Vibrio vulnificus IlpA-induced Cytokine Production Is Mediated by Toll-like Receptor 2*

Vibrio vulnificus is a pathogenic bacterium causing primary septicemia, which follows a classical septic shock pathway, including an overwhelming inflammatory cytokine response. In this study, we identified a putative lipoprotein of V. vulnificus, encoded by the ilpA gene, as one of the surface proteins that specifically reacted with the antibodies raised against outer membrane proteins of V. vulnificus. Using a mutant V. vulnificus in which its ilpA gene was knocked out, we found that IlpA is important in the production of interferon-γ in human peripheral blood mononuclear cells. Production of tumor necrosis factor-α and interleukin-6 is also induced by the recombinant IlpA (rIlpA) in human monocytes. Lipidation of the rIlpA was observed by in vivo labeling in Escherichia coli. Experiments using the mutant IlpA, which is unable to be modified by lipidation, indicate that the lipid moiety of this protein has an essential property for cytokine production in human cells. Pretreatment of monocytes with antibodies against Toll-like receptor 2 (TLR2) inhibited production of both tumor necrosis factor-α and interleukin-6. The role of TLR2 in IlpA-induced cytokine production was confirmed by an in vitro assay, in which only the TLR2-expressing cells showed a dramatic induction of nuclear factor-κB activity by rIlpA. In addition, rIlpA treatment resulted in induction of TLR2 transcription in human cells. In comparison with the wild type V. vulnificus, the ilpA mutant showed a reduced mortality in mice. These results demonstrate that IlpA of V. vulnificus functions as an immunostimulant to human cells via TLR2.

Vibrio vulnificus, a Gram-negative bacterium found commonly in the estuarine environment, has been frequently associated with primary septicemia following the consumption of contaminated shellfish. Over 50% of the primary septicemia patients caused by V. vulnificus died of multiorgan failure as a result of a rapidly progressive shock syndrome (1,2). Extracellular substances produced by V. vulnificus, such as hemolytic cytolysin (3,4) and elastase (5), had been extensively studied as candidate virulence factors responsible for its pathogenesis. Surface structures such as lipopolysaccharide (LPS) 2 (6,7) and outer membrane proteins (8,9) were also studied as candidates for V. vulnificus virulence factors. Based on the attenuated mouse lethality by a noncapsulated mutant V. vulnificus, capsular polysaccharide was also proven to be important in the pathogenesis of V. vulnificus (10). Type IV pilin was confirmed to be involved in the virulence of V. vulnificus via genetic deletion of the pilD or pilA genes (11,12). In addition, motility was discovered as a virulence determinant of V. vulnificus (13,14).
One of the distinct characteristics in V. vulnificus pathology is a rapidly progressing septic shock syndrome (15,16). Septic shock usually results from the overproduction and dysregulation of host cytokines in response to invading microorganisms. Inflammation-associated cytokines, such as tumor necrosis factor-␣ (TNF-␣), interferon-␥ (IFN-␥), interleukin-1 (IL-1), and IL-6, play pivotal roles in the host immune response to infection (17,18). A variety of bacterial products, including LPS, capsular polysaccharide, peptidoglycan, lipoarabinomannans, and porins, has been identified to elicit or modulate the release of cytokines from host cells in both in vivo and in vitro models (19). For example, initial interaction of Neisseria gonorrhoeae with mucosal epithelial cells triggers the release of inflammatory cytokines, including IL-6 and IL-8, which subsequently recruit and activate other immune cells at the site of infection (20,21).
Therefore, the involvement of V. vulnificus surface molecules(s) in its interaction with host cells can be postulated, which may be recognized by immune cells and thus trigger cytokine production. The interaction between the surface molecules of V. vulnificus and the immune cells may occur by specific recognition via the receptors on the immune cells. A variety of host cell receptors has been implicated in the recognition of bacteria or their components. One of them is the Toll-like receptor (TLR) family, which plays a central role in innate immune defenses (22,23). The human TLR family is categorized into at * This work was supported by the 21st Century Frontier Microbial Genomics and least 10 distinct receptors, which convey information in response to different microbial components resulting in activation of nuclear factor-B (NF-B), the transcription factor involved in the expression of proinflammatory cytokines, chemokines, and adhesion molecules (24). The ligands recognized by each TLR have the conserved molecular patterns shared by a broad range of pathogens. For instance, TLR4 is a principal signal transducer in the recognition of LPS (25,26). TLR2 confers responsiveness to various bacterial compounds, such as bacterial lipoprotein (27), lipoarabinomannan (28), lipoteichoic acid (29), and peptidoglycan (30).
In this study, we isolated an outer membrane protein of V. vulnificus, which is highly immunoreactive, and identified it as a lipoprotein. The lipidated form of this protein stimulates the production of proinflammatory cytokines in human monocytes. In vitro assays further indicate that TLR2 is required for recognition of this V. vulnificus lipoprotein.
Immunoscreening of the V. vulnificus Expression Library-Outer membrane proteins (OMPs) were prepared from V. vulnificus ATCC29307 and used for intravenous immunization of a rabbit as described (14). The resultant polyclonal antibodies specific to OMPs were used for the immunoscreening experiment. The V. vulnificus expression library was constructed by cloning genomic DNA partially digested with Sau3AI into a ZAPII vector (Stratagene). Recombinant phage DNA was packaged in vitro using Gigapack Gold as described by the manufacturer (Stratagene). The amplified library (with a titer of 9 ϫ 10 6 plaque-forming units/ml) was screened primarily using a 1:2,000 dilution of rabbit antibodies raised against V. vulnificus OMPs. Plaques distinct from the background level of antibody binding to the filter were further purified by the second and third screenings. Homogeneous plaques expressing the surface antigens were selected and excised to a pBK-CMV phagemid, as instructed by the manufacturer (Stratagene). The identities of the phagemid inserts were verified by automatic sequencing of the double-stranded plasmids.
