Salivary histatin 3 inhibits heat shock cognate protein 70-mediated inflammatory cytokine production through toll-like receptors in human gingival fibroblasts

The stable cell lines, 293-TLR4/MD2-CD14 (InvivoGen) and 293-TLR2/CD14 (InvivoGen) and HGFs were cultured in Dulbecco’s modified Eagle medium (DMEM; Sigma-Aldrich) with 10% fetal bovine serum (FBS), 100 units/ml of penicillin G, and 100 μg/ml of streptomycin at 37°C in 5% CO₂ and 95% air in a humidified incubator. HGFs were collected from volunteers after obtaining appropriate informed consent. The Ethics Committee of Matsumoto Dental University approved the study protocol.

The following materials and antibodies were purchased: LPS from Escherichia coli 0128:B12 (Sigma-Aldrich); LPS from P. gingivalis (InvivoGen); recombinant human HSP70 (Stressgen; endotoxin activity, <0.05 endotoxin units (EU)/μg); recombinant bovine HSC70 (Stressgen; <0.05 EU/μg); recombinant bovine HSC70 ATPase fragment (Enzo Life Sciences; 0.1 EU/μg); V8 protease (Roche); 15-deoxyspergualin (DSG) (Spanidin® Inj.; Nippon Kayaku); mouse monoclonal anti-TLR4 (HTA125) and anti-TLR2 (TL2.1) antibodies (Hycult Biotechnology); mouse monoclonal anti-CD14 (MY4) antibody (Beckman Coulter); rabbit polyclonal anti-p44/42, anti-p38, anti-phosphorylated p38, anti-phosphorylated IκB-α, mouse monoclonal anti-phosphorylated p44/42 antibodies (Cell Signaling); mouse monoclonal anti-JUN-N-terminal protein kinase (JNK) (D-2), anti-phosphorylated JNK (G-7), anti-IκB-α (H-4) antibodies (Santa Cruz Biotechnology); and mouse monoclonal anti-β-actin antibody (Abcam). Endotoxin levels of all materials used were determined using a ToxinSensor chromogenic LAL endotoxin assay kit (GenScript), and the resultant level was low (~0.1 EU/μg). The LPS, bovine serum albumin (BSA), HSP70, HSC70, and the HSC70 ATPase fragment were heated at 95°C for 20 min.

Human histatins 3, 4, and 5 and P3a peptide (Biosynthesis Inc.) were chemically synthesized [7]. The control peptide, (Pro-Pro-Gly)₁₀ · 9H₂O, was purchased from Peptide Institute Inc.

pIgκB-Luc (1 μg) and pRSV-β-gal (0.1 μg), as the standard plasmid, were mixed with TransIT-LT1 transfection reagents (Mirus) [35]. The mixtures were transfected into 3 × 10⁵ cells. One day after transfection, the cells were stimulated using LPSs from E. coli (20 ng/ml) or P. gingivalis (100 ng/ml). For 293-TLR4/MD2-CD14 cells, the incubation mixtures also contained 1.4 or 14 nM BSA, HSP70, or HSC70 or 14 nM heated HSP70, heated HSC70, or the HSC70 ATPase fragment (heated or unheated). For 293-TLR2/CD14 cells, the incubation mixtures contained 2.8 or 28 nM BSA, HSP70, or HSC70 or 28 nM heated HSP70, heated HSC70, or the HSC70 ATPase fragment (heated or unheated). Cells were harvested and lysed 6 h after stimulation. Luciferase and β-galactosidase activities in the lysates were measured as described previously [35]. The luciferase activities were compared after normalization against the standard (β-galactosidase activity). For the peptide experiments, the stimulant (20 ng/ml, E. coli LPS or 14 nM, HSP70 or HSC70 for 293-TLR4/MD2-CD14 cells; 100 ng/ml, P. gingivalis LPS or 28 nM, HSP70 or HSC70 for 293-TLR2/CD14 cells) and respective peptides (3 or 30 μM; stimulation with peptide alone, 30 μM) were mixed. For the DSG experiments, HSC70 (14 nM for 293-TLR4/MD2-CD14 cells and 28 nM for 293-TLR2/CD14 cells) and the peptides (30 μM) were mixed with DSG (10 μg/ml). All the above mixtures were placed on ice for 30 min before being added to transfected cells. The cells were then cultured for 6 h. Figures show representative examples of 3 identical experiments with essentially identical results.

