The Lip Lipoprotein from Neisseria gonorrhoeae Stimulates Cytokine Release and NF- (cid:1) B Activation in Epithelial Cells in a Toll-like Receptor 2-dependent Manner*

The human pathogen Neisseria gonorrhoeae produces an array of diseases ranging from urethritis to dissemi-nated gonococcal infections. Early events in the estab-lishment of infection involve interactions between N. gonorrhoeae and the mucosal epithelium, which leads to the local release of inflammatory mediators. Because of this, it is important to identify the bacterial virulence factors and host cell components that contribute to in-flammation. Using a series of column chromatography steps, we purified a lipoprotein from N. gonorrhoeae strain F62 called Lip. This outer membrane antigen expresses a conserved epitope known as H.8, which is common to all pathogenic Neisseria species. We found the purified preparation of Lip to be a potent inflammatory mediator capable of inducing the release of the chemokine interleukin (IL)-8 and the cytokine IL-6 by immortalized human endocervical epithelial cells and the production of IL-8 and the activation of the transcription factor NF- (cid:1) B by human embryonic kidney 293 (HEK) cells transfected with toll-like receptor (TLR) 2. Upon removal of Lip by immunoprecipitation, the ability of the lipopeptide derived fermentans the terminus. G-agarose

During the early stages of infection, the initial interaction of N. gonorrhoeae with mucosal epithelial cells triggers the release of inflammatory cytokines and chemokines including interleukin (IL)-6 1 and IL-8 (1)(2)(3), which serve to recruit and activate other immune cells to the site of infection. A variety of host cell receptors have been implicated in the recognition of and response to bacteria and their components. These include the toll-like receptor (TLR) family, which is central to innate immune defenses (4,5). The human TLR family consists of at least 10 distinct receptors and has been linked to the activation of NF-B (4, 5), a transcription factor that is involved in the expression of many proinflammatory cytokines, chemokines, costimulatory proteins, and adhesion molecules. The principal ligands that are recognized by the TLRs are conserved molecular patterns shared by a broad range of pathogens. For instance, TLR4 has been identified as the principle signal transducer in the recognition of LPS (5-7), whereas TLR2 confers responsiveness to bacterial lipoproteins/lipopeptides (8 -11), lipoarabinomannan (12), lipoteichoic acid (13), and peptidoglycan (13,14). It has been suggested that TLR2 can form functional interactions with other members of the TLR family, specifically TLR1 and TLR6 (15)(16)(17)(18), and that these interactions can distinguish between triacylated and diacylated lipidmodified lipoproteins/lipopeptides, respectively. For instance, peritoneal macrophages from TLR6-deficient mice do not produce significant levels of TNF␣, nitric oxide, and IL-12 in response to diacylated mycoplasmal macrophage-activating lipopeptide-2 (MALP-2) but maintain responsiveness to triacylated lipopeptides and lipoproteins (19). Moreover, macrophages derived from mice lacking TLR1 are hindered in the release of TNF␣ when treated with triacylated lipopeptides but respond normally to MALP-2, peptidoglycan, and LPS (17).
In light of reports that an LPS-deficient strain of N. meningitidis primarily activates macrophages through TLR2 (20,21) and that purified preparations of the Neisseria outer membrane porin protein signal through TLR2 (22), we examined other potential TLR2 ligands from the outer membrane of Neisseria. Many of the surface proteins of Neisseria are known to undergo significant antigenic variability; comparatively fewer are conserved within the pathogenic Neisseria species. One such conserved antigen is the Lip (or H.8 antigen) lipoprotein. The H.8 antigen is named after a mouse hybridoma anti-body that recognizes a surface epitope that is common to pathogenic Neisseria species including N. gonorrhoeae and Neisseria meningitidis but absent from other commensal Neisseria species (23). Although the size and amino acid sequence of Lip can differ between Neisseria species and strains (24), it primarily consists of 11-20 pentameric AAEAP amino acid repeats and a lipid-modified N-terminal cysteine residue and lacks any aromatic residues (25,26). This latter property makes it virtually undetectable by most protein assay and staining methods and thus makes it an elusive potential contaminant of preparations purified from outer membranes derived from Neisseria. In addition to recognizing Lip, the H.8specific monoclonal antibody is weakly cross-reactive to a second conserved protein from Neisseria known as Laz. This lipid-modified azurin-like protein contains imperfect AAEAP motifs at the N-terminal domain of the mature protein and a second domain with homology to the azurin protein of Pseudomonas aeruginosa (27)(28)(29).
In this report, we describe the proinflammatory activity of the Lip lipoprotein in terms of its ability to stimulate chemokine and cytokine production in human endocervical epithelial cells. In addition, we provide evidence using transfected human embryonic kidney (HEK) 293 cells suggesting that TLR2 and TLR1 are required for recognition of the Lip lipoprotein. Finally, preliminary biochemical analysis confirms the presence of lipid modification in the Lip lipoprotein and suggests that it may contain both short (C 8 -C 10 ) and medium (C 16 ) chain fatty acids.