Construction of the ilpA Knock-out Mutant V. vulnificus-An 840-bp region upstream of the ilpA open reading frame (ORF) was amplified from the genomic DNA of V. vulnificus ATCC 29307 using the following two primers: ilpA-upF (5Ј-GTC- sucrose, and sensitivity to chloramphenicol) were further confirmed by PCR using primers ilpA-upF and ilpA-downR, and finally given the name YS101.
Complementation of ⌬ilpA Mutant with the Intact ilpA Gene-An intact ilpA gene was amplified from the genomic DNA of wild type V. vulnificus ATCC29307 by PCR using the following primer set: ilpA-comF (5Ј-GGTTGGATCCATTG-GTGAGCT-3Ј; the underlined sequence denotes a BamHI restriction site) and ilpA-comR (5Ј-TATCAAGCTTCTCTT-GGGATCATTTGAAA-3Ј; the underlined sequence indicates a HindIII restriction site). The amplified ilpA DNA fragment of 1,159 bp was digested with BamHI and HindIII, and then cloned into a broad range host plasmid, pLAFR5 (32), resulting in the positioning of ilpA ORF downstream of the lac promoter in pLAFR5. The resultant plasmid, pLAFRilpA, in E. coli SM10 pir was transferred into the ilpA mutant (YS101) by conjugation, and the exconjugants were selected on thiosulfate citrate bile sucrose agar containing tetracycline (2 g/ml).
Expression and Purification of Recombinant IlpA Proteins-Two oligonucleotides, lipo-F (5Ј-CATGCCATGGCTATGAA-ATTTAGCCTTAAAGG-3Ј; the underlined sequence denotes an NcoI restriction site) and lipo-R (5Ј-CCCAAGCTTCCAG-CCTTTTACTACGCC-3Ј; the underlined sequence represents a HindIII restriction site), were used to amplify an 810-bp DNA fragment containing the complete ORF of the ilpA gene from the genomic DNA of V. vulnificus. NcoI and HindIII sites located at both ends of the resultant ilpA DNA were used to clone this DNA into the pET28b expression plasmid (Novagen) to generate the plasmid pETilpA. The recombinant IlpA (rIlpA) protein was overexpressed in E. coli BL21(DE3) carrying pETilpA by adding isopropyl thio-␤-D-galactoside (IPTG; Sigma) at a concentration of 1 mM, and purified using a Ni 2ϩnitrilotriacetic acid affinity column as directed by the manufacturer (Qiagen).
Using primers, mlipo-F (5Ј-CATGCCATGGATGGGC-GAAAAAGCGACTGAC-3Ј; the underlined sequence denotes an NcoI restriction site) and lipoR, we prepared the truncated ilpA gene, which lacked the 5Ј-end of DNA fragment (69 bp) containing the consensus sequence for lipidation. The resultant ilpAЈ DNA fragment of 741 bp was then used to clone into pET28b as described above. The truncated rIlpA protein was overexpressed in E. coli BL21 (DE3) carrying the resultant plasmid, pETmilpA, and purified in the same manner as the intact rIlpA protein.
Determination of the Recombinant IlpA Protein Lipidation-In vivo labeling of rIlpA protein with [ 3 H]palmitic acid was performed as described (33). Briefly, various E. coli strains, DH5␣ carrying pET28b, DH5␣ carrying pETilpA, BL21 (DE3) carrying pET28b, and BL21 (DE3) carrying pETilpA, were grown to the mid-log phase (A 600 ϭ 0.6 -0.8), and the incubation was continued for 12 h in the presence of 50 Ci of [ 3 H]palmitic acid (Amersham Biosciences) without or with IPTG. After the labeling was stopped by adding 10% trichloroacetic acid (w/v), the labeled bacterial cells were resuspended in boiling buffer (2% (w/v) SDS, 50 mM Tris-Cl, pH 8.0), boiled for 10 min, and then broken by sonication. The bacterial extracts were then phase-partitioned into aqueous and detergent phases with the addition of 5% Triton X-114 at 37°C. The pellet fraction was washed three times with radioimmunoprecipitation buffer (1% (v/v) Nonidet P-40, 0.5% (w/v) deoxycholic acid, 0.1% (w/v) SDS, 50 mM Tris-Cl, pH 8.0) before being resuspended in sample buffer for SDS-PAGE. An equal microgram of each fraction was subjected to 10% (w/v) SDS-PAGE and transferred to a polyvinylidene difluoride membrane. The membrane was treated with Amplify solution (Amersham Biosciences) and exposed to preflashed Hyperfilm-MP (Amersham Biosciences) for 2 weeks at Ϫ70°C.
Western Blot Analysis-Purified rIlpA (100 g) was mixed with 0.5 ml of complete Freund's adjuvant (Sigma), and injected intraperitoneally into a specific pathogen-free rat (CrjBgi: CD[S.D.]IGS, 7-week-old, female). Two additional immunizations were performed with the same amount of rIlpA protein mixed with incomplete Freund's adjuvant (Sigma) at 2 and 4 weeks after the primary immunization. A week after the third immunization, serum was obtained from the immunized rat and used for Western blot analysis.
