IL-1β strengthens the physical barrier in gingival epithelial cells

ABSTRACT

Periodontitis is one of the most common oral diseases worldwide and is caused by a variety of interactions between oral bacteria and the host. Here, pathogens induce inflammatory host responses that cause the secretion of proinflammatory cytokines such as IL-1β, IL-6, and IL-8 by oral epithelial cells. In various systems, it has been shown that inflammation compromises physical barriers, which enables bacteria to invade the tissue. In this study, we investigated the barrier properties of the oral mucosa under physiological and inflamed conditions. For this purpose, we assessed the influence of IL-1β on the transepithelial electrical resistance and in particular on tight junctions in vitro in human stratified squamous epithelium models. Indirect immunofluorescence and western blot analyses were performed to investigate localization and expression of tight junction proteins in primary gingival cells, immortalized gingival cells and native gingiva. Furthermore, the TEER of gingival keratinocytes was assessed. The results showed that IL-1β led to strengthening of the gingival keratinocyte barrier. This was demonstrated by an increase in TEER, the upregulation of TJ proteins, and an increase in the formation of TJ strands. The IL-1β-mediated upregulation of occludin was prevented by the NF-κB inhibitor BAY 11–7085. These observations provide insights into host responses in the early stages of periodontal disease and offer information about TJ formation in human gingival epithelial cells under physiological and inflammatory conditions. Comprehensive knowledge of the physical barrier during inflammation may help in developing strategies to effectively target the inflammatory barrier to improve the bioavailability of drugs for the treatment of periodontitis.

KEYWORDS:

• Gingiva
• oral mucosa
• tight junctions
• TEER
• IL1beta

Introduction

Periodontitis is a bacterially induced inflammatory disease of the oral mucosa¹ that extends to the periodontium and gradually destroys the periodontal fibers and alveolar bone.² Approximately 50% of the world’s population shows clear symptoms of mild or moderate forms of periodontitis and even 10–12% suffer from more severe forms of periodontitis.^3,4 If left untreated, periodontitis is associated with increased systemic inflammatory diseases, with potential for an increased risk of stroke, myocardial infarction, atherosclerosis, and diabetes mellitus.^5,6

In recent years, there has been a paradigm shift in which the role of bacteria has receded in favor of the inflammatory process and the host response. Periodontitis is, therefore, considered to be an inflammation that is initially induced by subgingival microorganisms that subsequently provoke an inflammatory host response, which is responsible for the tissue loss.⁷ As a first line of epithelial defense, oral mucosal epithelial cells secrete proinflammatory cytokines such as IL-1β, IL-6, and IL-8 [CXCL8].^8-10 Various studies have shown that IL-1β, in particular, is elevated in diseased gingival tissue and gingival crevicular fluid.^11,12 After activation and secretion, IL-1β induces a series of inflammatory reactions in cells of the periodontium, which contributes to the course of periodontitis.^13,14 Apart from this chemical barrier characterized by a number of soluble mediators, the gingiva also forms a physical barrier to resist mechanical forces and prevent the penetration of pathogens that may damage gingival cells. The physical barrier consists primarily of the stratum corneum, the outermost layer of the stratified squamous epithelium, and the underlying tight junctions (TJs).¹⁵ TJs are intercellular junctions that seal adjacent keratinocytes.¹⁵ Under physiological conditions, TJs act as a semipermeable barrier for the paracellular transport of ions, solutes, and water^16-19 and maintain tissue homeostasis.^20,21

In contrast to simple epithelia, the skin and gingiva consist of several layers whose cells undergo different stages of differentiation. Some TJ proteins appear to be expressed primarily in the stratum granulosum (SG), whereas individual TJ proteins are expressed beyond the SG).^22,23,24,25

