected by addition of sodium orthovanadate (supplemental Fig.  S2), this inhibitor reduced the proportion of PY signal-lacking cells from 19.3% to 9.8% (Fig. 5, A and B). Lysates were prepared from these infected cells and were subjected to the immunoblot assay (Fig. 5C). As reported previously (14), signal intensities of PY proteins were increased in the presence of the inhibitor (Fig. 5C). These results indicate that treatment with Na 3 VO 4 partially attenuates the extent of PY dephosphorylation induced by B. bronchiseptica infection. Next, to examine whether or not the B. bronchiseptica-induced dephosphorylation is associated with DNA damage to the host cells or injury to the host membrane, TUNEL and LDH release assays were performed on HeLa cells infected with the wild-type strain in the presence of an inhibitor. The rate of TUNEL-positive cells and the amount of released LDH were almost the same in the presence or absence of the inhibitor (Fig. 5, D-F). In fluorescent microscopy, TUNEL-positive cells were detected in both PY-positive (Fig. 5D, arrow in top row) and PY-lacking (Fig. 5D, arrowhead in top row) cells. In contrast, the PY signallacking phenotype was not detected in the TUNEL-negative cells. These findings suggest that the decrease in PY proteins in host cells infected with B. bronchiseptica occurred after BopC-dependent cell death.
BopC Seems to Be Translocated into Mammalian Cells by TTSS-It has been well known that type III effectors are delivered into host cells directly from bacterial cells. The TEM-1 reporter system (38) was adopted to examine whether or not BopC is translocated into the host cell during wild-type B. bronchiseptica infection. Plasmid encoding TEM-1 (␤-lactamase) fused at its N terminus with each of the following: the full-length of BopC (658 aa); BopC truncated versions (448, 248, or 48 aa of the BopC N terminus); BcrH2 (cytoplasmic protein of B. bronchiseptica (28)); or Map (one of the EPEC effectors) was constructed. The resulting plasmids and pTEM-1 (encoding TEM-1 alone) were  introduced into respective ⌬BopC, ⌬BopB, and type III Ϫ . These strains were cultured in SS broth to the stationary phase, and then whole cell lysates and supernatant fractions were prepared. These samples were subjected to immunoblot analysis with anti-TEM-1 antibodies (Fig. 6A). All signals of the TEM-1-fused proteins or TEM-1 alone were detected in whole cell lysates as their deduced molecular masses (Fig. 6A, asterisks). The band intensities were similar among ⌬BopC, ⌬BopB, and type III Ϫ . As expected, although the signal was detected for the full-length of BopC fused with TEM-1 (FL-TEM-1) in the supernatant sample prepared from ⌬BopC or ⌬BopB, it was not detected in that from type III Ϫ (Fig. 6A). Signals of BcrH2 fused with TEM-1 (BcrH2-TEM-1) or TEM-1 alone were not detected in the supernatant fractions prepared from ⌬BopC and ⌬BopB (Fig. 6A). Signals of the BopC truncated versions fused with TEM-1 (N1-TEM-1, N2-TEM-1, and N3-TEM-1) were also detected in supernatant fractions prepared from ⌬BopC and ⌬BopB (Fig. 6A) but not from type III Ϫ (data not shown). Interestingly, signals of EPEC Map fused with TEM-1 (Map-TEM-1) were detected in the supernatant fractions prepared from ⌬BopC and ⌬BopB. However, the signal intensity of the supernatant fraction prepared from ⌬BopB was much weaker than that from ⌬BopC (Fig. 6A). These results indicate that BopC fused with TEM-1 was secreted, as was native BopC, and 48 aa residues of N terminus were enough for BopC secretion via the TTSS. Next, HeLa cells were infected with B. bronchiseptica expressing TEM-1 fused protein, and then infected cells were stained with CCF2-AM solution. When TEM-1-fused proteins were translocated into host cell cytosol, the fluorescence emitted by CCF2-AM-stained cells turned from green to blue in fluorescence microscopy as observed by means of a 4Ј,6-diamidino-2-phenylindole filter set. Although the cells infected with ⌬BopC expressing FL-, N1-, N2-, BcrH2-, Map-TEM1, or TEM-1 alone still emitted green fluorescence (data not shown), ϳ70% of cells infected with ⌬BopC expressing N3-TEM-1 emitted blue fluorescence (Fig. 6B). As expected, the cells infected with ⌬BopB or type III Ϫ expressing TEM-1-fused proteins emitted green fluorescence (data not shown). These results suggest that BopC is translocated into the host cell via the TTSS, and the signal sequence of both secretion and translocation is located within 48 aa residues of the BopC N terminus.
