Regulation of ICAM-1 expression in gingival fibroblasts infected with high-glucose-treated P. gingivalis


For correspondence. E-mail; Tel. (+886) 5 271 7836; Fax (+886) 5 271 7780.


Porphyromonas gingivalis is a major pathogen in the initiation and progression of periodontal disease, which is recognized as a common complication of diabetes. ICAM-1 expression by human gingival fibroblasts (HGFs) is crucial for regulating local inflammatory responses in inflamed periodontal tissues. However, the effect of P. gingivalis in a high-glucose situation in regulating HGF function is not understood. The P. gingivalis strain CCUG25226 was used to study the mechanisms underlying the modulation of HGF ICAM-1 expression by invasion of high-glucose-treated P. gingivalis (HGPg). A high-glucose condition upregulated fimA mRNA expression in P. gingivalis and increased its invasion ability in HGFs. HGF invasion with HGPg induced increases in the expression of ICAM-1. By using specific inhibitors and short hairpin RNA (shRNA), we have demonstrated that the activation of p38 MAPK and Akt pathways is critical for HGPg-induced ICAM-1 expression. Luciferase reporters and chromatin immunoprecipitation assays suggest that HGPg invasion increases NF-κB- and Sp1-DNA-binding activities in HGFs. Inhibition of NF-κB and Sp1 activations blocked the HGPg-induced ICAM-1 promoter activity and expression. The effect of HGPg on HGF signalling and ICAM-1 expression is mediated by CXC chemokine receptor 4 (CXCR4). Our findings identify the molecular pathways underlying HGPg-dependent ICAM-1 expression in HGFs, providing insight into the effect of P. gingivalis invasion in HGFs.


Periodontal disease, the most common lytic disease of the bone, consists of a bacteria-induced inflammatory disorder that involves the structures that support the teeth (Pihlstrom et al., 2005). It is initiated by bacteria that induce an inflammatory response in the periodontal tissue, in most cases being caused by a Gram-negative anaerobic bacterium, Porphyromonas gingivalis (Darveau, 2010). Human gingival fibroblasts (HGFs), the major cell type found in periodontal connective tissue, play an active role in host defence. Previously, it has been reported that HGFs can recognize and respond to pathogens, and can produce inflammatory mediators when they are activated by pathophysiological stimuli (Souza et al., 2010; Scheres et al., 2011). Thus, it is clarified that HGFs regulate local inflammatory responses in inflamed periodontal tissue. Furthermore, the link between diabetes and periodontal disease has been well established, and it is clear that poorly controlled diabetes is a significant contributor to the development of periodontal disease (Lalla and Papapanou, 2011; Preshaw et al., 2012). Although previous studies have demonstrated that periodontal disease is one of the clinical manifestations of diabetes and is recognized as a common complication in diabetic patients, the effects of high-glucose (HG)- versus normal-glucose (NG)-treated P. gingivalis on the gene expression of HGFs have not yet been clearly evaluated.

Adherence of P. gingivalis to the host cells of periodontal connective tissue is the first step for P. gingivalis to build an infection (Bostanci and Belibasakis, 2012). The pathogenicity of P. gingivalis is attributed to a number of potential virulent factors. Among them, fimbriae of P. gingivalis have been considered a crucial virulent factor for periodontal disease initiation and progression (Yoshimura et al., 2009). P. gingivalis fimbriae are cell surface structures that facilitate adherence of the bacteria to host cells, such as oral epithelial cells or fibroblasts. Interactions between P. gingivalis and host cells by fimbriae not only promote bacterial attachment and invasion, but also stimulate a number of host responses that subsequently influence the effect of a bacterial infection (Azuma, 2006). One of the major fimbriae proteins of P. gingivalis is fimbrillin (Fim) A encoded by the fimA gene. The fimA gene, which has six variants (types I–V and type Ib), plays a critical role in determining the disease-associated genotypes of P. gingivalis involved in the development of periodontal disease (Kato et al., 2007).

The intercellular cell adhesion molecule-1 (ICAM-1) is a transmembrane molecule known to be involved in cell–cell adhesion and to regulate the binding and extravasation of leucocytes from the bloodstream to sites of inflammation. ICAM-1 seems to be overexpressed in inflamed gingival tissue, which is related to increased levels of soluble ICAM-1 (Schenkein et al., 2007), indicating a crucial role for ICAM-1 in inflammatory periodontal disease. When HGFs are activated in response to IL-1β, TNF-α and IFN-γ, the expression of ICAM-1 on their surface is upregulated (Hosokawa et al., 2006). Although it has been well established that HG induces the expression of ICAM-1 in different cell types, and many reports have been made on the expression of ICAM-1 in fibroblasts (Ozawa et al., 2003; Nizamutdinova et al., 2009; Song et al., 2011), little is known regarding HGFs in terms of ICAM-1 expression after infection with P. gingivalis cultured under HG conditions.

Because hyperglycemia has been associated with inflammatory responses linked to periodontal disease, it was hypothesized that P. gingivalis cultured in HG conditions may upregulate HGF ICAM-1 expression. In addition, it has been reported that the fimbriae bind the CXC-chemokine receptor 4 (CXCR4), leading to enhanced P. gingivalis persistence. P. gingivalis can bind and activate CXCR4 directly by its fimbriae and transduced signals to the intracellular compartment, further modulating cell function and causing disease (Hajishengallis et al., 2008). These signals may lead to the activation of mitogen-activated protein kinases (MAPKs) and phosphatidylinositol 3-kinase (PI3K)/Akt pathways. The goal of this study was to elucidate the signalling network that orchestrates the expression of HGF ICAM-1 by HG-treated P. gingivalis (HGPg). The results demonstrated that the ICAM-1 upregulation induced by HGPg is mediated through the CXCR4, the intracellular signalling cascades p38 MAPK and Akt, and the transcription factors nuclear factor-κB (NF-κB) and Sp1. Our findings provide a molecular basis for the mechanisms by which HGPg enhances ICAM-1 expression in HGFs.