Although TJs vary in composition, the tetraspan transmembrane proteins occludin (OCLN) and members of the claudin (CLDN) family have been identified in all stratified mammalian epithelia.²⁶ The cytoplasmic parts of these transmembrane proteins are associated with a dense plaque structure containing proteins such as ZO-1.^26,27 Knock-out experiments revealed that CLDN-1 is essential for sealing the epidermis.¹⁷ Since healthy gingival tissue can only be examined in a limited way, there are little data available to describe the healthy human situation in the gingiva. The expression of claudin-1, −2, and occludin has been shown in primary and immortalized gingival keratinocytes.^28,29,30,31 Gene expression analyses of multilayer gingival epithelial cell cultures showed a strong expression of claudin-1, −4, −12, −17, −25, JAM-A, and occludin.³² At the protein level, claudin-1, −4, −5, and occludin were found to be expressed in primary and immortalized human gingival keratinocytes.^29,33

It has been shown that TJs are influenced by cytokines in various tissues.³⁴ In the present study, we addressed the question of whether pro-inflammatory mediators affect the physical barrier of the gingival epithelium. We therefore investigated the influence of the periodontitis-relevant cytokine IL-1β on TJs in in vitro models derived from primary and immortalized gingival keratinocytes.

Materials and methods

Cells and cell culture

For experiments, hTERT-immortalized human gingival keratinocytes (OKG4; kindly provided by Susan Gibbs, Amsterdam) were used.^35,36 Primary gingival epithelial cells (GECs) were dissected from gingival tissue samples obtained from tooth extractions and kindly provided by the Department of Oral and Maxillofacial Surgery, Charité. The study was approved by the ethical committee of the Charité – Universitätsmedizin Berlin (Number of the request: EA2/185/16). All volunteers were informed before sampling of the tissues and gave written informed consent. All experiments followed the guidelines of good clinical/laboratory practice (GCP/GLP) and the WHO Declaration of Helsinki 1964, latest update Fortaleza 2013 (64th WMA General Assembly, Brazil, October 2013).

Epithelial cells were seeded on collagen A- (Biochrom, Berlin, Germany;  for GEC) or collagen  IV-coated (Sigma-Aldrich, St. Louis, USA; for OKG4) cell culture dishes (Corning, Corning, USA) and grown at 37°C and 5% CO₂ in DermaLife K  Serum-Free  Keratinocyte  Growth  Medium that contained 60 µM  Ca²⁺ (Lifeline® Cell Technology, Frederick, USA), supplemented with 1% Penicillin-Streptomycin (PAN Biotech, Aidenbach, Germany). GECs additionally received 1% Amphotericin B (Biochrom, Berlin, Germany).

Cell differentiation was induced by increasing the Ca²⁺  level in the medium from 60 µM  Ca²⁺ (low Ca²⁺) to 1.4 mM (high Ca²⁺). To achieve this, the medium was exchanged for DermaLife K  Serum-Free  Keratinocyte  Culture  Medium that contained 1.4 mM Ca²⁺. Inflammation of OKG4 and primary gingival cells was induced by administration of 10 ng/ml IL-1β (Immunotools, Friesoythe, Germany) to the cell culture medium. The medium containing IL-1β and the control medium were changed every day. Unstimulated cells served as controls.

To investigate the influence of the NF-κB pathway on occludin expression, cells were pretreated with 10 µM BAY 11–7085 (Sigma-Aldrich, St. Louis, USA) for 1 h.

Human cytokine antibody array

To perform the human cytokine antibody array, which allows the simultaneous detection of 42 different cytokines (Abcam, Cambridge, UK), 2.5 x 10⁵ OKG4 cells per well were seeded into 6-well dishes (Corning, Corning, USA). After the Ca²⁺ increase, the cells were cultured for 24 h and then stimulated using IL-1β. After 72 h, the medium was collected and stored at −20°C until use. The array was carried out according to the manufacturer’s instructions. A 1 ml volume of culture medium was added to the membrane. Signals were visualized using a VersaDoc^TM 4000 MP and QuantityOne® 4.6.5 software (BioRad Laboratories, Hercules, USA). For signal quantification, ImageJ’s Protein Array Analyzer plug-in was used (Wayne Rasband, National Institutes of Health, USA, version 1.52a). Signal intensities below 10% of the positive control were excluded from evaluation to avoid false-positive results.