HGFs (1.2 × 10⁵) were cultured in DMEM containing 0.1%-0.5% FBS for 24 h. Cells were stimulated using 10 ng/ml LPS or 0.5 μg/ml HSC70 for the indicated time periods. For peptide experiments, 1.5 μg/ml HSC70 and 0.05 or 0.5 μM peptides (stimulation of peptide alone, 0.5 μM) were mixed and placed on ice for 15 min. The mixtures were added to low serum-cultured HGFs. After 30 min, the cells were lysed with radioimmunoprecipitation assay (RIPA) buffer (50 mM, Tris–HCl (pH 7.5); 5 mM, EDTA; 150 mM, NaCl; 1% NP-40; 0.1% sodium dodecyl sulfate (SDS); 0.5% sodium deoxycholate; 1 mM, phenylmethylsulfonyl fluoride; 10 μg/ml, leupeptin; 10 μg/ml, aprotinin; 5 mM, Na₃VO₄; 10 mM, NaF). Western blotting analyses were performed as previously described [35, 36]. Figures of Western blotting are representative of 3 independent experiments with similar results. Relative band intensity was measured using the ImageJ software (http://​imagej.​nih.​gov/​ij/​).

HGFs (1 × 10⁴) were cultured with HSC70 (7 or 70 nM), heated HSC70 (70 nM), or the unheated or heated HSC70 ATPase fragment (70 nM) for 24 h. For antibody experiments, HGFs were cultured with 10 μg/ml anti-TLR4, anti-TLR2, or anti-CD14 antibodies for 1 h. LPSs (from E. coli, 10 ng/ml; from P. gingivalis, 5 ng/ml) and HSC70 (70 nM) were then added to the cells, and the cells were cultured for 24 h. For peptide experiments, HSC70 (70 nM) and peptides (0.15 μM and 1.5 μM) were mixed and placed on ice for 15 min. The mixtures were added to the cells and cultured for 24 h. The culture media from all the above experiments were collected, and levels of IL-6 and IL-8 were measured using CytoSet kits (Biosource). Figures show representative examples of three identical experiments with essentially identical results.

HSC70 (2.1 μM) and peptides (10.5 μM) in a buffer (50 mM, Tris-HCl (pH 8.0); 0.5 mM, DTT; 150 mM, NaCl) were placed on ice for 10 min. After the addition of V8 protease (75 or 750 ng), the mixtures were incubated at 30°C for 1.5 h. The reactions were terminated by the addition of 0.5 volume of 2× SDS sample buffer (140 mM, Tris–HCl (pH 7.0); 6% SDS; 10% mercaptoethanol; 22.4% glycerol; 0.02% bromphenol blue; and 2 mM, phenylmethylsulfonyl fluoride). The samples were heated at 100°C for 5 min, and the proteolytic fragments were resolved using 12% SDS-polyacrylamide gel electrophoresis (PAGE) gels, followed by staining with Coomassie Brilliant Blue. Figure is representative of 3 independent experiments with similar results.

All quantitative data were statistically analyzed using either one-way analysis of variance (ANOVA) or two-way ANOVA using the StatMate software (ATMS). Differences were considered statistically significant at P < 0.05.