Measurement of Cytokines-PBMCs and monocytes in 96-well plates (1 ϫ 10 5 cells per well) were coincubated with culture medium, bacterial lysates, or rIlpA protein at the indicated concentration for 18 h. For comparison, cells were also treated with commercially available E. coli LPS (Sigma). Cellfree supernatants were then collected by centrifugation, transferred to new tubes, and stored at Ϫ70°C until analysis. The levels of each cytokine in the supernatants were determined using enzyme-linked immunosorbent assays (BD Biosciences). Concentration of each cytokine was determined based on the standard curve of the corresponding cytokine, which had been purified and was available commercially (IFN-␥ from BD Biosciences; TNF-␣ and IL-6 from BIOSOURCE).
Blocking Experiments with Polymyxin B or with Anti-TLR2 Antibodies-Monocytes were preincubated for 1 h with medium containing 10 g/ml human TLR2 monoclonal antibodies (BioLegend) or its isotype control IgG (BioLegend). To examine the role of residual LPS in the prepared rIlpA protein fraction on cytokine production by monocytes, polymyxin B (Sigma) was added to the rIlpA at a concentration of 20 g/ml prior to being incubated with the prepared monocytes.
RT-PCR of TLR mRNAs-Using TRIzol reagent (Invitrogen), total RNAs were isolated from human PBMCs (2 ϫ 10 6 ) stimulated with either rIlpA or E. coli LPS (Sigma) at the concentration of 1.0 g/ml. As a control, total RNA was also prepared from the same number of PBMCs. In addition, HEK293-hTLR2/CD14 (indicated as 293-hTLR2) and HEK293-hTLR4/MD2-CD14 (indicated as 293-hTLR4) cell lines (Table 1) were treated with either rIlpA or LPS and subjected to RNA preparation.
LD 50 Determination-For determination of LD 50 , specific pathogen-free, 7-week-old, female ICR mice were used without pretreatment with iron dextran. Cultures of bacterial strains grown overnight in LBS medium were freshly cultivated in the same medium up to an A 600 of 0.7, harvested, washed once in PBS, and then resuspended in PBS. One hundred microliters of FIGURE 1. Genetic organization of the insert DNA in plasmid pEX3, and identification of the gene encoding the protein that reacted with anti-OMPvv antibodies. A, pEX3 is one of the immunoreactive clones, which was isolated by immunoscreening of the V. vulnificus expression library with the antibodies raised against OMPs of V. vulnificus (OMPvv). A 3.66-kb DNA fragment of pEX3 contains four complete ORFs that encode an ABC permease, a lipoprotein, a hypothetical protein, a SecY-interacting protein, and a partial ORF for a hypothetical protein. Sizes of the putative proteins are indicated in parentheses. Primers used for the construction and confirmation of a ⌬ilpA mutant V. vulnificus are denoted with arrows. B, 20 g of protein extracts from E. coli XLOLR and E. coli XLOLR carrying pEX3 was subjected to Western analysis using antibodies against OMPvv. An ϳ30-kDa protein appeared as an immunoreactive band and is indicated by an arrow. Lane 1, lysate prepared from XLOLR; and lane 2, lysate prepared from XLOLR carrying pEX3. serial dilutions of the bacterial suspension were then injected intraperitoneally into six mice per dilution group. The numbers of dead mice were determined 24 h after the injection, and the LD 50 value was calculated using the equation described previously by Reed and Muench (37).
Nucleotide Sequence Accession Number-The nucleotide sequences of the ilpA gene isolated from V. vulnificus ATCC29307 were deposited in the GenBank TM data base under the accession number of DQ177330.
Statistical Analysis-Data are presented as means Ϯ S.D. derived from at least three independent experiments. Student's t test (SYSTAT program, SigmaPlot version 9; Systat Software Inc.) was used to evaluate the statistical significance of the results.

RESULTS
Isolation of the Immunoreactive Clone from Immunoscreening V. vulnificus OMPs-Approximately 30,000 plaques from the ZAPII-based expression library of V. vulnificus were screened to isolate the clones expressing the proteins reacting with the antibodies against V. vulnificus OMPs. Twelve candidate plaques showing reproducible immune response with anti-OMP serum were isolated, and the insert DNAs in the isolated phages were excised to pBK-CMV phagemids. Among the 12 isolated clones, 3 of them, pEX3, pEX19, and pEX21, had an identical 3.66-kb DNA fragment (Fig. 1A). The DNA insert of pEX3 was found to contain four complete ORFs coding for an ABC permease, a putative lipoprotein, a hypothetical protein, a SecY-interacting protein (Syd), and a partial ORF encoding another hypothetical protein.
To examine which gene product was responsible for the immune reaction with anti-OMP serum, the lysate of E. coli XLOLR containing pEX3 was subjected to a Western blot analysis using anti-OMP serum (Fig. 1B). As a control, lysates of E. coli XLOLR without any plasmid were reacted with the same serum. A strong immunoreactive band of ϳ30 kDa was detected only in the lysate of XLOLR with pEX3. The molecular mass of this immunoreactive protein, ϳ30 kDa, is a similar size to the tentative protein encoded by the second ORF in pEX3, which codes for a putative lipoprotein. Thus, this putative lipoprotein may be an antigenic molecule of V. vulnificus found in the immunoscreening assays.
