Toll-like receptor 2-mediated interleukin-8 expression in gingival epithelial cells by the Tannerella forsythia leucine-rich repeat protein BspA

Abstract

Tannerella forsythia is a gram-negative anaerobe strongly associated with chronic human periodontitis. This bacterium expresses a cell surface-associated and secreted protein, designated BspA, which has been recognized as an important virulence factor. The BspA protein belongs to the leucine-rich repeat (LRR) and bacterial immunoglobulin-like protein families. BspA is, moreover, a multifunctional protein which interacts with a variety of host cells, including monocytes which appear to respond to BspA through Toll-like receptor (TLR) signaling. Since gingival epithelium forms a barrier against periodontal pathogens, this study was undertaken to determine if gingival epithelial cells respond to BspA challenge and if TLRs play any role in BspA recognition. This study was also directed towards identifying the BspA domains responsible for cellular activation. We provide direct evidence for BspA binding to TLR2 and demonstrate that the release of the chemokine interleukin-8 from human gingival epithelial cells by BspA is TLR2 dependent. Furthermore, the LRR domain of BspA is involved in activation of TLR2, while TLR1 serves as a signaling partner. Thus, our findings suggest that BspA is an important modulator of host innate immune responses through activation of TLR2 in cooperation with TLR1.

FIG. 1.

(A) Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of rBspA and its derivatives. Lane stnd, protein marker; lane 1, rBspA; lane 2, rBspLRR-1-2; lane 3, rBspLRR-1; lane 4, rBspΔLRR; lane 5, His-GST. The numbers on the left indicate the molecular masses of protein standards. (B) Schematic diagrams of BspA and its truncated derivatives expressed as recombinant proteins. The amino acid residues for each derivative are indicated in parentheses. Abbreviations: D1, LRR domain D1 containing 14 tandem LRRs (residues 91 to 412); D2, LRR domain D2 containing six tandem LRRs; Big_2, bacterial Ig-like domains.

FIG. 2.

IL-8 induction in epithelial cells challenged with the BspA protein. HGECs were treated with rBspA (5 μg/ml) or the TLR2 agonist Pam₃Cys (Pam3) (0.1 μg/ml) for 6 h. Induction of IL-8 in culture supernatants was determined by an ELISA. The data are the means ± standard deviations of triplicate determinations in one of three independent sets of experiments that yielded similar findings. Statistically significant (P < 0.05) blocking of cytokine release by TLR2 antibody (Ab) is indicated by an asterisk.

FIG. 3.

(A) TLR2 and TLR1 recognize BspA. HEK293 cells were transiently transfected with plasmids expressing the indicated human TLRs or an empty vector (EVC). Cells were incubated for 6 h with zymosan (10 μg/ml), BspA (5 μg/ml), highly purified E. coli LPS (Ec LPS) (1 μg/ml), or no agonist. NF-κB activity in supernatants was quantified by the luciferase reporter assay. (B) NF-κB is activated by BspA through TLR2 and TLR1 but not through TLR2 and TLR6. Cells were transiently transfected with the TLR2 and TLR1 plasmids or with the TLR2 and TLR6 plasmids and incubated for 6 h with the indicated reagents. The data are the averages ± standard deviations for triplicate wells, normalized to the unstimulated cells for each TLR combination. Statistically significant cellular activation (P < 0.05) compared with the corresponding unstimulated control (A) or a statistically significant difference between TLR2/TLR1 and TLR2/TLR6 (B) as calculated by Student's t test is indicated by an asterisk. Pam3, Pam₃Cys. (C) Anti-human TLR2 antibody (Ab) inhibited BspA-mediated activation of NF-κB activity in HEK293 cells stably expressing human TLR2 and human TLR1. HEK293 (hTLR2/hTLR1) stable cells were transiently transfected with reporter plasmids, the firefly luciferase gene conjugated with NF-κB promoters (NF-κB-luc), and Renilla luciferase genes (R-luc). Cells were preincubated with anti-TLR2 MAb (TL2.1; 5 mg/ml) or an Ig isotype-matched control (IgG2a; 5 mg/ml) for 30 min before stimulation with Pam₃Cys (1 μg/ml) or BspA (5 μg/ml). After 6 h of stimulation, supernatants were collected, and the NF-κB activity in supernatants was quantified by a luciferase reporter assay. The data are the averages ± standard deviations for triplicate determinations. Statistically significant inhibition (P < 0.05) as calculated by Student's t test compared to the isotype control treatment is indicated by an asterisk. (D) HEK293 or HEK293 (hTLR2/hTLR1) cells were stimulated with BspA protein (5 μg/ml) for 16 h, and the amount of IL-8 released in the medium was determined by an ELISA. The data are the means ± standard deviations for triplicate determinations. Statistical significance (P < 0.05) as calculated by Student's t test is indicated by an asterisk.

FIG. 4.

BspA binds TLR2 but not CD14. (A) Binding of biotinylated BspA (10 μg/ml) to the indicated receptors was determined colorimetrically using receptor-coated microtiter wells probed with peroxidase-conjugated streptavidin. (B) Dose-response binding curve at the indicated concentrations of BspA. The data are means ± standard deviations of duplicate determinations in typical experiments, which were repeated three times with similar findings. BSA, bovine serum albumin.

FIG. 5.

BspA LRR domain D1 activates TLR2/TLR1. HEK293 cells cotransfected with human TLR2 and human TLR1 plasmids along with the reporter plasmid were incubated with rBspA, rBspLRR-1-2, rBspLRR-1, rBspΔLRR, or His-GST (negative control protein), each at a final concentration of 5 nM, for 6 h. The cells were harvested and lysed for further determination of NF-κB activity. The data are expressed in relative luciferase units and are averages ± standard deviations for triplicate determinations, normalized to His-GST-treated cells (control). A statistically significant difference (P < 0.05) compared to the rBspΔLRR or His-GST treatment as calculated by Student's t test is indicated by an asterisk.