Angiopoietin-like protein 2 regulates Porphyromonas gingivalis lipopolysaccharide-induced inflammatory response in human gingival epithelial cells

Angiopoietin-like protein 2 (ANGPTL2) maintains tissue homeostasis by inducing inflammation and angiogenesis. It is produced in infiltrating immune cells or resident cells, such as adipocytes, vascular endothelial cells, and tumor cells. We hypothesized that ANGPTL2 might play an important role as a unique mediator in both systemic and periodontal disease. We demonstrated an increased ANGPTL2 concentration in gingival crevicular fluid from chronic periodontitis patients. Porphyromonas gingivalis lipopolysaccharide (LPS) treatment strongly induced ANGPTL2 mRNA and protein levels in Ca9-22 human gingival epithelial cells. Recombinant human ANGPTL2 increased interleukin 1β (IL-1β), IL-8, and tumor necrosis factor-α (TNF-α) mRNA and protein levels in Ca9-22 cells. Small-interfering (si)RNA-mediated ANGPTL2 knockdown in Ca9-22 cells reduced IL-1β, IL-8 and TNF-α mRNA and protein levels compared with control siRNA (p<0.01) in P. gingivalis LPS-stimulated Ca9-22 cells. Antibodies against integrin α5β1, an ANGPTL receptor, blocked induction of these inflammatory cytokines in P. gingivalis LPS-treated Ca9-22 cells, suggesting that secreted ANGPTL induces inflammatory cytokines in gingival epithelial cells via an autocrine loop. The classic sequential cascade of P. gingivalis LPS → inflammatory cytokine induction is well established. However, in the current study, we reveal a novel cascade comprising sequential P. gingivalis LPS → ANGPTL2 → integrin α5β1 → inflammatory cytokine induction, which might be responsible for inducing potent periodontal disorganization activity in gingival epithelial cells. Via this pathway, ANGPTL2 functions in the pathogenesis of periodontitis and contributes to prolonging chronic inflammation in patients with systemic disease.


Introduction
Periodontitis is a chronic inflammatory disease caused by periodontal pathogens. In this disease, the inflammatory response is the major contributor to structural component damage in the periodontium [1]. Gingival epithelial cells are the initial barriers to oral microbial intrusion, and their modulatory function in the inflammatory response in periodontal diseases has recently been highlighted [2]. Porphyromonas gingivalis, a gram-negative anaerobic bacterium, has been identified as one of the most important periodontal pathogens. P. gingivalis induces the release of large numbers of outer membrane vesicles containing lipopolysaccharide (LPS), which can penetrate periodontal tissue [3,4]. P. gingivalis LPS is reported to induce an inflammatory response in various cell types (such as macrophages, fibroblasts, endothelial cells, and gingival epithelial cells) via activating Toll-like receptor 2 (TLR2) or TLR4. These combined inflammatory responses produce imbalance between osteoblast and osteoclast numbers, resulting in alveolar bone resorption [5]. Periodontal disease is influenced by both systemic and environmental factors such as age, smoking, diabetes mellitus, and overweight or obesity [6]. Clinical studies have revealed an association between periodontal disease and systemic diseases such as diabetes, cardiovascular disease, metabolic syndrome, and cancer [7][8][9][10][11][12][13]. There is also increasing evidence that periodontal disease is an independent risk-factor for several systemic diseases, which is changing our understanding about the causality and directionality of associated oral and systemic diseases [14].
The angiopoietin-like protein (ANGPTL) family of eight secreted glycoproteins was recently identified [15,16]. ANGPTLs do not bind to the Tie2 angiopoietin receptor or to the related protein Tie1, and are classified as orphan ligands; thus, they appear to have distinct biological functions from angiopoietins [17]. ANGPTL2 maintains tissue homeostasis by inducing inflammation and angiogenesis [17,18]. In some contexts, ANGPTL2 overexpression promotes irreversible pathological tissue remodeling via integrin α5β1 and the development or progression of metabolic diseases, type 2 diabetes, atherosclerotic vascular disease, and some cancers caused by chronic inflammation [16,[19][20][21][22]. ANGPTL2 expression in increased in infiltrating immune cells or resident cells, such as adipocytes, vascular endothelial cells, and tumor cells [19,21,22]. It is therefore likely that enhanced ANGPTL2 autocrine or paracrine signaling accelerates disease development and progression. However, no studies have yet investigated a role for ANGPTL2 in the pathogenesis of inflammatory disease, including periodontal disease. Thus, further studies are required to understand its intracellular function.
We tested the hypothesis that ANGPTL2 is a unique mediator in periodontal disease by conducting a series of in vivo and in vitro experiments. We found increased ANGPTL2 concentrations in gingival crevicular fluid (GCF) samples from chronic periodontitis (CP) patients.