Blast search of this putative lipoprotein reveals several homologous proteins found in other pathogenic Gram-negative bacteria (Fig. 2). The deduced amino acid sequences of this V. vulnificus lipoprotein (GenBank TM accession number DQ_177330) showed 84, 68, 53, and 59% identities to those of Vibrio cholerae (GenBank TM accession number NP_230552), Salmonella typhimurium (GenBank TM accession number NP_ 459249), E. coli (GenBank TM accession number YP_859264), and Haemophilus influenzae (GenBank TM accession number P_317282), respectively. None of these proteins has yet been reported with respect to their functional role. In all five proteins, however, there is a conserved signal sequence and a cysteine residue, which have been known to be required for lipidation (38). Therefore, we putatively designated this protein as IlpA, which stands for immunogenic lipoprotein A.
Lipidation of the Recombinant IlpA Protein-The deduced amino acid sequences of this putative protein contain a consensus lipobox sequence for lipid modification at its N terminus, which is Leu-Ala-Gly(Ala)-Cys (38). Using an expression system of the rIlpA in E. coli, we investigated whether IlpA protein is modified by lipidation as suggested from its amino acid sequences (Fig. 3). The rIlpA protein was expressed in E. coli BL21 (DE3) as a C terminus histidine-tagged form using the pET28b-based expression plasmid, pETilpA, in which the expression of the recombinant protein is induced by IPTG. E. coli DH5␣ was also used as a control strain, in which expression of the rIlpA was not induced by IPTG.
Four different E. coli strains, DH5␣ with pET28b, BL21 (DE3) with pET28b, DH5␣ with pETilpA, and BL21 (DE3) with pETilpA, were grown in the presence of [ 3 H]palmitic acid and examined for the presence of radiolabeled protein(s). Specifically, two of the four strains, DH5␣ carrying pETilpA and BL21 (DE3) carrying pETilpA, were cultured in the absence or presence of 1 mM IPTG and then compared side by side to examine whether the IPTG-induced protein might be labeled with [ 3 H]palmitic acid. Because proteins modified by lipidation have detergent-soluble properties, lysates of each culture were subdivided into aqueous, detergent, and pellet fractions by their solubility in Triton X-114.
A protein labeled with [ 3 H]palmitic acid, the size of which is ϳ30 kDa, was detected only in the lysates prepared from BL21 (DE3) with pETilpA, when grown in the presence of IPTG. This result indicates that the rIlpA protein is lipidated as suggested from its amino acid sequences. The lipid-containing property of the rIlpA was also supported by observing the strongest radioactivity in the detergent phase of IPTG-induced BL21 (DE3) bacteria with pETilpA.
Construction of ilpA Knock-out Mutant and Complementation of the Mutant-We constructed a mutant V. vulnificus, in which its ilpA gene was deleted from the wild type V. vulnificus ATCC29307 to examine the functional role of this protein in V. vulnificus. Two sets of primers were used to construct a knockout mutant, i.e. a set of two primers specific to the upstream region (ilpA-upF and ilpA-upR) and a second set of primers specific to the downstream region (ilpA-downF and ilpA-downR) of the ilpA gene (Fig. 1A). The resultant mutant, YS101, lacked a complete ORF of the ilpA gene. Deletion of the ilpA gene in a chromosome of the mutant V. vulnificus was confirmed by PCR using primers ilpA-upF and ilpA-downR (Fig.  4A). The resultant PCR product of the ilpA mutant V. vulnificus was 1,855 bp long, whereas wild type V. vulnificus with the intact ilpA gene produced a larger PCR product of 2,665 bp.
The cellular level of the IlpA protein was also examined in wild type and ⌬ilpA mutant V. vulnificus by Western blot analysis using polyclonal antibodies against rIlpA (Fig. 4B). A strong immunoreactive band of ϳ30 kDa was detected only in the wild type lysate. An intact ilpA gene of V. vulnificus was cloned into the broad host range plasmid pLAFR5 to produce pLAFRilpA. The ⌬ilpA mutant V. vulnificus carrying pLAFRilpA clearly demonstrated a formation of the immunoreactive protein (Fig. 4B).

Cytokine Production of Human PBMCs Induced by Cell Extracts of Various V. vulnificus Strains-Repeated isolation of
ilpA-containing clones from immunoscreening of V. vulnificus expression library with anti-OMP serum led us to examine the possibility that IlpA of V. vulnificus acts as an immunostimulant to induce cytokine production in human cells. Thus, we prepared a V. vulnificus lysate devoid of the IlpA protein using  the ⌬ilpA mutant YS101 and compared its activity to induce cytokine production with that of wild type V. vulnificus. When human PBMCs were treated with various concentrations of wild type lysate (ranged from 0.5-10.0 g/ml), the secretion of IFN-␥ by PBMCs was increased up to 1,310 pg/ml in a dose-dependent manner (Fig. 5). In contrast, treatment of PBMCs with the lysate of the ⌬ilpA mutant V. vulnificus produced a much lower level of IFN-␥ (41 pg/ml), which was a comparable value to those of the control cells, such as PBMCs treated with medium only or bovine serum albumin (13 or 32 pg/ml, respectively).
When lysate of the ⌬ilpA mutant V. vulnificus carrying pLAFRilpA was used to challenge PBMCs, IFN-␥ production was restored to the levels induced by the equivalent concentrations of wild type V. vulnificus lysate. This result suggests that V. vulnificus IlpA protein stimulates an immune response in human cells.