ELISA detecting IL-6 and IL-8

Sandwich ELISAs (Immunotools, Friesoythe, Germany) were performed to quantify cytokine concentrations. The cell culture supernatants were collected as previously described. In brief, medium from 6-well plates (Corning, Corning, USA) was removed from the culture after 6, 24, or 72 h of stimulation time and stored at −20°C until use. Samples were subjected to ELISA. The assay was performed in accordance with the manufacturer’s instructions using 100 µl of culture medium per well.

MTT assay

Cells (1x10⁴) were seeded in 100 µl growth medium per well in 96-well plates. Wells without cells (blank), unstimulated cells and cells lysed with 4% Triton X-100 before performing the MTT assay served as controls. The stimulation was performed by changing the medium the following day. The MTT assay was performed 24 h, 48 h, and 72 h after stimulation. After 4 h of incubation with 10 µl MTT solution (Thiazolyl Blue Tetrazolium Bromide, Sigma-Aldrich, St. Louis, USA) per well, the supernatant was aspirated. By applying 100 µl 5% 1 N HCl (v/v) in isopropanol (Thermo Fisher, Waltham, USA) per well, the reduction of the dye to violet-colored formazan was achieved. The colorimetric reaction was measured photometrically at 570 nm and subsequently quantified.

SRB assay

The cells were seeded, cultured, and stimulated analogously to the MTT assay. After stimulation, 100 µl of 10% TCA (Carl Roth, Karlsruhe, Germany) per well was added and incubated at 4°C for 1 h. The cells were incubated with 100 µl SRB solution (0.057% sulforhodamine B) (m/v; Sigma-Aldrich, St. Louis, USA) in 1% acetic acid per well at room temperature. The cells were repeatedly washed using 200 µl 1% acetic acid and were dried at room temperature. Tris solution (10 mM; pH 10.5; Carl Roth, Karlsruhe, Germany) was added (200 µl per well) and incubated for 15 minutes at room temperature. The cell density was monitored using a microplate reader at 492 nm. Three independent experiments were performed in triplicate.

Measurement of the transepithelial electrical resistance (TEER)

Gingival keratinocytes (2.7 x 10⁴) were cultivated on collagen-coated polycarbonate filters (12 mm, 0.4 µm; Corning, Corning, USA) in DermaLife cell culture medium that contained 60 µM Ca²⁺. When cell confluency was achieved, the cells were transferred to high Ca²⁺ medium containing 1.4 mM Ca²⁺. TEER values were measured using an EndOhm-12 chamber (World Precision Instruments, Sarasota, USA) in combination with a Millicell ERS-2 volt-ohm meter (Merck Millipore, Burlington, USA). The measurement was performed after equilibration in the respective medium. To determine the influence of IL-1β on the course of the resistance, the TEER values were monitored after 24, 48, and 72 h in IL-1β-treated and untreated cells. The cells cultivated on filters in the presence of 60 µM Ca²⁺ were kept as negative controls. TEER values were calculated by background (blank filters) subtraction and multiplication of the surface area.

Western blot analysis

OKG4 cells (2.5 x 10⁵) were placed in 60 mm-culture dishes (Corning, Corning, USA) in DermaLife, differentiated as described above and stimulated with IL-1β for 24 and 72 h. For Western blotting, cells were washed with PBS and then lysed in lysis buffer (50 mM β-glycerophosphate pH 7.6, 1.5 mM EGTA, 1.0 mM EDTA, 1% (v/v) Triton X-100, 0.2% (v/v) protease inhibitor cocktail (PIC; Serva, Heidelberg, Germany), 0.4% (v/v) PMSF, 100 mM natrium vanadate, 500 mM NaF). Cell supernatants were denatured by boiling in Laemmli’s buffer for 5 min at 95°C. Then, 60 μg of protein was separated under reducing conditions by 10% SDS-PAGE and blotted onto nitrocellulose membranes (Thermo Fisher, Waltham, USA) for 1 h at 0.25 A. The membranes were subsequently blocked in 5% BSA/TBST and then incubated with the respective primary antibodies in blocking buffer overnight at 4°C. Primary antibodies and respective working concentrations are listed in Table A1. After incubation with appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies, signals were detected using Western Lightning® Plus ECL (PerkinElmer, Waltham, USA) chemiluminescence reagents and a VersaDoc^TM 4000 MP system with QuantityOne® software. Signals were quantified using ImageJ and subsequently normalized to the respective loading control.