BopC Has the Ability to Induce Cytotoxicity-To investigate whether cytotoxicity is induced by BopC alone, the bopC gene was introduced into mammalian cell lines. The mock vector or pcDNA-BopC was transfected to COS-7, HeLa, or 293T cells. After 18 h of incubation, the amounts of LDH released into the extracellular media were measured (Fig. 7). No significant LDH release was detected in the media of mock vector-introduced cells. When the bopC gene was introduced into COS-7, 293T, and HeLa cells, the amount of LDH release reached 22.0%, 13.4%, and 4.6%, respectively (Fig. 7). These results strongly suggest that BopC has the ability to induce cytotoxicity and that other factors were not involved in this series of events.

DISCUSSION
Thus far, no protein has been characterized as a definite type III effector in Bordetella. In this study, we identified a novel effector that plays a crucial role in necrotic cell death (Fig. 3). This 69-kDa protein is secreted via the TTSS of B. bronchiseptica (Fig. 2) and was designated as BopC (Bordetella outer protein involved in cytotoxicity) in this study. Proteins secreted via the TTSS are generally divided into two types: effectors on the one hand and pore-forming factors, called translocators, on the other (15). It is known that pore formation on the host membrane by translocators is a prerequisite for delivering the effector into the host cell, and this pore-forming activity is measurable by hemolytic activity (43,44). A translocator-deficient mutant exerts no hemolytic activity on red blood cells. ⌬BopC had clear hemolytic activity that was at the same level as that of the wild-type strain (Fig. 3G). Moreover, when the bopC gene was introduced into mammalian cell lines, a significant LDH release was induced (Fig. 7), strongly suggesting that BopC is the effector that triggers the host cell necrosis during Bordetella infection.
The results in Fig. 4 suggested that BopC is a protein-tyrosine phosphatase (PTPase), such as the Yersinia type III effector, YopH, which has PTPase activity (45). To examine this possibility, the PTPase activity in the culture supernatant of the wild-type B. bronchiseptica or purified recombinant BopC was measured by using a Universal Tyrosine Phosphatase Assay kit (Takara). However, no detectable PTPase activity was found (data not shown). It has been shown that the PTPase family members possess an essential cysteine residue in their catalytic site (46). For example, the YopH mutant, the 403rd cysteine residue of which was changed to an alanine residue, has no PTPase activity (47). In contrast, BopC has no cysteine residue in its sequence (supplemental Fig. S4), suggesting that the dephosphorylation of PY induced by the wild-type B. bronchiseptica was not a direct effect of BopC. To speculate as to the functions of BopC, several programs were used to detect homologous proteins and conserved domains. However, neither the BopC homolog nor the conserved domain was found in searches of BLAST, FASTA, Conserved Domain Data base (www.ncbi.nlm.nih.gov/Structure/cdd/ cdd.shtml), BLOCKS (48,49), ProDom (50,51), PRINTS (52,53), and Pfam (54). The Phyre (www.sbg.bio.ic.ac.uk/ϳphyre/) program and 3D-PSSM (www.sbg.bio.ic.ac.uk/ϳ3dpssm/index2.html) were also used to search structural homologous proteins against BopC. Again, no confident predictions were obtained. BopC in B. bronchiseptica RB50 has been registered in National Center for Biotechnology Information (NCBI, www.ncbi.nlm.nih.gov/) as a 69-kDa hypothetical protein with accession number is BX640449 (nucleotide, supplemental Fig. S3) and NP_890763 (amino acid; supplemental Fig. S4). BopC in B. pertussis Tohama I and B. parapertussis 12822 were also registered in NCBI as NP_879352 and NP_885936, respectively. Although the BopC expression is unknown in both species, BopC has high identity (Ͼ95%) among the three species (supplemental Fig. S4), suggesting that the function of BopC is probably equivalent among the Bordetella species. In B. bronchiseptica, the genes encoding BopB, BopD, BopN, and Bsp22 are located within the TTSS locus (26,27). In contrast, the bopC gene is located outside the locus, at 2.5 Mbp (in B. bronchiseptica), 1.7 Mbp (in B. parapertussis), and 1.8 Mbp (in B. pertussis) from their respective TTSS loci. It is unclear why the BopC locus on the genome is separated from the TTSS locus. However, similar observations has been made in other effectors, such as SopE (55) and SigD (56) in Salmonella, Orf3 in EPEC (57), and NleA in enterohemorrhagic E. coli (58). The 10-kbp upstream and downstream flanking regions of bopC in both B. bronchiseptica and B. parapertussis encode nearly identical proteins, which are mainly housekeeping proteins. One exception to this similarity is a gene encoding a transposase homolog located 5-kbp upstream from bopC in B. parapertussis but not found in B. bronchiseptica. In contrast, in B. pertussis, genes encoding transposase and integrase homologs are located 2.0 kbp upstream and 1.6 kbp downstream from bopC, respectively. Neither insertion is found in those regions of B. bronchiseptica or B. parapertussis, whereby the outside region of bopC in B. pertussis encodes different proteins from those of B. bronchiseptica or B. parapertussis. These findings suggest that the extensive genome arrangements, including recombinations, gene conversions, and transpositions, occurred during the divergence of B. pertussis from the progenitor, B. bronchiseptica (3).