Participants
CP patients (at least six teeth containing sites with a probing depth [PD] of !5 mm, a clinical attachment level [CAL] of !6 mm, and extensive bone loss by radiography) and healthy adult volunteers were recruited from the outpatient clinic of Aichi Gakuin University Dental Hospital, Nagoya, Japan. Exclusion criteria were: (1) smoking within the past 5 years; (2) antibiotic therapy in the last 6 months; (3) pregnancy; and (4) any systemic condition that could affect periodontitis progression (e.g. immunologic disorders, diabetes, or osteoporosis). Healthy volunteers had no signs of clinical periodontal attachment loss, a PD of 3 mm, percentage of bleeding on probing (full mouth) of <10%, and maximal bone loss (full mouth) of <10%. A single examiner recorded PD, CAL, and demographic data for all participants. A total of 46 participants comprising 24 CP patients (14 men and 10 women, mean age 55.4 ± 14.4 years) and 22 healthy volunteers (13 men and 9 women, mean age 44.2 ± 18.5 years) were included in the study.
Informed written consent was obtained from all participants at the first outpatient visit, and the study protocol for human participants was reviewed and approved by Aichi Gakuin University, School of Dentistry, Ethics Committee (permit number AGUD 383).

GCF sampling
GCF samples were collected from three buccal sites (mesial, center, and distal) of teeth with a single root, a PD of 3 mm (for all three sites; healthy individuals) or !4 mm (at least one site; CP patients), and prominent radiographic evidence of alveolar bone loss (CP patients only), based on a previously described method [23]. GCF samples were collected from three sites identified in the clinical examination by inserting paper strips (Oraflow, Hewlett, NY, USA) into the gingival crevice until mild resistance was felt and removed after 30 s. The volumes of GCF samples were quantified using a Periotron 8000 (Oraflow) that had been calibrated with pooled human serum. ANGPTL2 protein levels in these samples were measured (described below).

Pretreatment with anti-human integrin α5β1 monoclonal antibody
Before stimulation with P. gingivalis LPS or rhANGPTL2, Ca9-22 cells were incubated for 1 h with 1 ng/ml anti-human integrin α5β1 monoclonal antibody (Millipore, Darmstadt, Germany, Cat: MAB1969). We confirmed that 1 ng/ml antibody was sufficient to block integrin α5β1 (data not shown).

Small-interfering RNA assays
At 24 h after seeding into six-well plates, Ca9-22 cells were transfected with negative control or Silencer Select Predesigned small-interfering (si)RNA targeting TLR2, TLR4, or ANGPTL2 (Life Technology, Tokyo, Japan) using Lipofectamine 3000 (Life Technology), according to the manufacturer's instructions. After incubation for 48 h, cells were primed with rhIFN-γ for 12 h, washed three times with PBS, and treated with 30 ng/ml P. gingivalis LPS for 24 h.

Statistical analysis
Statistical analyses were performed using SPSS for Windows, version 15.0. (Chicago, IL, USA). Differences among groups were examined by one-factor analysis of variance (ANOVA); Student's t-test was used for comparisons. Data are expressed as means ± standard deviation (SD). Statistical significance was set at p<0.05.

ANGPTL2 levels are increased in GCF from CP patients
To examine whether ANGPTL2 protein levels are altered in CP patients, ANGPTL2 protein production was measured in GCF samples from 22 healthy participants and 24 CP patients. ANGPTL2 levels in healthy participants were relatively low (below the limit of detection in all but four), but were significantly higher in CP patients (p<0.05; Fig 1).