Histology and immunofluorescence analysis

For immunostaining, immortalized and primary gingival epithelial cells were seeded at 1.2 x 10⁴ cells per well on collagen-coated 8-well Nunc^TM Lab-Tek II Chamber Slides^TM (Thermo Fisher, Waltham, USA) and treated as described above. Furthermore, cryo-sections of human gingiva with a thickness of 7 μm were subjected to immunofluorescence analysis.

Cells and sections were fixed for 15 min at RT with 4% paraformaldehyde (m/v)  in PBS^+/+ (1 mM Ca²⁺, 0.5 mM Mg²⁺), pH 7.4. After blocking, the cells were incubated with primary antibodies overnight at 4°C (Appendix Table 1). After extensive washing with PBS^+/+, an appropriate secondary antibody (Thermo Fisher, Waltham, USA) was added for 1 h at RT. Cell nuclei were stained with Hoechst 33342 dye (Sigma-Aldrich, St. Louis, USA). Cover slips were mounted using Fluoromount^TM Mounting Medium (Sigma-Aldrich, St. Louis, USA). For histological evaluation, sections were stained with Mayer’s hematoxylin and eosin (HE; Carl Roth, Karlsruhe, Germany).

Images were captured with an LSM700MAT confocal laser scanning microscope (Zeiss, Oberkochen, Germany) or an Axiovert 200 fluorescence microscope (Zeiss, Oberkochen, Germany) and processed with ZEN software (Zeiss, Oberkochen, Germany) or AxioVision Release 4.8.2 SP3 (Zeiss, Oberkochen, Germany). For comparison, images were acquired using identical settings for exposure and processing.

Statistical analysis

All experiments were performed at least three times. Data were displayed as the mean value (bars) and standard deviation (error bars) unless stated otherwise. Pairwise datasets were tested for significant differences using Student’s t-test. Significance was indicated by P-values and P < .05 was considered significant.

To compare IL-1β-treated and untreated cells, the starting points of stimulation were set to 100% due to different TEER starting values. Subsequently, the test groups were compared to the controls at the respective time point.

For Western blot analyses, the controls were set to 1, and the test groups were presented as relative values.

Results

IL-1β induces an inflammatory response in gingival epithelial cells

Since IL-1β has been identified as an essential component in the pathogenesis of periodontitis,^12,37,38 we investigated whether this proinflammatory cytokine is able to induce an inflammatory response in the immortalized gingival keratinocyte cell line OKG4 using a human cytokine antibody array. After 72 h of IL-1β stimulation, not only the amount of IL-1β but also the amount of the inflammation marker IL-8 was increased (Figure 1(a);  Figure A1). The concentration determined for IL-6, a classical mediator of the early inflammatory response, was very low in both experiments.

Figure 1. IL-1β induces the secretion of the proinflammatory cytokines IL-6 and IL-8 in gingival keratinocytes.