MATERIALS AND METHODS
Reagents-The anti-H.8 murine monoclonal antibody (mAb 2C3) was obtained from Drs. Sunita Gulati and Peter Rice (Boston Medical Center, Boston, MA; Ref. 30). LPS from Escherichia coli strain K235 was purchased from List Pharmaceuticals (Woburn, MA) and repurified by phenol re-extraction to remove potentially contaminating proteins, as described previously (31,32). Lipooligosaccharide (LOS) was purified from N. gonorrhoeae strain F62 using a modification (33,34) of the hot phenol water extraction method (35). Recombinant human IL-1␤ and TNF␣ were obtained from Endogen (Woburn, MA). The lysate from Borrelia burgdorferi strain cN40 was a gift from Dr. Timothy Sellati (Albany Medical College, Albany, NY). Recombinant human soluble CD14 was obtained from Dr. Henri Lichenstein (Amgen, Thousand Oaks, CA). The synthetic lipopeptide Pam 3 Cys-Lip, the nonlipidated Lip peptide, the Pam 3 Cys-SK 4 lipopeptide, and the mycoplasmal lipopeptide MALP-2 were purchased from EMC Microcollections (Tuebingen, Germany). The Pam 3 Cys-Lip, which contains a tripalmitoyl-S-glyceryl cysteine at the N terminus, was based on the N-terminal sequence of the N. gonorrhoeae F62 Lip protein (sequence CGGEKAAE-APAAEAS). The MALP-2 lipopeptide was derived from the MALP-2 membrane lipoprotein from Mycoplasma fermentans (sequence CGNNDESNISFKEK) and possesses a dipalmitoyl-S-glyceryl cysteine at the N terminus. Protein G-agarose was from Sigma.
Bacterial Strains-The N. gonorrhoeae laboratory strain F62 (nonpiliated, nonopaque; pil Ϫ /opa Ϫ ), which was originally isolated from an individual with pelvic inflammatory disease, was used for these studies (36). Strain FA1090 and the H.8 deletion mutant FA1090⌬H.8 and the plasmid pHSS6 (⌬H.8) containing the FA1090 H.8 gene insertionally inactivated with a chloramphenicol resistance gene were gifts from Dr. Janne Cannon (University of North Carolina, Chapel Hill, NC) (26). Piliated N. gonorrhoeae strain F62 was transformed with purified plasmid pHSS6 as previously described (37,38). Chloramphenicol-resistant N. gonorrhoeae F62 colonies were selected on standard gonococcal culture medium supplemented with chloramphenicol at a concentration of 1 g/ml. Transformants were confirmed by immunoblot analysis showing the absence of the 2C3 epitope as described below.
Nucleotide Sequencing of Lip from N. gonorrhoeae-N. gonorrhoeae strain F62 was grown overnight on chocolate agar. Crude DNA was isolated by resuspending approximately five colonies in 100 l of sterile water, as described by Hobbs et al. (39). The Lip gene was amplified by polymerase chain reaction using Taq DNA polymerase (Promega) and chromosomal DNA from N. gonorrhoeae strain F62. The primers, derived from Trees et al. (24) that were used to amplify the Lip gene were CAAATTCAGCGATGAATTTCCAACCC-3Ј (forward primer) and 5Ј-TATGAAGGTCAGGCATGTTTGTCGG-3Ј (reverse primer). The PCR conditions consisted of 25 cycles of denaturation at 94°C for 30 s, annealing at 52°C for 30 s, and elongation at 68°C for 2 min 30 s. The amplified PCR product of ϳ500 base pairs was gel-purified and sequenced by the Boston University Molecular Genetics Core Facility using the ABI Prism Big DYE Terminator Cycle Sequencing Ready reaction kit (Applied Biosystems) performed on the ABI Prism 377-96 DNA sequencer.
Outer Membrane Preparations-N. gonorrhoeae strain F62 was grown overnight on chocolate agar, and the outer membranes were prepared as described previously (33). Briefly, bacteria were harvested into shearing buffer (PBS containing 10 mM EDTA, pH 7.4). The suspension was incubated at 60°C for 30 min, and the outer membranes were sheared by passage through a 25-gauge needle, followed by homogenization for 10 s in a Waring blender. After two rounds of centrifugation at 12,000 ϫ g for 20 min to remove unlysed bacteria and large cellular debris, the resulting supernatant was centrifuged at 80,000 ϫ g for 2 h. The pellet from this ultracentrifugation step was resuspended in deoxycholate buffer (50 mM glycine, 1 mM EDTA, and 1.5% deoxycholate, pH 9.5). Uniform solubilization of the membranes was achieved by increasing the pH of the suspension to 13.5 with 1 M NaOH, followed by immediate readjustment back to pH 9.5 with 0.5 N HCl.