Invasion of the HGFs by P. gingivalis cultured under different conditions

The effects of P. gingivalis cultured on mannitol (M), NG and HG conditions in mediating bacterial invasion by HGFs were investigated using gentamicin protection assays. As shown in Fig. 1A, P. gingivalis cultured under HG conditions (HGPg) invaded HGFs much more readily than P. gingivalis cultured in mannitol (MPg) and NG (NGPg) conditions. The invasion ability of HGPg was enhanced when the glucose concentration in culture medium was increased (Fig. 1B). In addition, the invasion number of HGPg on HGFs was dose-dependent (Fig. 1C). The fimA genotype of P. gingivalis CCUG25226 was assessed by PCR using primer pairs specific for the six different fimA genes (I–V and Ib). P. gingivalis CCUG25226 used in this study was found to carry type III fimA genotype (data not shown). The fimA mRNA expression in P. gingivalis under NG and HG conditions was detected by real-time PCR. As shown in Fig. 1D, different expression levels of fimA mRNA were found in NGPg and HGPg, with the significantly higher expression in HGPg. In addition, transmission electron micrographs clearly showed that the HGPg had decreased capsule expression on the cell surface (Fig. 1E).

Figure 1.

Invasion of the HGFs by P. gingivalis.

A. Invasion percentage of HGFs by P. gingivalis grown in normal glucose (NGPg), mannitol (MPg), and high-glucose (HGPg) conditions. *P < 0.05 versus NGPg or MPg.

B. Invasion percentage of HGFs by either MPg or HGPg cultured in different concentrations of mannitol or glucose. *P < 0.05 versus MPg at the same concentrations.

C. Invasion percentage of HGFs by 20 or 50 multiplicity of infection (moi) of NGPg or HGPg. *P < 0.05 versus NGPg at the same moi.

D. Analysis of fimA mRNA expression in NGPg, MPg and HGPg. RNA samples from P. gingivalis were isolated and subjected to real-time PCR analysis. The mRNA data are presented as fold changes in fluorescent density from control P. gingivalis normalized to the 16S rRNA levels. *P < 0.05 versus NGPg or MPg.

E. Representative transmission electron images of P. gingivalis grown in NG or HG conditions. The capsule thickness is markedly reduced on the cell surface of HGPg.

HGPg enhanced ICAM-1 expression in HGFs

The effects of P. gingivalis from a HG condition on the expression of ICAM-1 in HGFs were studied using infected cells with HGPg (versus NGPg). The changes in ICAM-1 expression compared with the control cells at the same time points were analysed by real-time PCR, and cell surface ICAM-1 expression was detected by ELISA. The time-courses determined for the ICAM-1 mRNA levels (Fig. 2A) revealed an increase after 1 h of HGPg infection and a peak expression at 4 h. At the 4 h time point, the invasion of the HGFs by HGPg-induced increases in the ICAM-1 mRNA levels by 8.6-fold compared with NGPg-infected HGFs (Fig. 2A). The invasion of HGFs by HGPg also caused significant increases in the cell surface ICAM-1 expression at 4 h after stimulation (Fig. 2B). In addition, the induction of ICAM-1 mRNA expression and cell surface ICAM-1 by HGPg invasion was dose-dependent (Fig. 2C and D). HGPg invasion also induced an increase in total ICAM-1 protein expression in HGFs in a time-dependent manner (Fig. 2E).

Figure 2.

Induction of ICAM-1 expression in HGFs invasion with P. gingivalis. RNA samples were isolated at the indicated time periods or moi. All bar graphs represent folds of control cells (CL) normalized to 18S rRNA by real-time PCR analysis (A and C). Cell surface ICAM-1 was determined by cell surface ELISA analyses (B and D). (A and B) Cells were invaded by NGPg or HGPg (20 moi) for the times indicated, or (C and D) invasion with NGPg or HGPg at various moi for 4 h. Data are shown as mean ± standard error of the mean (SEM). *P < 0.05 versus CL HGFs or HGFs infected with NGPg. (E) The expression of total ICAM-1 in HGF cell lysate after HGPg invasion for the times indicated was determined using Western blotting.

HGPg invasion only had marginal effects on VCAM-1 mRNA expression in HGFs (Fig. 3A). The effect of glucose and insulin on HGF ICAM-1 expression is shown in Fig. 3B. The increases in ICAM-1 mRNA expression were 2.3-fold in HGFs stimulated with 25 mM glucose for 4 h. HGFs stimulated with 5 mM glucose and 50 microunits of insulin had a minor effect on ICAM-1 expression.

Figure 3.

A. HGFs were maintained as untreated controls (CL) or invaded by HGPg for the times indicated. RNA samples were isolated and subjected to real-time PCR analysis. The mRNA data are presented as the fold changes compared with CL HGFs normalized to the 18S rRNA level.