Cytokine Production of Human PBMCs and Monocytes Induced by rIlpA Protein-Based on the drastic difference in immunogenically stimulating activity between lysates of wild type and ⌬ilpA mutant V. vulnificus, we asked whether the IlpA protein itself was able to induce production of IFN-␥ in PBMCs (Fig. 6A). Indeed, when rIlpA protein was added at a concentration of 1.0 g/ml to PBMCs, IFN-␥ was increased to 1,080 pg/ml. This was equivalent to the IFN-␥ level in a positive control (1,320 pg/ml), in which PBMCs were treated with 1.0 g/ml concanavalin A, a well known molecule to induce production of IFN-␥ (39).
In subsequent experiments, we examined the ability of rIlpA to induce the production of cytokines in monocytes. Monocytes prepared from human PBMCs were challenged with various concentrations of rIlpA (0.01-1.0 g/ml) and assayed for production of TNF-␣ (Fig. 6B) and IL-6 ( Fig. 6C). As a positive control, E. coli LPS was used to stimulate cytokine production by monocytes. When these rIlpA-treated monocytes were assayed for cytokine production at 18 h post-stimulation, increasing amounts of both cytokines were detected in a dosedependent manner of rIlpA (up to 14,500 pg/ml for TNF-␣ and  were incubated with 1.0 g/ml rIlpA for 18 h, and the supernatant of each culture was used to determine the level of IFN-␥ production using a human IFN-␥ ELISA kit (BD Biosciences). PBMCs exposed to concanavalin A (con A) (1.0 g/ml) were included as a positive control for IFN-␥ production. In addition, PBMCs were also treated with medium or bovine serum albumin and assayed for cytokine production as negative controls. B and C, confluent monocytes (1 ϫ 10 5/ ml), which were prepared as described under "Experimental Procedures," were exposed to various concentrations of rIlpA protein ranging from 0.01 to 1.0 g/ml. Monocytes treated with E. coli LPS were used as a positive control for cytokine production. Cytokine concentrations were measured using a human TNF-␣ ELISA kit (BIOSOURCE) and a human IL-6 ELISA kit (BIOSOURCE). Error bars represent the standard deviations from three independent experiments. Measurement of each cytokine induced by rIlpA was performed in triplicate for each experiment.
10,200 pg/ml for IL-6). In response to the increased amount of LPS, monocytes displayed increased formation of these two cytokines, up to 8,050 pg/ml for TNF-␣ and 7,220 pg/ml for IL-6. These results indicate that rIlpA of V. vulnificus is as potent as E. coli LPS in inducing cytokine production by monocytes.
Treatment of rIlpA Protein with Polymyxin B-LPS of Gramnegative bacteria is known as a potent immunostimulant (40), and the rIlpA protein used in this experiment was prepared from the Gram-negative bacterial cell, E. coli BL21 (DE3). Thus we examined the possibility that cytokine production by rIlpAstimulated monocytes was caused from the residual contamination of LPS in the purified fraction of rIlpA. To exclude this possibility, polymyxin B, an amphipathic LPS-binding molecule, was used to antagonize the function of LPS. The polymyxin B-treated rIlpA was used to stimulate monocytes, upon which the production of TNF-␣ (Fig. 7A) and IL-6 ( Fig. 7B) was determined. As a positive control to confirm the proper functioning of polymyxin B, LPS was also pretreated with polymyxin B before it was used for monocyte activation. As expected, the ability of polymyxin B-treated LPS to stimulate cytokine production in monocytes was dramatically attenuated compared with the ability of untreated LPS-stimulated monocytes. In contrast, preincubation of rIlpA with polymyxin B did not affect the ability of rIlpA to induce cytokine production in monocytes. This result clearly demonstrates that the rIlpA protein used in this experiment was free of LPS contamination, and the cytokine production by rIlpA-exposed monocytes is derived from immunogenic activity of IlpA of V. vulnificus.