Purification of the H.8/Lip Lipoprotein from N. gonorrhoeae-The procedure for purifying H.8/Lip from Neisseria was described by Zollinger et al. (40). Briefly, the solubilized outer membrane preparation was loaded onto a Sephadex G-100 column (2.5 cm ϫ 60 cm) equilibrated with deoxycholate buffer. Fractions (5 ml) were collected at a flow rate of 0.25 ml/min. The presence of H.8 was confirmed by dot-blot analysis using the anti-H.8 mouse monoclonal IgG antibody 2C3, and the presence of other proteins and LOS was monitored by silver staining of SDS-PAGE gels and/or BCA protein assay reagent A (Pierce). The fractions that were positive for H.8 and contained a low level of other proteins were pooled and dialyzed for 18 h against PBS (NaPO 4 , pH 7.4, and 50 mM NaCl). This fraction was further clarified by ion exchange chromatography on a DEAE-Sephacel CL-4B (Amersham Biosciences) column (1.5 ϫ 12 cm; 20 ml) that was equilibrated with PBS. The bound material was eluted with a 50 -400 mM NaCl gradient in PBS at a flow rate of 0.5 ml/min, and the fractions were collected every 4 min.
Fatty Acid Analysis-The purified H.8/Lip fraction, the synthetic Pam 3 Cys-Lip lipopeptide, and gonococcal LOS were hydrolyzed in 6 M HCl for 20 h at 95°C. After extraction of the hydrolyzed products with isopropanol/hexane and 1 N H 2 SO 4 (40:10:1), the hexane phase containing total lipid was concentrated by evaporation of the organic solvent under nitrogen gas, and the resulting product was redissolved in 100 l of hexane and separated by TLC (hexane/ethyl ether/acetic acid, 70:30: 1). The fraction that contained free fatty acids was scraped off and eluted with hexane. After drying down under N 2 gas, methylation was performed by incubation in BF3-methanol solution (14%, v/v, BF3 in methanol) at 60°C for 30 min. The fatty acid methyl ester was extracted into a hexane solution. The hexane solution was dried with anhydrous sodium sulfate and used directly for gas/liquid chromatography (GLC) analysis without further condensation. GLC analysis was performed on a Shimuzu 14A gas chromatograph with a Supelco SP™-2380 capillary column, with an initial oven temperature of 150°C, a final temperature of 240°C, a heating rate of 4°C/min, an injector temperature of 220°C, and a detector temperature of 240°C. The carrier gas (helium) was at 50 kPa, make-up carrier gas (helium) was at 100 kPa, hydrogen gas was at 55 kPa, and compressed air was at 50 kPa. The sample was injected in 1-1.5 l with splitting rate of 1:25. The peak assignment was done using Nu-Chek GLC reference for free fatty acids (063A) and bacterial acid methyl ester reference standards (catalog number 47080-U) from Supelco (Bellefonte, PA). The results are presented as the percentages of each fatty acyl chain in the sample.
Assays for IL-6 and IL-8 Secretion-End/E6E7, HEK 293, and HEK-TLR2 cell were grown in 96-well tissue culture dishes at a density of 5 ϫ 10 4 /well and allowed to grow for 24 h. The cells were treated with various concentrations of purified DEAE H.8/Lip fraction in a reaction volume of 100 l. Control treatments were PBS or DEAE column elution buffer, IL-1␤ (5 ng/ml), TNF␣ (5 ng/ml), and LPS (100 ng/ml). Treatments of the End/E6E7 cells were in keratinocyte serum-free growth medium. Soluble CD14, at a concentration of 500 ng/ml, was added to the medium when indicated (41). Treatments for the HEK 293 cells were in medium supplemented with 10% fetal bovine serum. After incubation for 20 h at 37°C, the culture supernatants were removed, diluted to the appropriate concentrations in keratinocyte serum-free medium or Dulbecco's modified Eagle's medium, and assayed for IL-6 or IL-8 using Eli-pair enzyme-linked immunosorbent assay kits from Cell Sciences (Norwood, MA). All of the cytokines were assayed in duplicate or triplicate, and the data are reported as the means Ϯ the standard error or standard deviation, respectively.
Expression Plasmids-The ELAM-1-luciferase reporter plasmid (pELAM-luc) was generated as described (5) by cloning a fragment (Ϫ241 to Ϫ54 base pairs) of the human E-selectin promoter into the pGL3 reporter plasmid (Promega). This expression plasmid reports luciferase in an NF-B-dependent manner. Expression plasmids for human TLR1 (in the pFLAG-CMV vector), TLR2 (in the pFLAG-CMV vector), and TLR6 (in the pEF-BOS vector) were a gift of Douglas Golenbock (University of Massachusetts Medical School, Worcester, MA). All of the plasmid DNA was isolated with Endo-Free TM Maxi-prep columns from Qiagen.