B. HGFs were grown as untreated CL or stimulated with 5 mM, 25 mM glucose (Glc), 50 microunits insulin, or invaded with HGPg for 4 h. RNA samples were then isolated and subjected to real-time PCR analysis. *P < 0.05 versus control HGFs. #P < 0.05 versus 25 mM Glc.

HPGg-induced ICAM-1 expression in HGFs is mediated by the p38 and Akt pathways

To determine whether HGPg-induced ICAM-1 expression is mediated through the MAPK- and PI3K/Akt-dependent pathways, HGFs were incubated with specific inhibitors (dissolved in DMSO) for ERK (PD98059, 30 μM), JNK (SP600125, 20 μM), p38 (SB203580, 10 μM) and PI3K/Akt (LY294002, 20 μM) for 1 h before and during invasion with HGPg. The HGPg-induced mRNA expression (Fig. 4A) and HGF surface protein expression (Fig. 4B) of ICAM-1 was found to be significantly inhibited by SB203580 and LY294002 but not by PD98059 and SP600125. To further confirm the involvement of p38 and Akt in the modulation of ICAM-1 expression by HGPg invasion, we examined the effects of expressing specific shRNAs that target these signalling pathways upon HGPg-induced ICAM-1 expression in HGFs. HGPg-induced ICAM-1 mRNA (Fig. 4A) and cell surface protein expression (Fig. 4B) on HGFs was inhibited by transfection with p38- and Akt-specific shRNAs but not by transfection with control shRNA (100 μmol ml−1 for each). The effectiveness of the silencing was validated because p38 and Akt shRNA (compared with control shRNA) caused an 85% reduction in p38 and Akt protein expressions respectively (data not shown). The phosphorylation of p38 and Akt in HGFs increased rapidly after HGPg invasion, reaching maximal levels at 30 min (Fig. 4C). HGPg significantly caused p38 and Akt phosphorylation after 30 min of treatment, whereas NGPg had a minor effect on p38 and Akt phosphorylation in HGFs (Fig. 4D). In addition, the phosphorylation levels of p38 and Akt were not affected in control cells without HGPg infection (data not shown).

Figure 4.

p38 and Akt pathways are required for HGPg-induced ICAM-1 expression in HGFs. HGFs were kept as CL or invaded by HGPg (20 moi) for 4 h. Before being kept as CL or invaded by HGPg, HGFs were pretreated with DMSO, PD98059 (PD), SP600125 (SP), SB203580 (SB), or LY294002 (LY) individually for 1 h, or transfected with control shRNA (sh-CL), or a specific shRNA of sh-p38 or sh-Akt.

A. All bar graphs represent multiple increases of CL HGFs normalized to 18S rRNA.

B. The cell surface ICAM-1 was determined by cell surface ELISA analyses. The results are shown as mean ± SEM. *P < 0.05 versus CL. #P < 0.05 versus vehicle control (DMSO) or control shRNA (sh-CL) with HGPg invasion.

C. Control or HGPg-infected HGFs were maintained for the times indicated, and the phosphorylation of p38 and Akt in these cells was determined using Western blotting.

D. The phosphorylation of p38 and Akt in HGFs after 30 min of NGPg or HGPg invasion was determined using Western blot analysis. Bar graphs represent folds of CL HGFs, mean ± SEM. *P < 0.05 versus CL (C and D).

NF-κB and Sp1 binding sites are essential for the HGPg induction of human ICAM-1 promoter activity

To identify the cis-acting elements in the ICAM-1 gene promoter that mediate HGPg invasion-induced ICAM-1 transcription, luciferase assays were conducted with the p540-Luc plasmid and several deletion promoter constructs (Fig. 5A). In HGFs, the +540/-1 region of ICAM-1 was found to direct maximum luciferase activity. A sequence deletion from −323 to −238 caused this reporter activity to decrease to 70–75% of the control levels, whereas a further deletion to the −99 position almost completely eliminated the activity (Fig. 5A). To investigate whether HGPg invasion can induce NF-κB and Sp1 binding activities in HGFs, we performed ChIP analysis of the Sp1 and NF-κB p65-DNA binding activities. Immunoprecipitated chromosomal DNA with an Sp1 antibody was subjected to PCR using primers designed to amplify the ICAM-1 promoter region harbouring the Sp1 binding site. Sp1 was indeed found to bind the ICAM-1 promoter region containing the Sp1 site (Fig. 5B). Similarly, the region containing the NF-κB site was specifically immunoprecipitated with NF-κB p65 antibody (Fig. 5C). NF-κB p65 binding activity was further analysed in vitro by using a TF ELISA kit from Panomics. The infection of HGFs with HGPg caused NF-κB p65-DNA binding activity to increase after 1 h and remain elevated for at least 2 h (Fig. 5D). In addition, as shown in Fig. 5E, HGPg infection also induced degradation of IκB protein in HGFs in a time-dependent manner.

Figure 5.

Luciferase reporter analysis of the ICAM-1 promoter and the roles of Sp1 and NF-κB in HGPg-induction of ICAM-1 promoter activity and transcription in HGFs.

A. HGFs were cotransfected with 5′-deletion constructs and invasion with HGPg for 4 h. ICAM-1 promoter activity was measured by luciferase assay normalized to β-galactosidase activity and is shown relative to that of HGFs transfected with p540-Luc (set at 100%). *P < 0.05 versus p540-Luc with HGPg invasion.

B and C. ChIP assays were performed for Sp1 (B) and NF-κB (C) by using p65 antibody.

D. HGFs were kept as CL or infected with HGPg for the times indicated, and NF-κB p65 activation was determined by a TF ELISA assay.