HSP70 stimulates inflammatory cytokine production (e.g., IL-6 and tumor necrosis factor-α) in monocytes and dendritic cells through TLR2 and TLR4 [25, 37]. However, the activation of inflammatory cytokine production by HSC70, even though it belongs to the same HSP family as HSP70, is poorly understood. Therefore, we investigated whether cytokine production by HSC70 is mediated by TLRs and their interacting proteins such as HGFs, particularly in oral cells. Because HGFs express TLR2 and TLR4, it is necessary to perform systematic experiments investigating the dependence of HSC70 stimulation on TLR2 or TLR4. It has been reported that HSP70 stimulation induces NF-κB activation in 293T cells transiently expressing TLR4 and MD2, but not in the cells expressing TLR4 alone [38]. In contrast, till now there has been no evidence that HSC70 activates NF-κB in HEK293 cells expressing TLRs. Therefore, we examined NF-κB activation in the stable cell line 293-TLR4 (HEK293 cells constitutively expressing TLR4) transiently transfected with an MD2 expression vector and an NF-κB-dependent luciferase reporter plasmid. The results indicated that HSC70 stimulated NF-κB-dependent promoter activity through TLR4/MD2 signaling pathway. Moreover, when CD14 and MD2 were both expressed in 293-TLR4 cells, the promoter activation by HSC70 was more effective than that in 293-TLR4 cells transiently expressing MD2 alone (data not shown). We then examined whether this activation was dose-dependent. The stable cell line 293-TLR4/MD2-CD14 (HEK293 cells constitutively expressing TLR4, MD2, and CD14) transfected with the NF-κB-dependent reporter plasmid (293-TLR4/MD2-CD14/NF-κB) was stimulated with HSC70, and luciferase assays were performed. As shown in Figure 1A, both HSC70 and HSP70 (control) induced promoter activation in a dose-dependent manner, whereas BSA did not. Promoter activity levels with heated HSC70 and HSP70 were similar to those in unstimulated cells. When 293-TLR4/MD2-CD14/NF-κB cells were stimulated with the heated or unheated HSC70 ATPase fragments (HSC70 with deletions of the substrate-binding and lid domains), the promoter activation was at a lower level (Figure 1B). These results suggest that HSC70 as well as E. coli LPS induces NF-κB activation through TLR4/MD2/CD14 system.

HSP70 also induces NF-κB activation in stable HEK293-TLR2 cells transiently expressing CD14, but not in HEK293-TLR2 without CD14 [25]. Therefore, we examined NF-κB activation by HSC70 in the stable cell line 293-TLR2/CD14 (HEK293 cells constitutively expressing TLR2 and CD14) transfected with the NF-κB-dependent reporter plasmid (293-TLR2/CD14/NF-κB). As shown in Figure 1C, HSC70 as well as HSP70 (control), but not BSA, induced the promoter activation through TLR2/CD14 in a dose-dependent manner. Heated and unheated P. gingivalis LPSs induced promoter activation, whereas heated HSC70 and heated HSP70 reduced activation compared with the respective unheated HSPs. When the cells were stimulated with heated and unheated HSC70 ATPase fragments, only low levels of the promoter activation were observed (Figure 1D). These results suggest that both HSC70 and P. gingivalis LPS induce NF-κB activation through TLR2/CD14.