The BvgAS system was shown to be down-regulated when bacteria were cultured in the presence of MgSO 4 (12). To analyze whether or not BopC is regulated by the BvgAS system, the whole cell lysate was prepared from the wild-type B. bronchiseptica cultured in SS broth containing MgSO 4 and was subjected to an immunoblot assay using anti-BopC antibodies. No BopC signal was detected in the whole cell lysate (data not shown), suggesting that the expression of BopC is regulated positively by the BvgAS system.
Although the deduced molecular mass of BopC is 69 kDa, the secreted BopC bands appeared at ϳ150 kDa and at over 200 kDa as multimers on SDS-PAGE (Figs. 1A and 2B). The monomer band of BopC was detected only as a faint band on overexposed x-ray film (data not shown). It has been reported that polytopic membrane proteins, which have been defined generally as proteins having multiple transmembrane regions, form an SDS-resistant multimer (60). For example, DotA, which is secreted via the type IV secretion machinery in Legionella pneumophila, has eight transmembrane domains (61) and forms SDS-resistant multimers that were detected in the immunoblot assay, although the signal of the DotA monomer was stronger than that of the DotA multimer (62). The DotA multimer disappeared when the sample was loaded on the gel without boiling (62). In contrast, the BopC multimer was still detected and the monomer band intensity was not increased, even when the whole cell lysate or the supernatant sample was prepared without boiling (data not shown). Prediction of the transmembrane domain in BopC by the TMpred program (www.ch.embnet. org/software/TMPRED_form.html) showed that one probable transmembrane domain exists in the C terminus (amino acid residues 619 -643). This suggests that BopC is unlikely to be a typical polytopic protein. When TEM-1 was fused with the C terminus of the full-length BopC, the monomer band was detected clearly, although the multimer bands of the fused protein were still detected (Fig. 6A, lane FL-TEM-1). In contrast, multimer bands were not detected in the C-terminal truncated versions of BopC (BopC-N1-TEM-1), suggesting that the C-terminal region (amino acid residues 449 -658) is needed to form multimers. These findings suggest that BopC would interact with itself tightly like polytopic proteins, such as DotA.
The Xanthomonas TTSS can secrete the Yersinia effector YopE (63), and the Yersinia TTSS recognizes Pseudomonas effectors AvrB and AvrPto as substrates (64). In this study, we also confirmed that the EPEC effector Map fused with TEM-1 (Map-TEM-1) was secreted via the B. bronchiseptica TTSS (Fig. 6A). However, the translocation of Map-TEM-1 into infected HeLa cells was not detected in the TEM-1 assay (data not shown). One possibility is that Map may be unstable in B. bronchiseptica. Indeed, degraded Map-TEM-1 products were detected, and a small amount of the intact length of Map-TEM-1 was detected in the supernatant sample (Fig. 6A). Presumably, CesT, a chaperone for Map (65), is required to stabilize Map in B. bronchiseptica. The amount of secreted Map-TEM-1 from ⌬BopB is lower than that from ⌬BopC, probably due to the competition between the endogenous BopC and Map-TEM-1 in the secretion capacity of the TTSS. The amounts of type III-secreted proteins, such as BopB, BopD, BopN, and Bsp22, in ⌬BopC are higher than they are in the wild-type strain and were decreased by the introduction of pBopC into ⌬BopC (Fig. 2A). These findings suggest that BopC has a regulatory effect on other type III-secreted proteins.
We showed that the BopC-N3-TEM-1 protein was translocated into HeLa cells during infection (Fig. 6B). However, in TEM-1 assays, translocation into host cells was not detected in three BopC derivatives, including full-length BopC. Conceivably, these BopC-TEM-1 derivatives are unstable in B. bronchiseptica proteins, because the degraded products were detected in the supernatant samples (Fig. 6A); or because the amount of delivered BopC-TEM-1 proteins is insufficient for detection by TEM-1 assay. The precise reason why only BopC-N3-TEM-1 was detected in the TEM1 translocation assay is unknown. However, the signal intensities of BopC-N3-TEM-1 in whole cell lysates and supernatant samples seemed to be stronger than those of any other fusion protein used (Fig. 6A, asterisks). In this context, we speculate that detectable amounts of TEM-1 fusion protein were translocated to host cells in the case of BopC-N3-TEM-1. Indeed, even in the case of HeLa cells infected with ⌬BopC expressing BopC-N3-TEM-1, the infected cells turned greenish-blue, whereas HeLa cells infected with EPEC expressing Map-TEM-1 turned from green to rich blue (data not shown).
Collectively, BopC is the first type III effector to be identified in Bordetella, and this effector is crucial for inducing cytotoxicity in cultured mammalian cells. The mechanisms of necrotic cell death induced by B. bronchiseptica infection in mammalian cells will be clarified by analyses of the detailed function(s) and localization of translocated BopC.