Role of Lipid Moiety of IlpA in Cytokine Production in Monocytes-The amino acid sequence of IlpA has a putative sequence for lipid modification, and lipidation of rIlpA was confirmed by in vivo labeling assay (Fig. 3). In the case of the Borrelia burgdorferi surface lipoprotein, its lipid moiety has been reported to be crucial for immunogenic function (41). To examine the role of the IlpA lipid moiety in cytokine production by monocytes, we synthesized a truncated rIlpA protein, in which 23 N-terminal residues were deleted. Because the missing region of the mutant rIlpA protein contains important amino acid residues required for lipidation, this protein was not expected to be modified by lipidation, as shown previously in other lipoproteins (41). Monocytes were stimulated with the mutant rIlpA at a various concentrations from 0.001 to 1.0 g/ml, and then assayed for production of TNF-␣ (Fig. 8A) and FIGURE 7. Effect of polymyxin B on cytokine production by rIlpA-treated monocytes. Polymyxin B was added to rIlpA at a concentration of 20 g/ml to neutralize residual LPS, which might be coeluted during the purification process for rIlpA. Monocytes (5 ϫ 10 5 ) were seeded in 24-well culture plates and exposed to rIlpA protein or to polymyxin B-treated rIlpA at a concentration of 1.0 g/ml. LPS of E. coli was also treated simultaneously with polymyxin B to provide a control for the neutralizing effect of polymyxin B against LPS. Following an 18-h exposure to the rIlpA protein or LPS, the culture supernatants were collected and then assayed for TNF-␣ (A) or IL-6 (B) by ELISA as described under "Experimental Procedures." Error bars represent the standard deviations from three independent experiments. Induction of each cytokine by IlpA, polymyxin B-treated rIlpA, LPS, or polymyxin B-treated LPS was performed in triplicate for each experiment. FIGURE 8. Role of the lipid moiety of rIlpA in cytokine production by monocytes. Confluent monocytic cells (1 ϫ 10 5/ ml) were exposed for 18 h to various concentrations of the wild type rIlpA protein or nonlipidated mutant rIlpA protein (rmIlpA) ranging from 0.01 to 1.0 g/ml. LPS of E. coli was used as a positive control for induction of cytokine production in monocytes. Culture supernatants were then collected and assayed for TNF-␣ (A) and IL-6 (B) by ELISA as described under "Experimental Procedures." Error bars represent the standard deviations from three independent experiments. Production of each cytokine by wild type rIlpA or rmIlpA at various concentrations was measured in triplicate for each experiment. IL-6 (Fig. 8B). As a control, another set of monocytes was treated with intact IlpA at the same concentrations as the mutant IlpA, and then assayed for cytokine production. When monocytes were treated with the mutant rIlpA at low concentrations (less than 0.1 g/ml), only basal levels of cytokines were detected. Formation of significant amounts of cytokines was observed in monocytes stimulated with relatively high concentrations of the mutant rIlpA (0.1 and 1.0 g/ml). On the other hand, considerable amounts of both TNF-␣ and IL-6 were formed in monocytes challenged with even the lowest concentration (0.001 g/ml) of the intact IlpA, and cytokine production increased in proportion with IlpA concentration as shown in Fig. 6. Monocytes exposed to 0.1 g/ml LPS were also examined for their induction of TNF-␣ and IL-6 production to monitor the proper experimental procedures, such as a stimulus treatment and cytokine assays. This result indicates that the nonlipidated mutant rIlpA was significantly impaired in its ability to induce cytokine production in monocytes compared with the ability of intact lipidated IlpA.
Role of TLR2 in Cytokine Production by IlpA-stimulated Monocytes-In this experiment, we investigated how the IlpA protein of V. vulnificus was recognized by human monocytes to result in cytokine production. In particular, we examined whether IlpA-induced cytokine production is mediated by a TLR family member, e.g. well known receptors for pathogenassociated molecular patterns, which relay information with respect to microbial components resulting in innate immunity. In Mycobacterium tuberculosis, antigenic lipoproteins were reported to be recognized by TLR2 (42). In addition, TLR2, which has been reported to function in inflammatory events associated with Lyme arthritis, recognizes lipoproteins of B. burgdorferi (43). Thus, we examined whether the IlpA of V. vulnificus also activates human monocytes via its interaction with TLR2.
To examine the possibility that TLR2 is involved in IlpAinduced cytokine production of monocytes, monocytes were pretreated with antibodies specific to TLR2 prior to being stimulated with rIlpA. As a control, another set of monocytes was pretreated with isotype control IgG instead of TLR2 antibodies, and then stimulated with rIlpA by the identical manner. These rIlpA-challenged monocytes were then assayed for production of TNF-␣ (Fig. 9A) and IL-6 ( Fig. 9B). As shown in the previous figures, both TNF-␣ and IL-6 were produced in monocytes challenged with rIlpA. Pretreatment of monocytes with the IgG control did not affect the formation of these two cytokines by rIlpA. In contrast, cells pretreated with TLR2 antibodies showed significant decrease in both TNF-␣ and IL-6 production compared with isotype IgG-treated monocytes (p Ͻ 0.02, Student's t test).
LPS is known to be recognized by another TLR, TLR4 (26,44). Thus, another set of monocytes was challenged with LPS FIGURE 9. Role of TLR2 in cytokine production by rIlpA-stimulated monocytes. A and B, human monocytes were preincubated for 1 h with either anti-TLR2 antibodies or with the relevant isotype IgG control. These cells were then stimulated with rIlpA (1.0 g/ml) for 18 h, and then culture supernatants were assayed for TNF-␣ (A) and IL-6 (B) by ELISA as described under "Experimental Procedures." A control set of monocytes pretreated with either the TLR2 antibodies or with the control IgG was stimulated with LPS instead of rIlpA and then assayed for cytokine production. C, HEK 293 cells (5 ϫ 10 5 ), which have been transfected with pFLAG-TLR2, p(IL6B) 3 50hu.IL6P-lucϩ, and pCH110, were stimulated with rIlpA (1.0 g/ml), incubated for an additional 20 h, and lysed to determine both luciferase and ␤-galactosidase activities. The NF-B activities were shown by dividing the luciferase activities (relative light units) by the ␤-galactosidase activities of the identical sample. As controls, plasmid pFLAG-CMV or pFLAG-TLR4 was included in the transfection experiments. Error bars represent the standard deviations from three independent experiments. Production of IL-6 or TNF-␣ was measured in triplicate for each sample. Asterisks indicate cytokine levels that were significantly different from that of untreated monocytes by the Student's t test. Data with a p value of Ͻ0.01 are indicated with **, and data with a p value between 0.01 and 0.05 are represented with *.
instead of rIlpA to determine whether the neutralizing activity of anti-TLR2 antibodies occurred only in rIlpA-treated cells but not in LPS-treated cells. Both the isotype IgG and anti-TLR2 antibodies were ineffective in inhibiting cytokine production in LPS-stimulated monocytes. This result suggests that the IlpA protein of V. vulnificus may act as an immunostimulant to monocytes by being recognized via TLR2.