We have previously reported that salivary protein histatin 3 binds to HSC70, but not HSP70 [7]. Here, we examined the effects of histatin 3 on NF-κB activation by HSC70 through TLR2 and TLR4. 293-TLR4/MD2-CD14/NF-κB and 293-TLR2/CD14/NF-κB cells were stimulated with mixtures of HSC70 and a control peptide, P3a (a peptide derived from the clathrin light chain that stimulates ATP hydrolysis through HSC70 [39]), or histatin 3, and luciferase assays were performed. The results showed that NF-κB-dependent promoter activation, stimulated by HSC70 through TLR4/MD2/CD14 or TLR2/CD14, was significantly suppressed by histatin 3 in a dose-dependent manner. Figure 2A shows results for TLR4/MD2/CD14 (histatin 3, 3 μM (P < 0.001) and 30 μM (P < 0.001)), and Figure 2B for TLR2/CD14 (histatin 3, 3 μM (P < 0.01) and 30 μM (P < 0.001)). However, histatin 3 did not affect promoter activation caused by LPS (from E. coli or P. gingivalis) or HSP70, in either cell type. Neither the control peptide nor P3a inhibited the LPS-, HSP70-, or HSC70-induced activation. We then tested whether other members of the histatin family were capable of suppressing NF-κB-dependent activation through TLR4/MD2/CD14 or TLR2/CD14. 293-TLR4/MD2-CD14/NF-κB and 293-TLR2/CD14/NF-κB cells were stimulated with mixtures of HSC70 and histatins 3, 4, or 5, and luciferase assays were performed. As shown in Figures 2C and 2D, HSC70-induced promoter activation was significantly reduced in the presence of histatin 3 compared with that in the presence of histatin 5 (P < 0.05 in both cell types). The suppressive effects of histatin 4 on the HSC70-mediated promoter activation were at low levels. These results suggest that histatin 3 bound to HSC70 inhibits HSC70-induced NF-κB activation both in TLR4/MD2/CD14 and TLR2/CD14 mode.

We investigated whether the specific interaction of histatin 3 and HSC70 was necessary for inhibition of HSC70-induced NF-κB activation. The NF-κB-driven luciferase assays were performed on 293-TLR4/MD2-CD14/NF-κB and 293-TLR2/CD14/NF-κB cells treated with HSC70 along with or without histatin 3 in the presence or absence of DSG. DSG is a synthetic analog of spergualin, a natural product from Bacillus laterosporus, and has a peptidomimeric structure. It binds particularly well to HSC70 and is considered to preclude peptide binding to HSC70 [40, 41]. The results demonstrated that in the presence of histatin 3 and DSG HSC70-stimulated NF-κB-dependent promoter activity was significantly higher than with histatin 3 alone in both 293-TLR4/MD2-CD14 (Figure 3A, P < 0.001) and 293-TLR2/CD14 (Figure 3B, P < 0.001) cells. These results suggest that the specific binding of histatin 3 to HSC70 is important for inhibition of HSC70-mediated activation of TLR2 and TLR4 signaling activation.

It is unknown whether HSC70 is involved in inflammatory response in HGFs, especially in the production of inflammatory cytokines. To examine this possibility, HGFs were stimulated with HSC70 for 24 h, and the amounts of IL-6 and IL-8 in the culture media were measured by ELISAs. As shown in Figures 4A and 4B, unheated HSC70 significantly induced cytokine production in HGFs in a dose-dependent manner, whereas heated HSC70 did not induce cytokine production (HSC70, 7 nM (P < 0.01) and 70 nM (P < 0.001) in Figure 4A; 7 nM (P < 0.05) and 70 nM (P < 0.01) in Figure 4B). When HGFs were stimulated with the unheated and heated HSC70 ATPase fragments, the levels of IL-6 and IL-8 were similar to those in unstimulated cells. These results suggest that HSC70 induces inflammatory cytokine production in HGFs.

To examine whether inflammatory cytokine production stimulated by HSC70 was dependent on TLR2 or TLR4 in HGFs, we performed experiments using anti-TLR2 and anti-TLR4 antibodies. As shown in Figure 5A, levels of HSC70-stimulated IL-6 production in HGFs with anti-TLR4, anti-TLR2, and both antibodies significantly decreased (P < 0.001). LPS-stimulated IL-6 production also significantly decreased in the presence of anti-TLR4 and both anti-TLR2 and anti-TLR4 antibodies (P < 0.001). The patterns of HSC70- and LPS-stimulated IL-8 production in the presence of the antibodies were similar to those of IL-6 production under the same condition (Figure 5B). Next, we examined the effect of anti-CD14 antibody on HSC70- and LPS-stimulated inflammatory cytokine production in HGFs. HSC70-stimulated IL-6 and IL-8 production significantly decreased in the presence of this antibody (P < 0.001) (Figures 5C and 5D). The antibody also caused a decrease in cytokine production in LPS-stimulated HGFs. The levels of IL-6 and IL-8 production under HSC70 stimulation were approximately1.6-fold higher than those by LPS stimulation. This behavior is similar to the previously reported findings for HSP70 stimulation of cytokine production through a CD14-dependent pathway [37]. These results suggest that HSC70-stimulated inflammatory cytokine production in HGFs is dependent on TLR2, TLR4, and CD14.