E. HGFs were kept as CL or infected with HGPg for the times indicated, and the IκB protein expression was determined by Western blot analysis.

F. ICAM-1 mRNA and p540-Luc activity were determined in HGFs transfected (48 h) with sh-CL, sh-Sp1, or sh-p65 and then invaded by with HGPg for 4 h. *P < 0.05 versus CL. #P < 0.05 versus sh-CL-transfected HGFs with HGPg invasion.

We further tested whether Sp1 and NF-κB activation are involved in the signal transduction pathway leading to the HGPg induction of ICAM-1 gene expression. HGFs were transfected with shRNAs for Sp1 and p65, followed by invasion with HGPg for 4 h. The HGPg-induced ICAM-1 mRNA and ICAM-1 p540-Luc promoter activity levels (Fig. 5F) were significantly downregulated by the inhibition of Sp1 or p65 by shRNA, indicating that Sp1 and NF-κB are involved in the regulation of ICAM-1 gene expression. The Sp1- and p65-specific shRNAs (compared with control shRNA) caused an 80% reduction in Sp1 and p65 protein expressions respectively (data not shown).

The p38 and Akt signalling pathways are involved in HGPg invasion-induced ICAM-1 promoter activity

To evaluate whether the inhibition of ICAM-1 expression by the p38 and Akt signalling pathways occurs at the transcriptional level, we studied the effects of inhibitors or shRNAs against p38 and Akt upon HGPg-induced ICAM-1 p540-Luc promoter activity. Invasion of the HGFs by HGPg increased the luciferase activity compared with unstimulated cells after normalization with a transfection control (Fig. 6A). Pretreatment of the cells with SB203580 and LY294002 or transfection with p38 shRNA and Akt shRNA resulted in a marked inhibition of HGPg-induced ICAM-1 promoter activity (Fig. 6A). In addition, ChIP analysis revealed that the pretreatment of HGFs with SB203580 or LY294002 inhibited the HGPg induction of the Sp1 (Fig. 6B) and NF-κB promoter (C) binding activities.

Figure 6.

p38 and Aky signalling pathways are involved in HGPg-induced ICAM-1 promoter activity.

A. ICAM-1 p540-Luc activity was determined in HGFs pretreated with PD98059 (PD), SP600125 (SP), SB203580 (SB), or LY294002 (LY) individually for 1 h, or transfected with sh-CL, sh-p38, or sh-Akt and then invaded by HGPg for 4 h. All bar graphs represent the fold increases over the control HGF levels (CL), calculated as the mean ± SEM from three independent experiments. *P < 0.05 versus CL. #P < 0.05 versus DMSO-treated or sh-CL-transfected HGFs with HGPg invasion.

B and C. ChIP assays were performed for Sp1 (B) and NF-κB (C, using p65 antibodies) in HGFs pretreated with DMSO, SB203580 (SB), or LY294002 (LY).

CXCR4 regulates the HGPg-induced activation of NF-κB and Sp1 and the expression of ICAM-1

To assess the role of CXCR4 in HGPg-induced ICAM-1 expression in HGFs, we evaluated the effects of CXCR4 antagonist AMD3100 (20 μg ml−1, dissolved in DMSO) and CXCR4-siRNA (100 μmol ml−1) on HGPg-induced ICAM-1 expression, and NF-κB and Sp1 activation. Pretreatment of HGFs with AMD3100 or transfection with CXCR4-siRNA significantly inhibited the HGPg-induced ICMA-1 mRNA expression (Fig. 7A). In addition, the HGPg-induced Sp1 and NF-κB promoter binding activities were also inhibited by cells pretreated with AMD3100 (Fig. 7B and C). We further examine whether HgPG can regulate the CXCR4 expression on HGFs. As shown in Fig. 7D, HgPG infection decreased CXCR4 expression by HGFs.

Figure 7.

Pretreatment of HGFs with inhibitor or siRNA against CXCR4 inhibited HGPg-induced ICAM-1 expression.

A. HGFs were kept as controls (CL), or pretreated with vesicle (DMSO) and AMD3100, or transfected with the control siRNA (si-CL) and si-CXCR4, and then invasion with HGPg (20 moi) for 4 h (A). All bar graphs represent folds of control HGFs (CL), mean ± SEM. *P < 0.05 versus CL HGFs. #P < 0.05 versus DMSO or si-CL-transfected HGFs with HGPg invasion.

B and C. ChIP assays were performed for Sp1 (B) and NF-κB (C, using p65 antibodies) in HGFs pretreated with vesicle (MeOH) or AMD3100 (AMD).

D. Flow cytometry shows the cell surface expression of CXCR4. HGFs were kept as CL or infected with HGPg. Cells were incubated with mAb against CXCR4 or isotype-matched IgG served as a blank.

HGPg-induced neutrophil cell adhesiveness was inhibited by AMD3100

We examined the alterations in adhesiveness of the HGPg-stimulated HGFs for neutrophils. HGFs invasion with HGPg for 4 h significantly increased the adhesion of neutrophils in comparison to that of the controls and NGPg respectively (Fig. 8A). Pretreatment of HGFs with neutralizing antibody against ICAM-1 or CXCR4 (Fig. 8B) significantly inhibited the adhesion of neutrophils to the HGPg-infected HGFs.

Figure 8.

Neutrophil adhesiveness in HGPg-infected HGFs. All bar graphs represent the fold increases over the control HGFs (CL), calculated as the mean ± SEM from three to four experiments.