P. gingivalis LPS stimulates ANGPTL2 in Ca9-22 cells and HGECs
To investigate the relationship between altered ANGPTL2 mRNA and protein levels and CP, further in vitro experiments were performed using the Ca9-22 gingival epithelial cell line. ANGPTL2 mRNA expression was measured in primed Ca9-22 cells stimulated with 0 (control), 10, 30, 100, 300, or 1000 ng/ml P. gingivalis LPS for 24 h. ANGPTL2 mRNA induction was maximal in cells treated with 30 ng/ml (Fig 2A). At 4,8,12, and 24 h after treatment with 30 ng/ml P. gingivalis LPS, ANGPTL2 mRNA levels were significantly higher in primed Ca9-22 cells than in control cells (Fig 2B). The ANGPTL2 mRNA level was also significantly higher in primed HGECs than in control cells (Fig 2C).
Western blot analysis showed that ANGPTL2 (60 kDa) protein expression was also higher in primed Ca9-22 cells stimulated with P. gingivalis LPS compared with unstimulated controls (Fig 2D), indicating that P. gingivalis LPS induces ANGPTL2 protein. In addition, ANGPTL2 protein levels in culture supernatants from Ca9-22 cells and HGECs were increased at 24 and 48 h after stimulation with 30 ng/ml P. gingivalis LPS (Fig 2E and 2F). Because a similar trend was observed in the response of Ca9-22 cells and HGECs to P. gingivalis LPS stimulation, we mainly used Ca9-22 cells in the subsequent experiments.

TLR2 and TLR4 siRNA reduce ANGPTL2 levels in Ca9-22 cells
Since P. gingivalis LPS is a TLR2 and TLR4 agonist, we knocked down TLR2 and TLR4 with siRNA in Ca9-22 cells. TLR2 and TLR4 expression was completely suppressed after siRNA treatment (Fig 3A and 3B). Both TLR2 and TLR4 siRNA strongly reduced ANGPTL2 mRNA levels compared with control siRNA (Fig 3C and 3D). Western blot analysis confirmed that ANGPTL2 (60 kDa) protein levels were also decreased by TLR2 and TLR4 siRNA compared with control siRNA (Fig 3E and 3F). These data indicate that P. gingivalis LPS induction of ANGPTL2 is mediated by TLR2 and TLR4.
We next confirmed that integrin α5β1 (the ANGPTL2 receptor) is expressed in Ca9-22 cells (data not shown). Pretreatment with anti-integrin α5β1 IgG completely suppressed the induction of inflammatory cytokines IL-1β, IL-8 and TNF-α in P. gingivalis LPS-stimulated Ca9-22 cells, confirming that interaction between ANGPTL2 and integrin α5β1 is needed to stimulate inflammatory cytokine production (Fig 4G-4I).