Histatin 3 was found to inhibit NF-κB activation through TLR2 and TLR4 (Figure 2). We then examined the effect of histatin 3 on HSC70-stimulated inflammatory cytokine production in HGFs. HSC70 and histatin 3 were added to HGFs, and the amounts of IL-6 and IL-8 released into the culture media were measured by ELISAs. As shown in Figures 6A and 6B, HSC70-stimulated cytokine production significantly decreased in the presence of histatin 3, in a dose-dependent manner (histatin 3, 0.15 μM (P < 0.001) and 1.5 μM (P < 0.001)). The control peptide and P3a did not appreciably affect the cytokine production. We then investigated whether other members of the histatin family were capable of inhibiting HSC70-stimulated cytokine production in HGFs. Histatins 3, 4, and 5 were mixed with HSC70 and used to stimulate HGFs, and the amounts of IL-6 and IL-8 in the culture media were measured. As shown in Figures 6C and 6D, the inhibitory effect of histatin 3 on the cytokine production was significantly higher than that of histatin 5 (IL-6, P < 0.01; IL-8, P < 0.001). These results suggest that histatin 3 is an inhibitory factor of HSC70-stimulated inflammatory cytokine production in HGFs.

We next examined whether HSC70 stimulated MAPKs (extracellular signal-regulated kinases (ERK (p42/44)), JNK, and p38) phosphorylation and IκB-α degradation in HGFs. HGFs were cultured with HSC70 for 10, 30, and 60 min, and extracted proteins were analyzed by Western blotting with anti-MAPK and anti-IκB-α antibodies. As shown in Figures 7A and 7B, enhanced phosphorylation of p42/44, JNK, p38, and IκB-α and stimulated significant degradation of IκB-α by HSC70 were observed. LPSs also stimulated significantly both the phosphorylation of MAPKs and IκB-α and the degradation of IκB-α.

To examine whether HSC70-induced ERK phosphorylation and IκB-α degradation were inhibited by histatin 3, HGFs were stimulated with HSC70 in the presence of histatin 3, and Western blotting was performed. As shown in Figures 7C and 7D, p42/44 phosphorylation induced by HSC70 in the presence of histatin 3 was lower than that in the absence of histatin 3 or in the presence of the control peptide. Stimulation of either the control peptide or histatin 3 alone decreased levels of p42/44 phosphorylation. NF-κB activation by stimulation with the control peptide or histatin 3 alone was also diminished in 293-TLR4/MD2-CD14 and 293-TLR2/CD14 cells (Figures 2A and 2B). HSC70 stimulated IκB-α degradation in the presence of the control peptide, but not in the presence of histatin 3. These results suggest that histatin 3 inhibits HSC70-mediated MAPK phosphorylation and IκB-α degradation in HGFs.

Histatin 3 prevented HSC70-stimulated NF-κB activation through TLR2 and TLR4 (Figure 2) and inflammatory cytokine production in HGFs (Figure 6). Therefore, we assumed that histatin 3 binding to HSC70 had some effect on the conformation of HSC70. To examine this assumption, HSC70 was mixed with histatin 3, and the mixture was analyzed by limited V8 protease proteolysis. As shown in Figure 8, HSC70 with histatin 3 was relatively more resistant to digestion by V8 protease with HSC70 with the control peptide or P3a. Histatin 3 did not inhibit the activity of V8 protease (data not shown). These results suggest that histatin 3 may affect the conformation of HSC70 upon binding.