A. Neutrophils were labelled with calcein acetoxymethyl and added to HGFs that were invaded by NGPg and HGPg for 4 h. *P < 0.05 versus CL HGFs or HGFs with NGPg invasion.

B. Calcein acetoxymethyl-labelled neutrophils were added to HGFs treated with isotype-matched IgG or 20 μg ml−1 neutralizing antibodies against ICAM-1 (Ab-ICAM) or CXCR4 (Ab-CXCR4) during 4 h of HGPg invasion. *P < 0.05 versus CL. #P < 0.05 versus IgG-treated HGFs with HGPg invasion.

Higher ICAM-1 expression in human periodontitis tissues from diabetes patients

Representative images of periodontal tissue sections with ICAM-1 immunostaining are presented in Fig. 9A (patients with periodontal disease alone) and Fig. 9B (patients with both diabetes and periodontal disease). These images show that the intensity and area of ICAM-1 immunostaining was increased in diabetic patients with periodontal disease compared with patients with periodontal disease alone. ICAM-1 mRNA expression levels were further analysed in gingival samples obtained from diabetic patients with periodontal disease and patients with periodontal disease alone. As shown in Fig. 9C, when compared with ICAM-1 gene expression in patients with periodontal disease alone, ICAM-1 gene expression was significantly upregulated in samples from diabetic patients with periodontal disease. In addition, we also performed real-time RT-PCR to detect fimA gene expression in gingival samples. Figure 9D shows the relative expression for fimA gene in diabetic patients with periodontal disease and patients with periodontal disease alone. When compared with fimA gene expression in patients with periodontal disease alone, fimA gene expression shows increased relative levels in diabetic patients with periodontal disease.

Figure 9.

ICAM-1 expression is increased in gingival tissues from diabetic patients with periodontal disease. Representative images of immunohistochemical localization of ICAM-1 in (A) patients with periodontal disease alone and (B) patients with both diabetes and periodontal disease. The black arrows indicate periodontal granulation tissues. (C and D) Comparison of ICAM-1 (C) and fimA (D) expression levels in gingival tissues from patients with periodontal disease alone and patients with both diabetes and periodontal disease. A statistically significant difference was observed between the study groups (*P < 0.05).


Recent studies indicate that bacterial fimbriae expression and infection causes intracellular signalling pathways in host cells that induce inflammatory responses and cellular dysfunction (Hajishengallis et al., 2005; Zheng et al., 2011; Chen et al., 2011a). It has also been reported that the ability of P. gingivalis to invade cells in gingival tissues plays a key role in the pathogenesis of periodontal disease (Andrian et al., 2006). In addition, periodontal disease is considered a complication of diabetes; it seems that diabetes favours the occurrence of periodontopathy by inflammatory responses to the pathogen infections (Nishihara et al., 2009; Lalla and Papapanou, 2011). Our present study aimed to link the HG conditions and P. gingivalis infection to HGF ICAM-1 expression. Our results are significant in several major respects: (1) P. gingivalis CCUG25226 carries type III fimA genotype and an environment with an HG favourable fimA mRNA expression; (2) HGPg has a greater ability than NGPg to invade HGF; (3) HGPg invasion enhances the expression of ICAM-1 in HGFs compared with NGPg-treated HGFs; (4) HGPg-induced expression of ICAM-1 is mediated by the p38 and Akt, and NF-κB/Sp1 pathways; and (5) CXCR4 is required for HGPg-induced NF-κB and Sp1 activation and ICAM-1 expression.

It is becoming increasingly evident that P. gingivalis strains with different fimA genotypes display distinct virulence features (Kato et al., 2007). The fimA type II was reported to be associated with periodontal disease, whereas fimA type I was prevalent in healthy adults. In addition, types II, Ib and IV were found to be involved in the progress of apical periodontitis (Wang et al., 2010) and caused severe inflammatory responses in mice, whereas type I and III induced slight inflammation (Nakano et al., 2004). It has been observed that tobacco smoke increased the expression of fimA, and promoted P. gingivalis colonization and infection (Bagaitkar et al., 2010). In addition, a recent study also indicated that the capsule of P. gingivalis makes it less efficient in invading gingival fibroblasts (Irshad et al., 2012). The findings presented here have established that the P. gingivalis strain CCUG25226 carries a fimA type III genotype. In addition, NGPg has a lower fimA mRNA expression level and a lesser ability to invade HGFs; however, when cultured under HG conditions, the fimA mRNA was markedly upregulated, the thickness of the capsule was decreased. and the invasion efficiency to HGFs was also significantly increased by HGPg.

The presence of P. gingivalis is necessary for periodontal disease development, as the role of the host inflammatory response appears to be the critical determinant for susceptibility, especially in patients with diabetes. However, the relationship between HG conditions and P. gingivalis invasion is not totally understood. In the gingival tissues, a hyperglycemia status is considered to create an inflamed environment triggered by the penetration of microbial pathogens from the oral cavity. Previously it was reported that for type II diabetes patients with periodontitis, the prevalent types of P. gingivalis fimA include type III (Davila-Perez et al., 2007). The pathogenesis of periodontal disease is characterized by an infiltration of neutrophils into the periodontal tissues (Nussbaum and Shapira, 2011). A greater influx of neutrophils and pro-inflammatory mediators may contribute to the greater incidence and severity of periodontal disease observed in patients with diabetes (Yoon et al., 2012). HGFs are critical for regulating the homeostasis of healthy and diseased periodontal tissues. Upregulation of ICAM-1 on the surface of HGFs may play a key role in infiltration and retention of leucocytes at sites of periodontal diseased tissue. Furthermore, increased accumulation of neutrophils and upregulation of ICAM-1 have been shown to relate to the severity and activity of periodontal disease. It has also been reported that HGFs can be stimulated by the direct surface interaction of ICAM-1 with neutrophils (Liu et al., 2001), indicating that direct ICAM-1-mediated interaction of HGF with leucocytes may cause further activation of HGF.