Transient Transfection and NF-B Reporter Luciferase Assay-HEK 293 or HEK-TLR2 cells were plated in 6-well tissue culture plates (4 ϫ 10 5 cells/well) and maintained in Dulbecco's modified Eagle's medium plus 10% fetal bovine serum for 18 h. Using the SuperFect™ Transfection Protocol (Qiagen), the cells were transfected with 250 ng of pELAM-luc reporter construct and one of the following expression vectors (described above): 100 ng of human TLR2, 500 ng of human TLR1, or 10 ng of human TLR6. The empty vector pcDNA3 (Invitrogen) was used as a control and to normalize the DNA concentration for all of the transfection reactions. The transfected cells were then allowed to recover for a period of 20 h. After triplicate treatments for 5 h, the cells were harvested in lysis buffer (Promega), snap frozen on dry ice, and assayed for luciferase activity by the use of a commercial luciferase assay kit (Promega) according to the manufacturer's protocol. The amount of luciferase activity in each sample was quantified by a Monolight TM 3010 Luminometer (PharMingen, San Diego, CA). The data are reported as the means Ϯ the standard deviation. All of the transfection experiments were repeated at least twice.
Immunoprecipitation of H.8 -Immunoprecipitations were carried out for 2 h at 4°C in 70-l reaction volumes containing 5 l of purified H.8/Lip, 15 g of protein G-agarose, 4 g of bovine serum albumin, and the indicated concentrations of either anti-H.8 2C3 antibody or control IgG (0 -10 g/ml). After centrifugation, the IP supernatants were decanted, and the resulting IP pellets were washed three times with PBS. The proteins from both the pellet and the supernatant were separated by SDS-PAGE and transferred onto an Immobilon-P polyvinylidene difluoride membrane (Millipore, Bedford, MA). H.8 antigen was detected using mAb 2C3 and visualized by alkaline phosphatase-conjugated goat-anti-mouse IgG (Sigma) and substrate as described previously (42).
Gel Purification of Lip-H.8/Lip derived from the DEAE-Sephacel column was further purified by elution after electrophoresis on a 4 -12% Bis-Tris gel under denaturing conditions in the presence of lithium dodecyl sulfate sample buffer (Invitrogen). Using the prestained molecular mass samples as a guide, the gel fragment that corresponded to the predicted migration of Lip was excised and incubated with 1 ml of endotoxin-free H 2 O for 18 h at 37°C. The eluted material was analyzed by silver staining to assess for contaminants, and the presence of Lip was confirmed by immunoblotting with mAb 2C3 as described above. As a negative control, a gel fragment spanning the 60 -70-kDa regions was also excised and treated similarly. Both samples were tested for contaminating endotoxin using the Limulus Amebocyte Lysate Pyrochrome kit from Associates of Cape Cod (Falmouth, MA). The limit of detection for the Pyrochrome kit is 0.005 endotoxin units/ml (12 endotoxin units ϭ ϳ1 ng of endotoxin). Endotoxin contamination in the concentrated H.8/Lip and gel control preparations was 6 and 5 endotoxin units/ml, respectively; preparations were used in the biological assays at dilutions of 1:50 to 1:1000.

RESULTS
Purification of the H.8 Lipoprotein Lip from N. gonorrhoeae F62 Outer Membranes-It has been documented that small amounts of contaminating lipoproteins in LPS preparations can lead to TLR2-dependent cell activation (32). We tested an LOS preparation from N. gonorrhoeae for activation of NF-B in human endocervical epithelial cells. This cell line had previously been shown by our laboratory to be deficient in TLR4 and MD-2, the receptors required for responses to most species of endotoxin (41). Rather unexpectedly, we found dose-dependent cell activation with LOS concentrations greater than 1 g/ml (data not shown). However, upon repurification of the crude LOS preparation by phenol re-extraction (32), the LOSdependent activation disappeared. This suggested that cell activation was being driven by a phenol-soluble lipoprotein. Immunoblot analysis using mAb 2C3 revealed that the gonococcal outer membrane protein Lip (or H.8 antigen) was present in the original LOS preparation but not after re-extraction (data not shown). This observation prompted us to further examine the potential of the Lip lipoprotein to act as a ligand for TLR2mediated signaling.
We first cloned the gene encoding Lip from N. gonorrhoeae strain F62 by PCR and sequenced it (GenBank TM accession number AY363393). The predicted amino acid sequence was determined from the Lip DNA sequence and aligned with that from two other strains of N. gonorrhoeae, FA1090 and R10, and with N. meningitidis strain MC58. As previously reported in these other Neisseria strains, the N-terminal region contained the typical N-terminal 17-residue signal peptide for lipoproteins, along with a series of pentapeptide AAEAP repeats (14 repeats in strain F62) (27,28).