Discussion
Various substances involved in the progression or activity of periodontal disease, including cytokines (ILs, TNF-α, receptor activator of nuclear factor-κB ligand, and osteoprotegerin) and lipids (prostaglandin E2 and leukotriene), have been detected in GCF [29,30], and some have been verified as markers of periodontal disease [31,32]. To date, functional data on the pathophysiological roles of ANGPTL2 in periodontal disease are scarce. In this study, we demonstrated for the first time that ANGPTL2 levels are significantly increased in GCF from CP patients.
ANGPTL2 induction by Escherichia coli LPS was previously reported in human retinal pigment epithelial cells [33]. We similarly showed a significant increase in ANGPTL2 mRNA and protein levels after P. gingivalis LPS stimulation in cultured cells. Since P. gingivalis is one of the most important periodontal pathogens, it is possible that ANGPTL2 is produced from gingival epithelial cells during periodontitis progression. The presence of an ANGPTL2 signaling cascade in periodontal tissue (and especially in gingival epithelial cells) was previously unknown. TLR proteins are expressed on the surface of periodontal tissue component cells such as gingival epithelial cells and fibroblasts. Previous reports indicate that TLR2 and TLR4 are specifically expressed on periodontal cells [34] and involved in the immune response of (D-F) Protein levels of inflammatory cytokines were measured in Ca9-22 cells after stimulation with 10 ng/ml rhANGPTL2 for 3,6,9 and 12 h. Differences among groups were analyzed by one-way ANOVA. Data are expressed as means ± SD (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001 vs. control. (G-I) Protein levels of inflammatory cytokines after anti-integrin α5β1 IgG pretreatment followed by stimulation with 10 ng/ml rhANGPTL2 for 9 h. Data are expressed as means ± SD (n = 4). *p < 0.05, **p < 0.01 vs. control. ND: not detectable.
In vascular endothelial cells or monocytes, ANGPTL2 binding to its receptor, integrin α5β1, induces the production of various inflammatory cytokines. Therefore, increased ANGPTL2 levels can prolong chronic inflammation [16]. We confirmed that integrin α5β1 is constitutively expressed in Ca9-22 cells and observed that pretreatment with anti-integrin α5β1 IgG suppressed inflammatory cytokine induction in epithelial cells, indicating that ANGPTL2-induced overproduction of inflammatory cytokines is mediated by integrin α5β1. ANGPTL2 receptor consists of two forms, integrin α5β1 and LILRB2 [37]. However, we confirmed by flow cytometry that the alternative ANGPTL2 receptor, LILRB2, was not expressed on Ca9-22 cells (S1 Fig). In addition, the expression of LILRB2 was not enhanced upon stimulation with P.gingivalis LPS or rhANGPTL2. Therefore, we think that the alternative ANGPTL2 receptor, LILRB2, was not involved in the induction of proinflammatory cytokines in Ca9-22 cells.
In rheumatoid arthritis patients, the ANGPTL2 concentration is increased (to about 25 ng/ml) in synovial fluid and the synovium [38]. In addition, Tabata et al. reported that high concentrations (1-20 μg/ml) of rhANGPTL2 did not have a cytotoxic effect on human umbilical vein endothelial or arterial endothelial cells [19]. Thus, we used rhANGPTL2 concentrations in the order of ng/ml to stimulate Ca9-22 cells for clarifying the role of ANGPTL2 in gingival epithelial cells. However, the average ANGPTL2 concentration in GCF was even lower, at about 300 pg/ml (Fig 1). There is a therefore discrepancy between the ANGPTL2 concentration used in in vitro experiments and those detected in vivo. However, we hypothesize that ANGPTL2 accumulation at sites of inflammation results in high local ANGPTL2 concentrations.
A relationship between periodontal disease and various systemic diseases has been widely reported [7,8,10,12,13]. For example, it is generally accepted that a two-way relationship exists between periodontal disease and diabetes mellitus, and that TNF-α may mediate this association. The TNF-α level in GCF and serum is an inflammatory marker in CP and type 2 diabetes patients [39]. As ANGPTL2 is constitutively present in the blood and can act systemically [19], we speculate that ANGPTL2 expressed in the periodontium could have systemic effects as well as its role in irreversibly remodeling periodontal tissue. In the present study, ANGPTL2 was also detected in GCF (Fig 1). In addition, ANGPTL2 is constitutively expressed We identified a novel signaling cascade comprising sequential P. gingivalis LPS stimulation ! ANGPTL2 induction ! integrin α5β1 ! inflammatory cytokine induction, which results in potent periodontal disorganization activity in gingival epithelial cells. The bold line is generally considered as the main pathway mediating inflammatory cytokine production by periodontal pathogens such as P. gingivalis LPS. Non-bold line is shown as a novel signaling pathway via ANGPTL2 in P. gingivalis LPS stimulation. The dashed line is based on a hypothesis that ANGPTL2 is one of key mediator between periodontal disease and systemic disease.
https://doi.org/10.1371/journal.pone.0184825.g007 in adipocytes and secreted by lung cancer cells, prostate cancer cells, and peritoneal cells, where it has topical effects [22,40,41]. We demonstrated that the ANGPTL2 concentration is higher in GCF from CP patients compared with healthy controls; however, we were unable to investigate a possible correlation between ANGPTL2 concentrations in GCF and serum from patients with CP and/or systemic disease. Therefore, further studies are needed to determine whether ANGPTL2 links CP and systemic diseases.
A schematic diagram illustrating the possible involvement of ANGPTL2 in periodontal disease is shown in Fig 7. In periodontal disease, the bold line is generally considered the main pathway mediating inflammatory cytokine production by periodontal pathogens such as P. gingivalis LPS. In the current study, we discovered a novel pathway that acts via ANGPTL2. Interestingly, during periodontal disease progression, the pathway has an autocrine effect in which ANGPTL2 acts as an exacerbating factor. ANGPTL2 overexpression might therefore prolong chronic inflammation via an autocrine loop to promote periodontal tissue destruction. The inflammatory cytokine production by P. gingivalis LPS stimulation was partially suppressed by siRNA or neutralizing antibody against integrin α5β1 in gingival epithelial cells. Although we demonstrated that the key role of ANGPTL2 was inducing inflammatory cytokines upon P. gingivalis LPS stimulation in gingival epithelial cells, it is likely that ANGPTL2-independent signaling pathways for the production of cytokines exist in the cells. Further studies are therefore necessary to determine how ANGPTL2 functions as a key mediator in periodontal disease and systemic diseases.