To investigate the function of TLR2 in IlpA-induced cytokine production in human monocytes, we also reconstituted an in vitro system using HEK 293 cells, in which either TLR2 or TLR4 was expressed along with their common adaptor protein MyD88 (45). As a control, another set of HEK 293 cells was transfected with pFLAG-CMV1, the empty vector for pFLAG-TLR2 (a TLR2-expressing plasmid; see Ref. 35) or pFLAG-TLR4 (a TLR4-expressing plasmid; see Ref. 35). To monitor cytokine production using a reporter gene, this system includes a luciferase reporter for NF-B activities, p(IL6B) 3 50hu.IL6Plucϩ (36). For each transfection, the lacZ ϩ -plasmid pCH110 was included to normalize luciferase activities by dividing with the ␤-galactosidase activities of the same transfectant.
The ability of rIlpA to induce cytokine production was monitored by determining luciferase activities of the TLR2-expressing HEK 293 cells, the TLR4-expressing HEK 293 cells, and the control HEK 293 cells (Fig. 9C). In the absence of rIlpA, all of the HEK 293 cells showed basal levels of the luciferase activities. Upon stimulation with rIlpA, all three HEK 293 cells showed increased luciferase activities. Interestingly, the HEK 293 cells transfected with pFLAG-TLR2 demonstrated the most dramatic increase in their NF-B activities. On the contrary, the increase in luciferase activity was significantly less in the TLR4expressing cells and the control HEK 293 cells than the TLR2expressing HEK 293 cells.
Induction of TLR2 Transcription in Human PBMC and 293-hTLR Cell Lines by IlpA Treatment-To determine whether TLRs are expressed in the response to IlpA protein of V. vulnificus, we examined the levels of TLR2 and TLR4 mRNAs in the IlpA-treated cells by RT-PCR. When human PBMCs were challenged with rIlpA (1.0 g/ml) for 1 h, the level of TLR2 mRNA was significantly increased (Fig. 10A). On the other hand, LPStreated PBMC demonstrated increased levels of TLR4 mRNA at 5 h of post-stimulation (Fig. 10A). IlpA did not affect the level of TLR4 mRNA, and LPS did not influence the TLR2 mRNA in PBMC.
Two 293-hTLRs cell lines expressing either TLR2 or TLR4 were also examined for their response to IlpA by RT-PCR (Fig.  10B). Expression of TLR2 was increased in 293-hTLR2 cells at 1 h of exposure to rIlpA, whereas it was not affected by LPS stimulation at all. The amount of TLR4 mRNA was dramatically increased at 5 h after LPS stimulation in 293-hTLR4 cells. On the other hand, the level of TLR4 mRNA was only slightly increased by rIlpA treatment at 5 h in the 293-hTLR4 cells. The levels of GAPDH mRNA were examined as the controls for RNA amount.
Role of IlpA in Lethality of V. vulnificus to Mice-We examined the role of the IlpA protein in pathogenesis of V. vulnificus by using a mouse infection model. Upon intraperitoneal injection of various numbers of bacterial cells into mice, the numbers of dead mice were determined at 24 h after the injection ( Table 2). One of two experiments was presented as a representative experiment. Mice infected with the wild type showed an LD 50 of 9.5 ϫ 10 4 cells, whereas mice injected with the ⌬ilpA mutant showed an 8-fold higher LD 50 (7.5 ϫ 10 5 cells).

DISCUSSION
Human innate immune system recognizes microbial pathogens via action of its highly conserved pattern recognition receptors for microbial molecules, including LPS, peptidogly- FIGURE 10. Induction of Toll-like receptor mRNAs in human PBMCs, 293-hTLR2, and 293-hTLR4 cells. A, total RNA was prepared from PBMCs (2 ϫ 10 6 ) incubated with 1.0 g/ml rIlpA for 1 h or with 1.0 g/ml LPS for 5 h, and analyzed for their transcript levels of GAPDH, TLR2, and TLR4 using RT-PCR as described under "Experimental Procedures." B, levels of TLR2 mRNA and TLR4 mRNA were monitored by RT-PCR in 293-hTLR2 and 293-hTLR4 cells, which were pretreated with rIlpA or E. coli LPS. cDNA from a single RT reaction (derived from either 293-hTLR2 or 293-hTLR4) was used as a template for individual PCR with TLR2-specific primers or with TLR4 specific primers, respectively. Specific amplicons were visualized by 1.8% agarose gel electrophoresis and staining with ethidium bromide. GAPDH mRNA encoding glyceraldehyde-3-phosphate dehydrogenase was served as a loading control for RNA amount. cans, lipoteichoic acids, and lipoprotein. Lipoproteins are produced by both Gram-negative and Gram-positive bacteria, and some lipoproteins are known as the abundant proteins in certain Gram-negative bacteria (19). Lipoproteins of several microorganisms are known to activate a variety of host cells to produce cytokines. In Mycoplasma lacking a cell wall, surface lipoproteins such as VlpA and VlpC have been reported to induce cytokine production in macrophages (46). Several lipoproteins, LpqH, LpqG, and LprA, have been documented in M. tuberculosis to trigger innate inflammation (43,47,48). In addition, the signaling pathway involved in lipoprotein-induced activation of immune cells was found to be distinct from that involved by the LPS-induced process (49).