To purify Lip from F62, outer membranes were solubilized in deoxycholate buffer and fractionated on a Sephadex G-100 size exclusion column. The fractions containing H.8/Lip were detected by immunoreactivity with mAb 2C3. The H.8-positive fractions were pooled, dialyzed in PBS, and loaded on a DEAE-Sephacel ion exchange column. The bound material was eluted with a 50 -400 mM NaCl gradient in PBS. As Lip consists almost entirely of AAEAP amino acid repeats and lacks aromatic residues, the amount of protein in the final DEAE H.8/ Lip fraction could not be determined by conventional protein detection assays. As Fig. 1A demonstrates, immunoblotting with mAb 2C3 revealed the presence of a protein with an apparent molecular mass of ϳ18 kDa in each step of the purification, which is consistent with the H.8/Lip previously reported in other strains of Neisseria (23,28,43). This band was absent from the F62 H.8 deletion mutant (data not shown). The other faintly 2C3-immunoreactive bands evident in the total F62 outer membranes were excluded from the final DEAE pool. In addition, the final DEAE fraction containing H.8/Lip did not show any discernable protein or LOS by silver-stained SDS-PAGE gel analysis.
We next wanted to demonstrate that the purified H.8/Lip fraction contained fatty acids. The purified H.8/Lip, a synthetic Pam 3 Cys-Lip lipopeptide, and phenol re-extracted gonococcal LOS from strain F62 were acid-hydrolyzed, and the lipid fractions were analyzed by TLC. The synthetic Lip lipopeptide and gonococcal LOS purified from strain F62 served as positive controls for fatty acid detection. The migration pattern on TLC revealed the presence of free fatty acids in the H.8/Lip sample as well as the positive controls (data not shown). To further analyze the free fatty acids, the methyl ester derivatives of the free fatty acids of H.8/Lip were subjected to gas chromatography. The total H.8/Lip preparation and synthetic Pam 3 Cys-Lip lipopeptide were run simultaneously for comparison. Fatty acid structures were determined by comparing the retention signals to known standards. Unlike the synthetic compound, there was no single, dominant peak seen in the two H.8/Lip preparations. However, there was the suggestion of two dominant structures: a medium chain fatty acid with a length of C 16 and a shorter chain fatty acid with a length of C 8 -C 10 (Fig. 1B). The C 16 component would be consistent with the typical Pam 3 Cys lipoprotein described for most bacterial lipoproteins, whereas the C 8 -C 10 component has not been documented in other lipoprotein structures. It is possible that the additional signals noted reflect side reactions during the harsh acid treatment, because they were also noted in the synthetic Pam 3 Cys-Lip.
The Purified Preparation of H.8/Lip Stimulates IL-6 and IL-8 Release from Human Endocervical Epithelial Cells-We wanted to investigate the influence of the purified preparation of H.8/Lip on interleukin release from a relevant human cell line. Our laboratory recently established an in vitro model to study epithelial-microbial interactions and the downstream effects of cell signaling in the female urogenital tract using immortalized endocervical epithelial cells (41). Our previous investigations showed that these cells respond to whole N. gonorrhoeae as well as gonococcal lysates by up-regulation of cytokines and adhesion molecules. Although this cell line was unresponsive to bacterial endotoxin by virtue of the fact that it lacked TLR4 and MD-2, it was sensitive to stimulation with known TLR2 ligands, including MALP-2 and the Pam 3 Cys-SK 4 lipopeptide.
To investigate the pro-inflammatory activity of H.8/Lip, we treated End/E6E7 cells with dilutions of the H.8/Lip DEAE fraction ranging from 1:10,000 to 1:10 and measured the release of the cytokine IL-6 and the chemokine IL-8. The cells were also treated with medium supplemented with DEAE column elution buffer, LPS (1 g/ml), lysate from B. burgdorferi (1 g/ml), TNF␣ (5 ng/ml), and IL-1␤ (5 ng/ml). The buffer and LPS served as negative controls for IL-6/IL-8 secretion, whereas the B. burgdorferi lysate (a known TLR2 ligand), TNF␣, and IL-1␤ were positive controls. As shown in Fig. 2, the secretion of IL-6 ( Fig. 2A) and IL-8 ( Fig. 2B) was enhanced by H.8/Lip at 1:10 and 1:100 dilutions, with levels comparable with that seen with the TNF␣ and IL-1␤ controls. In contrast, no cytokines were secreted above basal levels by the End/E6E7 cells with LPS treatment. Thus, any trace quantities of neisserial LOS or environmentally derived endotoxins potentially contaminating our H.8/Lip preparation at a level of detection below that of silver staining would likely have a negligible impact on this LPS-unresponsive cell line. We saw no significant difference in IL-6/IL-8 production when cells were incubated with the H.8 deletion mutant F62 or the parental strain (data not shown). The most likely explanation for this observation is that although the H.8 deletion mutant is lacking H.8, it still expresses other inflammatory ligands. This would include porin, peptidoglycan, and a variety of lipoproteins such as the H.8-related lipoprotein Laz. In addition, when we examined whole cell lysates of the H.8 deletion mutant and the parental isolate from both strains F62 and FA1090 by silver staining, we found there were differences in the expression pattern of certain dominant outer membrane proteins, suggesting that they were not truly isogenic mutants (data not shown).