ICAM-1 plays an important role in leucocyte adhesion and transendothelial migration at sites of inflammation (Ley et al., 2007). Therefore, investigation of factors influencing the induction of ICAM-1 is important in regulating inflammatory processes. It is noteworthy that the expression of inflammation biomarkers in the development of periodontal disease refers to ICAM-1. ICAM-1 has been reported as a therapeutic target against inflammatory processes of the human gingiva based on local topical application of ICAM-1-directed antisense oligonucleotides (Nedbal et al., 2002). In addition, it was shown that ICAM-1 plays an important role in P. gingivalis invasion into oral epithelial cells, contributing to the recruitment of leucocytes (Tamai et al., 2005). Another previous study indicated that P. gingivalis increased ICAM-1 expression in vascular endothelial cells directly, rather than via inflammatory mediators released from cells in an autocrine manner (Zhang et al., 2011). In the present study, we investigated the molecular mechanism by which HGPg stimulates ICAM-1 expression in HGFs. The results of this study demonstrate that HGPg invasion not only promotes the cell surface ICAM-1 but also induces their gene transcription and total protein expression in HGFs. We have obtained several lines of evidence indicating that the HGPg-induced ICMA-1 expression in HGFs is mediated via p38 and Akt, and NF-kB and Sp1 activations. First, invasion of HGPg to the HGFs stimulated ICAM-1 mRNA expression and protein production by HGFs in an in vitro culture system. Second, luciferase reporters and ChIP assays demonstrated an increase in NF-κB and Sp1 binding to the promoter region of the ICAM-1 gene in HGFs. Third, the inhibition of NF-κB and Sp1 activations in HGFs through pretreatment with inhibitors, or transfection with specific shRNAs of p38, Akt, NF-κB p65 and Sp1, abolished HGPg-induced ICAM-1 expression. Our results suggest that when HGFs encounter a stimulus of HGPg, HGPg enhances ICAM-1 expression via CXCR4-mediated signal transduction. Our findings suggest that the augmented expression of ICAM-1 on HGFs may be involved in HGPg-associated gingival inflammation specifically, contributing to periodontopathic bacteria-mediated gingival connective tissue injury.

In order to induce gene expression, cells need to transduce signals from cellular receptors into the interior of the cells. It is clear that the immune evasion ability of P. gingivalis depends on its surface fimbriae to instigate functional toll-like receptor 2 (TLR2)/CXCR4 co-association (Hajishengallis et al., 2008). However, whether CXCR4 is involved in the initiation of the cascade of P. gingivalis invasion is controversial. Here, we showed that CXCR4 mediates HGPg-induced ICAM-1 expression through the activation of NF-κB and Sp1. By using the inhibitor AMD3100, the blockade of CXCR4 affected HGPg-induced HGF NF-κB and Sp1 activation, and ICAM-1 expression. We have also demonstrated the functional consequence of the CXCR4 modulation of the HGPg-induced HGF gene expression by measuring the adhesiveness of the neutrophils to HGPg-treated HGFs. HGFs pretreated with AMD3100 blocked the HGPg-induced neutrophil adhesion on HGFs. Recent study indicated that TLRs mediated the infection of P. gingivalis into host cells (Hajishengallis et al., 2009). However, since there is no direct evidence to show the interaction between CXCR4 and TLRs in HGFs in this study, further investigation is required to better understand their relationship.

In summary, our results contribute information regarding the mechanisms by which HGPg induces ICAM-1 expression in HGFs. We found that the invasion of HGFs with HGPg results in the activation of the signalling pathways mediated by CXCR4. Inhibition of CXCR4 activity may be an effective way to control P. gingivalis periodontal infection. HGPg also induces the activation of the p38/Akt and NF-κB/Sp1 signalling pathways and ultimately enhances ICAM-1 expression in HGFs. These findings provide insights into the mechanisms underlying the interplay between hyperglycaemic P. gingivalis with HGFs in modulating HGF signalling and gene expression, which may be involved in the development of complications in patients with diabetes. Our data indicate potential relevant clues regarding mechanisms, for future therapeutic interventions.

Experimental procedures


All culture materials were purchased from Gibco (Grand Island, NY). PD98059 (ERK inhibitor), SP600125 (JNK inhibitor), SB203580 (p38 inhibitor) and LY294002 (PI3K/Akt inhibitor) were purchased from Calbiochem (La Jolla, CA). Mouse monoclonal antibodies (mAB) against phosphor-Akt and Akt were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit polyclonal antibodies against phosphor-p38 and p38 were purchased from Cell Signalling Technology (Beverly, MA). Akt, p38, NF-κB p65, Sp1 shRNA and control shRNA (scrambled negative control containing random DNA sequences) were purchased from the National RNAi Core Facility in Academic Sinica, Taipei, Taiwan. The CXCR4-siRNA and control siRNA (scrambled negative control containing random DNA sequences) were purchased from Invitrogen (Carlsbad, CA). Human insulin and other chemicals of reagent grade were obtained from Sigma (St Louis, MO).