In this study, a lipoprotein of V. vulnificus was identified as a surface protein showing a potent antigenic property (Fig. 1). Therefore, we examined if this protein was able to stimulate the human cells to produce cytokine. Because the intact V. vulnificus exhibits too strong a cytotoxicity toward various types of human cells, 3 it is not applicable to measure cytokine production by directly treating PBMCs with intact V. vulnificus. Thus, bacterial lysates or recombinant proteins were used in this study. Indeed, wild type V. vulnificus lysate was effective to trigger cytokine production by human PBMCs, but V. vulnificus lysate lacking the isolated lipoprotein was significantly defective in cytokine induction (Fig. 5). Based on an observation that this lipoprotein of V. vulnificus functions as a major immunostimulant, we designated this protein as IlpA, which stands for immunogenic lipoprotein A. This is the first report on a lipoprotein of Vibrio spp. with immunostimulating activity. The lipoprotein from other Gram-negative bacteria, such as N. gonorrhoeae and Legionella pneumophila (50,51), have the ability to stimulate the production of proinflammatory mediators in mammalian cells.
TLRs, a group of well known receptors of the innate immune response, contain a cytoplasmic domain, which is homologous to the signaling domain of the IL-1 receptor. Each TLR senses the presence of the specific components of pathogens and transfers the information to a signaling pathway leading to the activation of NF-B. Activation of NF-B results in the transcription of genes coding for various cytokines (24). In the case of lipoproteins, TLR2 has been found as a main cellular receptor in several pathogenic bacteria. For example, a 19-kDa lipoprotein of Mycobacterium has been shown to trigger the signaling pathway of the host cells primarily through TLR-2, resulting in a production of IL-12/NO by monocytes/macrophages and an apoptosis of monocytes (42,52). In the case of V. vulnificus IlpA, it activates monocytes to produce TNF-␣ and IL-6 in a similar manner as LPS (Fig. 6). The receptor for Gram-negative bacterial LPS has been shown to be TLR4 (26). Therefore, the receptor for IlpA on monocytes was examined using two independent assays that clearly demonstrated the involvement of TLR2 in IlpA-mediated cytokine induction in monocytes. Cytokine production in rIlpA-stimulated monocytes is blocked by pretreatment with antibodies specific to human TLR2 (Fig.   9, A and B). In addition, V. vulnificus rIlpA induced activation of NF-B only in the HEK 293 cell line expressing TLR2 (Fig. 9C), whereas it did not induce NF-B activation in HEK 293 cells expressing TLR4.
Lipoproteins are present mainly in two forms, diacylated or triacylated (53). At present, biochemical properties of lipoproteins were extensively studied in Mycoplasma spp. (54). Mycoplasma gallisepticum and Mycoplasma mycoides have both diacylated and triacylated forms of lipoproteins (53). In TLR1-deficient mice, the macrophage-activating lipopeptide from Mycoplasma fermentans (MALP-2) containing two acyl chains failed to induce cytokine production, whereas N-palmitoyl-(S)-[2,3-bis(palmitoyloxy)-(2R,2S)-propyl] (Pam3)/Cys-Ser-Lys-4 (CSK4) (Pam 3 CSK 4 ) containing three acyl chains induced cytokine formation (55), suggesting that only triacylated lipoproteins are recognized by the combined action of TLR1 and TLR2. In vivo research using TLR6-knock-out mice indicates that diacylated lipoproteins are recognized by TLR6 in conjunction with TLR2 (56). Our data indicated that TLR2 is the main receptor responsible for V. vulnificus IlpA-mediated immune response. However, a role for other TLRs in this process is not yet known, and thus future investigation will be performed to unveil the biochemical characteristics of IlpA, and the roles of TLR1 and TLR6 in V. vulnificus IlpA-induced activation of human monocytes.
In many known lipoproteins having immunological activity, the portions of the proteins responsible for their immunological activity are located at the N-terminal triacylated lipopeptide region (42,57,58). The importance of lipidation in V. vulnificus IlpA was also confirmed by an observation that cytokine production in human monocytes triggered by nonlipidated rIlpA was significantly less than that by the intact lipidated rIlpA (Fig.  8). However, the nonlipidated form of V. vulnificus IlpA still retains an ability to induce cytokine production in monocytes at high concentrations. This finding raises the possibility that some domain(s) of V. vulnificus IlpA protein portion functions as a minor immunostimulant.
One of the distinct pathologies in V. vulnificus-causing septicemia is the clinical progression of the systemic inflammatory response (18). The potent immunostimulating activity of IlpA we observed suggests that this microbial factor may be responsible for the progressive pathogenesis of V. vulnificus. The importance of surface lipoprotein in systemic inflammation has been already reported in spirochaetal microorganisms lacking LPS (59). Increased levels of proinflammatory cytokines, such as TNF-␣, IL-1␤, and IL-6, in the serum from V. vulnificus-infected patients have been reported (60), which supports the speculated role for dysregulation of the cytokine response for the pathogenesis of V. vulnificus. Here we also provide evidence supporting the above clinical observation by showing an increased production of the proinflammatory cytokines, TNF-␣ and IL-6, in monocytes upon exposure to IlpA of V. vulnificus. Thus, V. vulnificus IlpA protein, which is capable of activating monocytes to produce cytokines, would be a potential molecule in the development of prevention and a therapy agent for V. vulnificus infection.