Transfection of TLR2 Enhances Responsiveness of HEK 293 Cells to Purified H.8/Lip-The observation that the immortalized endocervical cell line expresses mRNA for TLR2 but not for the LPS signal transducing molecules TLR4 and MD2 (41) prompted us to evaluate the role of TLR2 in mediating chemokine secretion induced by the purified preparation of H.8/Lip. HEK 293 cells, which do not express endogenous TLR2, were transiently transfected with human TLR2 or control vector DNA; treated with 10-fold dilutions of the purified H.8/Lip preparation ranging from 1:10 to 1:1000, control medium, LPS (negative control), or B. burgdorferi lysate (positive control) for 24 h; and then assessed for secretion of IL-8 into the culture medium. We found that in cells transfected with TLR2, purified H.8/Lip stimulated IL-8 secretion in a dose-dependent manner (Fig. 3A). In contrast, no change in the basal IL-8 secretion in control vector-transfected cells that were treated with H.8/Lip was detected.
Because the transcription factor NF-B has been implicated in the expression of numerous cytokines and chemokines, we investigated the role of TLR2 in mediating NF-B activation in HEK 293 cells treated with purified H.8/Lip. HEK 293 cells were transiently cotransfected with the NF-B-dependent pELAM-luc reporter construct and the TLR2-FLAG plasmid or the control vector pcDNA3. Upon exposure of the HEK-TLR2 For the silver stain gel and immunoblot, the amount of measurable protein loaded from the F62 outer membrane and G-100 H.8 fractions were 3.9 and 0.03 g, respectively. B, total H.8/Lip (H8), H.8 Lip free fatty acids (H8-FA), and synthetic Pam 3 Cys-Lip (Pam) were subjected to GLC as described under "Material and Methods." Fatty acid structures were determined by comparing the retention signals to known standards. The peaks that correspond to the reference fatty acid methyl ester standards were graphed as the percentages of the total derivatized fatty acids in each preparation. The asterisks represent the level of certainty of the identity of each peak. *, confirmed; **, speculation with confidence. iC15, methyl 13-methyltetradeconoate; aC15, methyl 12-methyltetradeconoate; iC17, methyl 15-methylhexadecanoate. cells to H.8/Lip at a dilution of 1:10,000, NF-B-mediated luciferase expression was detected, and this activation was enhanced in a dose-dependent manner (Fig. 3B). In contrast, control cells that were transfected with pcDNA3 did not show any NF-B-dependent luciferase activity even with the highest concentration of the purified H.8/Lip fraction used (1:10). NF-B was activated in HEK-TLR2 cells that were treated with the synthetic triacylated lipopeptides Pam 3 Cys-Lip and Pam 3 Cys-SK 4 (Fig. 3C), which contain a tripalmitoyl-S-glyceryl cysteine at their N termini. As a control, the synthetic, nonlipidated Lip peptide was not able to stimulate NF-B activation in these cells. These data provide evidence that H.8/Lip triggers inflammatory mediator production from epithelial cells and can activate downstream signaling events from TLR2.
Immunoprecipitation Blocks the Ability of the Purified H.8/ Lip to Stimulate NF-B Activation-We considered it important to rule out the contribution of other undetected proteins in the H.8/Lip preparation that had similar proinflammatory activity or acted in concert with H.8/Lip. To demonstrate this, H.8/Lip was depleted by IP using the anti-H.8 antibody 2C3. Immunoprecipitation of the final H.8/Lip DEAE fraction with between 3 and 10 g/ml of 2C3 antibody resulted in the removal of Lip from the supernatant with the concomitant appearance of H.8 in the IP pellet (Fig. 4A). Incubation of the H.8 DEAE fraction with control mouse IgG, however, did not result in any removal of Lip from the IP supernatant (data not shown). To measure the ability of the IP supernatants to simulate NF-B, HEK 293 cells were transfected with TLR2-FLAG and the pELAM-luciferase reporter plasmid and treated for 5 h with the IP supernatants. Immunoprecipitation of H.8/Lip with the control IgG antibody did not result in the removal of NF-B activation (Fig. 4B). On the other hand, immunoprecipitation of the H.8/Lip fraction with 3 or 10 g/ml of the anti-H.8 2C3 antibody resulted in the loss of NF-B activation in the IP supernatant (Fig. 4B), which correlates with depleting the sample of Lip (Fig. 4A). This suggests that Lip, or a factor tightly associated with Lip, is responsible for the induction of NF-B activation.