Bacterial strains and growth conditions

Porphyromonas gingivalis strain CCUG25226 was obtained from the Bioresources Collection and Research Center (BCRC) of the Food Industry Research and Development Institute (Hsinchu, Taiwan). The bacteria were grown anaerobically at 37°C in brain-heart infusion (BHI) broth supplemented with 0.5% yeast extract, haemin (5 mg l−1) and vitamin K3 (1 mg l−1). Incubation of P. gingivalis was monitored by measuring the optical density at 660 nm.

For studying the effect of glucose on P. gingivalis, and the further effect of HGPg on HGF ICAM-1 expression, P. gingivalis was cultured in BHI medium containing 5 (NG) or 25 (HG) mmol l−1 glucose, or 25 mmol l−1 mannitol. Then P. gingivalis was collected and defined as NGPg, MPg and HGPg. For the infection assay, P. gingivalis with different cultured conditions at the indicated multiplicity of infection (moi) was added according to different treatments of HGFs.

Transmission electron microscopy

Porphyromonas gingivalis cultured under HG and NG conditions was grown to the late logarithmic phase and placed on a high-resolution carbon substrate, followed by negative staining with 1% ammonium molybdate for 1 min. The samples were examined with a transmission electron microscope (JOEL JEM-2100).

Cell culture

Human gingival fibroblasts were purchased from ScienCell Research Laboratories (San Diego, CA). Cells were cultured in DMEM supplemented with 10% fetal bovine serum (FBS), penicillin (100 U ml−1) and streptomycin (100 μg ml−1), and were incubated at 37°C in a humidified atmosphere containing 5% CO2. HGFs from passage levels 4–6 were used in this study.

Invasion assays

Human gingival fibroblasts were infected with an moi of 20 or 50 bacteria per host cell. After 1 h incubation at 37°C, cells were washed twice with PBS and then incubated for another 2 h in medium containing gentamicin and metronidazole to kill extracellular bacteria. Cells were then washed with PBS, lysed in 1 ml of 0.1% Triton X-100, and plated on blood agar plates. Bacteria present in these lysates, representing the number of bacteria present intracellularly, were titred. Invasion frequencies were calculated as the number of bacteria surviving divided by the total number of bacteria present (Tsai et al., 2009).

PCR identification of fimA genotype

PCR was performed to detect P. gingivalis strain CCUG25226 using specific primers for 16S rRNA; then to identify its genotypes, specific primers for fimA were used, as described previously (Wang et al., 2010). Positive and negative controls were included in each PCR set. The PCR products were analysed by electrophoresis in a 2% agarose gel.

Real-time quantitative PCR

For detecting HGF ICAM-1 expression, and P. gingivalis fimA mRNA expression, real-time PCR was performed, and products were detected using an ABI Prism 7900HT with the FastStart DNA SYBR Green I kit (Roche Diagnostics GMbH, Mannheim, Germany). The designed primers in this study were ICAM-1 forward primer, 5′-GTGAC ATGCA GCACC TCCTG-3′; ICAM-1 reverse primer, 5′-TCCAT GGTGA TCTCT CCTCA-3′; 18S rRNA forward primer, 5′-CGGCG ACGAC CCATT CGAAC-3′, 18S rRNA reverse primer, 5′-GAATC GAACC CTGAT TCCCC GTC-3′; fimA forward primer, 5′-CAGCA GGAAG CCATC AAATC-3′; fimA reverse primer, 5′-CAGTC AGTTC AGTTG TCAAT-3′; 16S rRNA forward primer, 5′-TGTAG ATGAC TGATG GTGAA A-3′; and 16S rRNA reverse primer, 5′-ACTGT TAGCA ACTAC CGATG T-3′. Quantification was performed using the 2−ΔΔCt method (Chen et al., 2011a). All samples were measured in duplicate. The average value of both duplicates was used as the quantitative value.

ELISA for cell surface ICAM-1 expression

ICAM-1 expression on the HGF surface was measured by cell surface ELISA as described previously (Chen et al., 2011b). Briefly, HGFs cultured in 96-well plates were fixed by 4% paraformaldehyde. Cell surface ICAM-1 expression was assessed using the mouse antihuman ICAM-1 mAb followed by a horseradish-peroxidase-conjugated secondary antibody. The absorbance of each well was measured at 490 nm after the reactions were stopped.

Western blot analysis

Cells were lysed with a buffer containing 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, and a protease inhibitor mixture (PMSF, aprotinin and sodium orthovanadate). The total cell lysate (50 μg of protein) was separated by SDS-polyacrylamide gel electrophoresis (PAGE) (12% running, 4% stacking) and analysed using the designated antibodies and the Western-Light chemiluminescent detection system (Bio-Rad, Hercules, CA).

shRNA and siRNA transfection

For shRNA transfection, HGFs were transfected with the designated shRNA plasmids by using shRNA plasmid transfection reagent (Santa Cruz). For siRNA transfection, HGFs were transfected with the specific CXCR4-siRNA or control siRNA by using an RNAiMAX transfection kit (Invitrogen, Carlsbad, CA).

Luciferase assays

Human ICAM-1 promoter constructs containing −550/+16, −413/+16, −323/+16, −238/+16 and −99/+16 of ICAM-1 5′-flanking DNA linked to the firefly luciferase reporter gene of plasmid pGL4 (Promega, Madison, WI) were used. DNA plasmids at a concentration of 1 mg ml−1 were transfected into HGFs by lipofectamine (Gibco). The pSV-β-galactosidase plasmid was cotransfected to normalize the transfection efficiency. Values obtained were normalized to the levels of β-galactosidase in the cell lysates. β-Galactosidase activities were determined with an assay kit and exhibited < 20% variation between samples.