Gel-purified Lip Stimulates NF-B Activation and IL-8 Secretion-Having shown that immunoprecipitation of Lip from the purified H.8/Lip preparation resulted in the complete loss activity, we wanted to confirm that Lip alone, and not a molecule tightly associated with Lip, could stimulate the activation of epithelial cells. To remove any noncovalently associated molecules from Lip, we subjected the purified H.8/Lip preparation to SDS-PAGE under denaturing conditions, excised the region of the gel containing the 18-kDa Lip protein, and eluted the protein into endotoxin-free H 2 O by passive diffusion. As a negative control, a strip of gel between 60 and 70 kDa was similarly treated. These gel-purified samples were then used to treat HEK-TLR2 cells that were transiently transfected with the pELAM-luciferase reporter plasmid. Whereas the control gel sample only weakly stimulated NF-B activation in these cells, the gel-excised Lip protein activated NF-B to a much greater extent in a dose-dependent fashion (Fig. 5A). Similarly, the gel-purified Lip at a dilution of 1:1000 stimulated IL-8 production from the End/E6E7 endocervical cells, whereas the control gel slice elution did not stimulate IL-8 secretion above base-line levels (Fig. 5B). These data demonstrate that the Lip protein is sufficient to stimulate NF-B activation and IL-8 secretion in epithelial cells.
The TLR2-dependent Activation of NF-B by Purified H.8/ Lip in HEK-TLR2 Cells Is Enhanced by Transfection of TLR1 but Not TLR6 -The recognition of di-and triacylated lipopeptides by the TLR complex is postulated to involve the cooperation of TLR2 with other members of the TLR family such as TLR1 and TLR6 (15)(16)(17)(18). Expression of TLR2 with TLR6 appears to be selective for cell signaling induced by primarily diacylated lipopeptides, whereas coexpression of TLR2 with TLR1 permits signaling by both di-and triacylated lipopeptides. To determine the effect of other lipoprotein-responsive TLRs on Lip-mediated signaling, we cotransfected TLR1 or TLR6 with the NF-B-dependent pELAM-luc reporter plasmid in HEK-TLR2 cells and stimulated them with increasing concentrations of purified H.8/Lip (Fig. 6A) or 1 ng/ml of the synthetic diacylated lipopeptide MALP-2 (Fig. 6B). The levels of activated NF-B attributed to TLR2 alone (control vector transfections) increased in response to the purified H.8/Lip and MALP-2 lipopeptide. Transfection of HEK-TLR2 cells with TLR1 greatly enhanced NF-B activation upon treatment with the purified H.8/Lip, whereas transfection with TLR6 did not augment activation above levels seen with cells transfected with empty control vector (Fig. 6A). In contrast, MALP-2mediated activation of NF-B was only slightly enhanced by transfection of TLR1, but unlike H.8/Lip, expression of TLR6 increased NF-B activation in response to MALP-2 (Fig. 6B). Collectively, these data suggest that Lip signals through the association of TLR2 with TLR1 but not TLR6, suggesting that the Lip lipopeptide shares characteristics with triacylated lipoproteins.

DISCUSSION
The family of Toll-like receptors plays a key role in the activation of the innate immune response by recognition of pathogen-associated microbial patterns (44). Many membrane products that are constituents of or are associated with the bacterial outer membrane and cell wall have been demonstrated to activate cells via a TLR-dependent mechanism. We have purified a lipoprotein called Lip from the outer membrane of N. gonorrhoeae and demonstrated its ability to stimulate the production of proinflammatory mediators from epithelial cells in a TLR2-dependent manner. The DNA sequence of the Lip gene from F62, similar to that seen in other strains of N. gonorrhoeae and N. meningitidis (26,45), reveals that Lip from this strain contains a lipoprotein leader sequence and a N-terminal cysteine. It is the lipid modification of this cysteine residue that is believed to account for the anomalous migration pattern of H.8/Lip seen on SDS-PAGE (46,47).
Although the outer membranes of Gram-negative bacteria contain a variety of lipoproteins, very few have been characterized biochemically. Lipoproteins found in most bacteria and spirochetes appear to be triacylated at the N terminus. The classic triacylated bacterial lipoprotein structure is the Braun murein lipoprotein of E. coli (48). In contrast, the lipoproteins found in mycoplasmas are diacylated and therefore have a free amino group (49). Although the nature of the lipid modification of Lip is not precisely known, our preliminary analysis by GLC suggests that the associated lipid component of purified Lip may be made up of at least two elements: the classic C 16 fatty acid, which has not been detected in the prior H.8/Lip analyses, and a shorter C 8 -C 10 lipid, which has been previously demonstrated as a component of neisserial H.8/Lip (46,47). Evidence to support that Lip can be modified with a C 16  It has been suggested that TLR2 cooperates with TLR1 and/or TLR6 to allow responsiveness to triacylated and diacylated lipopeptides, respectively (17). Because transfection of TLR6 failed to enhance NF-B activation in the HEK-TLR2 cells while TLR1 transfection enhanced responsiveness, we believe that the N-terminal lipid modification of Lip is most The purified H.8/Lip preparation was clarified by SDS-PAGE, and the Lip lipoprotein was excised from the gel and eluted into endotoxin-fee H 2 O. A control gel slice (in the range of 60 -70 kDa) was also treated in a similar manner and served as a negative control. A, dilutions of these samples were then used to treat HEK-TLR2 cells that were transiently transfected with 250 ng of the pELAM-luciferase construct to measure NF-B activation, represented as relative light units (RLU) from triplicate treatments as described in Fig. 3. B, End/E6E7 endocervical epithelial cells were also incubated overnight with the gel-purified Lip and control gel eluate (1:100 dilutions), and the supernatants were then assayed for the presence of IL-8 by enzyme-linked immunosorbent assay. The values represent the means Ϯ S.D. of triplicate readings of one representative experiment. These results are representative of at least two independent experiments.
FIG. 6. Stimulation of NF-B in HEK-TLR2 cells by purified H.8/Lip is enhanced by transfection of TLR1 but not TLR6. HEK 293 cells that were stably transfected with TLR2 (HEK-TLR2) were transiently transfected with 250 ng of the pELAM-luc reporter construct and one of the following: 10 ng of human TLR6 in pEF-BOS, 500 ng of human TLR1 in pFLAG-CMV, or 500 ng of empty control vector pCDNA3. In all transfections, pcDNA3 was used to normalize the DNA concentration. The transfected cells were then treated for 5 h at 37°C with the indicated concentrations of purified H.8/Lip (A) or with the synthetic diacylated lipopeptide MALP-2 (B). Luciferase activity was measured as described in the legend to Fig. 3. The results are representative of at least two independent experiments. likely triacylated. We tested only one synthetic lipopeptide in our assays composed of the 15 N-terminal amino acids of Lip and triacylated at the N-terminal cysteine with palmitic acid. This compound was a potent activator of NF-B, similar to the Pam 3 Cys-SK 4 lipopeptide commonly used in many laboratories. Although other groups have reported that synthetic lipopeptides modified with C 12 and C 14 fatty acids have biological effects in vitro (17), the possible activity of a much shorter C 8 -C 10 lipopeptide similar to what appears to be present in our H.8/Lip product is unclear and warrants further investigation. Data from our laboratory suggest that neisserial ligands other than LOS can stimulate the release of proinflammatory cytokines from cervical epithelial cells, where the machinery for responses to endotoxin are absent, thereby contributing to disease pathogenesis (41). One activity of H.8/Lip that we report here is its ability to activate epithelial cells and to trigger the release of cytokines and chemokines such as IL-6 and IL-8 in a TLR2-dependent manner. These soluble factors would then serve to amplify the immune response by the recruitment of other immune effector cells, such as polymorphonuclear cells and macrophages, to the site of infection. The infiltrating neutrophils and macrophages subsequently release soluble mediators, such as TNF␣, that can be beneficial for fighting off an infection. However, when produced in large amounts, these mediators can also have consequences that lead to epithelial cell death. This damage to the protective cell layer could provide a means for the infecting organisms to gain access to the subepithelial compartment and allow for bacterial replication and dissemination. Unfortunately, the lack of an adequate animal model makes this a difficult hypothesis to investigate in vivo. In addition, we have found that inactivation of Lip expression in two N. gonorrhoeae strains tested leads to changes in the expression of other outer membrane proteins. Thus, a true isogenic mutant of Lip is not available.
Other potential roles for Lip in pathogenesis may result from its unique amino acid sequence, where the evenly spaced glutamic acid residues in the repetitive consensus AAEAP sequence on the surface of the molecule would produce a highly negatively charged protein in the outer membrane of the bacteria. The highly negatively charged surface of Lip could play a role in the interactions of the pathogen with host cells and extracellular matrix or act as a defense mechanism by binding to cationic bactericidal molecules. Although these latter functions are purely speculative, the unique amino acid composition of H.8/Lip has one indisputable consequence: it makes the antigen virtually undetectable by traditional protein detection methods, such as Coomassie Blue and silver staining, Bradford and Lowry colorimetric protein assays, and UV absorbance. Given that Lip is an abundant lipoprotein in gonococcal and meningococcal membranes, that Lip copurifies with LOS and presumably other lipid/lipoprotein preparations, and that purified Lip can activate NF-B and trigger cells to release cytokines, it should be considered a potential contaminant in purified components derived from the outer membrane of N. gonorrhoeae and N. meningitidis.
Although the role of H.8/Lip in the pathogenesis of Neisseria infections has not been fully elucidated, sera from patients with gonococcal infection contain antibodies that react with the H.8 antigen (52). Furthermore, the distribution of the H.8 antigen in primarily pathogenic species of Neisseria hints at a possible role in infection, survival, and/or disease progression. It is very likely that the host response to this complex pathogen is driven by a variety of virulence factors that activate multiple host receptors simultaneously. Based on our findings, we believe that H.8/Lip should be considered one of immunologically relevant molecules in the pathogenesis of neisserial infections.