Chromatin immunoprecipitation (ChIP)

ChIP assays were performed following a published protocol (Sung et al., 2009). Briefly, chromatins were sheared by sonication (3 times, 10 s on, 60 s off). Precleared extracts were immunoprecipitated with rabbit anti-p65 and c-Jun antibodies or rabbit IgG at 4°C overnight. DNA was isolated from precipitated complexes and analysed by PCR. The human ICAM-1 promoter region −305/−91 was amplified with the PCR primer pairs 5′-ACCTT AGCGC GGTGT AGACC-3′ and 5′-CTCCG GAACA AATGC TGC-3′.

NF-κB p65 transcription factor assay (TF ELISA assay)

Nuclear extracts of cells were prepared as previously described (Sung et al., 2009). Equal amounts of nuclear extracts were used for quantitative measurements of NF-κB p65 activation using commercially available ELISA kits (Panomics, Redwood City, CA) that measure NF-κB p65 binding activities.

Flow cytometry analysis

The expressions of CXCR4 on the surface of HGFs were measured by indirect immunofluorescence using flow cytometry and the mouse monoclonal antibody (20 μg ml−1; R&D) against CXCR4.

Neutrophil adherence experiment

Human neutrophils were isolated as described previously (Chen et al., 2006) and maintained in culture medium RPMI 1640 supplemented with 10% FBS. Before the adhesion experiments, neutrophils (5 × 106 cells ml−1) were labelled with 5 μM calcein acetoxymethyl (Molecular Probes, Eugene, OR) in RPMI-1640 medium for 30 min at 37°C. The labelled neutrophil cells (1 × 105 cells ml−1) were added to NGPg- or HGPg-infected HGFs and incubated for 30 min. In parallel experiments, HGFs were treated with isotype-matched IgG or neutralizing antibodies against ICAM-1 or CXCR4 (20 μg ml−1) during HGPg infection. Non-adherent cells were removed by washing with RPMI. The adherent neutrophils were evaluated using a cuvette fluorometer at 494 nm excitation and 510 nm emission. Serial dilutions of calcein-labelled neutrophils were used as a standard to calculate the number of adherent neutrophils.

Subjects and collection of gingival tissue samples

The Ethics Committees of Chang Gung Memorial Hospital approved the study protocol, and written informed consent was obtained from all patients before enrolment. The study group consisted of 10 patients with both periodontal disease and type 2 diabetes, and 10 control subjects with periodontal disease alone. The diabetic patients had a mean (± SEM) age of 47.6 ± 5.7 years, a body mass index of 29.6 ± 3.1 kg m−2, fasting glucose of 11.7 ± 1.1 mmol l−1 and haemoglobin A1C of 7.4 ± 1.3%. In addition, the control subject group had a mean (± SEM) age of 46.3 ± 6.4 years, a body mass index of 25.8 ± 1.6 kg m−2, fasting glucose of 4.7 ± 0.1 mmol l−1 and haemoglobin A1C of 4.4 ± 0.6%. None of the control subjects had infectious or inflammatory conditions or cardiac, renal, or pulmonary decompensated diseases. Patients who smoked cigarettes or who used alcohol or medications (hormonal replacement therapy, non-steroidal anti-inflammatory drugs, corticosteroids and anticoagulant drugs) were excluded from this study. Subjects had not been treated for periodontitis over the previous 2 years and had taken no antibiotics in the 6 months preceding surgery. Gingival samples were obtained from subjects undergoing surgery to treat periodontitis.

Immunohistochemical analysis of ICAM-1 expression

Gingival tissue samples were frozen immediately after surgery in liquid nitrogen and stored at −80°C. Serial paraffin sections of biopsies were cut 5 μm thick, and immunohistochemistry was performed using the immunoperoxidase procedure. The tissue sections were incubated with 2% normal goat serum and 5% non-fat dry milk for 20 min to block nonspecific binding. Subsequently, the sections were incubated overnight at 4°C with monoclonal antibody against ICAM-1. Sections were then incubated with biotinylated goat anti-mouse antibody for 30 min, followed by incubation with avidin–biotin–peroxidase complex for 30 min with the addition of 3,3′-diaminobenzidine tetrahydrochloride for 5 min. Between the reaction steps, the sections were each washed with PBS three times for 5 min. Counterstaining was performed with haematoxylin. Sections stained with normal mouse IgG as primary antibody were used as a negative control.

Statistical analysis

The results are expressed as mean ± standard error of the mean (SEM). Statistical analysis was determined by using an independent Student's t-test for two groups of data and analysis of variance (anova) followed by Scheffe's test for multiple comparisons. P-values less than 0.05 were considered significant.


This study was supported by grants CLRPG8C0091, CMRPF6A0072, CMRPF6C0031 and EZRPF6C0011 from Chang Gung Memorial Hospital-Kaohsiung Medical Center, Chang Gung Memorial Hospital, and Chang Gung University of Science and Technology, Chia-Yi Campus, Taiwan; and by the National Science Council, Taiwan (NSC101-2622-B-255-001-CC3 and NSC 101-2320-B-415-003-MY3). We thank Dr Cheng-Liang Huang (Department of Applied Chemistry, National Chiayi University, Taiwan) for TEM sample preparation.

Competing interests

The authors declare they have no conflict of interest.