(a) The upper panel shows a scheme of the positions for putative cytokine binding of cytokine arrays. Below, representative cytokine arrays incubated with supernatants from control and IL-1β-treated cells are shown. Signals for IL-1β, IL-6, and IL-8 are marked with boxes. Three independent experiments were performed. Concentrations of IL-1β (p = .044) and IL-8 (p = .04) were significantly increased by IL-1β after 72 h. IL-6 (dotted line) was hardly detectable in both experiments. (b) Secretion of the proinflammatory cytokines IL-6 and IL-8 was further analyzed by ELISA after 6 h, 24 h, and 72 h of IL-1β treatment. In accordance with the array, IL-6 could hardly be detected after 72 h. At this time point, IL-8 was significantly increased after IL-1β-treatment (p < .001). After 6 h and 24 h, IL-1β significantly enhanced the secretion of IL-6 (p < .001 and p = .01, respectively). For IL-8, significantly enhanced secretion was observed as early as after 24 h (p = .003). Three independent assays were performed. Error bars represent the standard deviation, and * indicates significant differences between the test and control groups (* 0.01 ≤ p < .05; ** 0.001 ≤ p < .01; *** p < .001).

We verified the results of the cytokine array for IL-8 and investigated the influence of IL-1β on the secretion of IL-6 at earlier time points using the ELISA technique (Figure 1(b)). After 72 h of stimulation, the ELISA results confirmed the data obtained with the cytokine array for both cytokines. While IL-6 could hardly be detected in the supernatants of IL-1β-treated and untreated cells (IL-1β: 1.77 ± 1.37 pg/ml, control: 3.66 ± 1.24 pg/ml), IL-8 was significantly upregulated by IL-1β after 72 h (IL-1β: 330.99 ± 14.39 pg/ml, control: 118.12 ± 6.02 pg/ml).

We found IL-6 to be expressed earlier. After 6 h of stimulation, the secretion of IL-6 was highly significantly increased in IL-1β-treated cells (296.85 ± 21.84 pg/ml) compared to control cells (37.62 ± 4.32 pg/ml). We found this cytokine to be still expressed at significantly higher levels in IL-1β-treated cells (319.47 ± 48.87 pg/ml) compared to the control cells (26.86 ± 6.37 pg/ml) after 24 h.

Unexpectedly, the IL-8 concentration of untreated cells after 6 h was higher compared to untreated cells after 72 h. IL-1β slightly reduced the concentration of IL-8 (202.36 ± 27.91 pg/ml) compared to control cells (285.29 ± 11.47 pg/ml) after 6 h. After 24 h, IL-8 was significantly elevated in cells treated with IL-1β (423.73 ± 26.72 pg/ml) compared to control cells (259.55 ± 30.32 pg/ml).

Overall, IL-6 and IL-8 showed different secretion patterns in unstimulated cells. IL-1β caused a rapid upregulation of IL-6, which decreased over time, whereas the IL-1β-induced secretion of IL-8 was delayed but more stable over time. These data showed that IL-1β induced an inflammatory response in gingival epithelial cells.

Even though the number of proinflammatory mediators secreted upon IL-1β stimulation was limited, we were able to show that this cytokine leads to a persistent inflammatory response over a period of 72 h in gingival epithelial cells.

IL-1β does not affect proliferation or viability but enhances the physical barrier of gingival epithelial cells

We further investigated whether IL-1β causes cytotoxic effects or affects cell proliferation. We found that IL-1β did not have an impact on metabolic activity/cell viability nor on cell proliferation using the MTT and SRB assays, respectively (Figure 2(a,b)). Furthermore, nuclei of IL-1β-treated and untreated cells were stained with Hoechst dye. Fluorescence analysis and the counting of cell nuclei from randomly chosen visual fields of each experimental condition confirmed that IL-1β had no influence on cell proliferation (Figure 2(c)).

Figure 2. IL-1β does not influence the metabolic activity or proliferation of OKG4 cells.

(a) MTT assay; 1 × 10⁴ OKG4 cells per well were seeded in 96-well collagen-coated plates and stimulated with IL-1β for 24 h, 48 h, and 72 h. Subsequently, the MTT solution was applied, followed by an acidic isopropanol solution and readout of the converted dye quantity at 570 nm in the photometer. The error bars represent the standard deviation; the p-values for 24, 48, and 72 h are p = .269, 0.838, and 0.919, respectively. (b) SRB assay; cultivation and stimulation of cells were performed analogously to the MTT assay. The cells were then stained with SRB solution and solubilized in Tris solution, and the data were obtained photometrically at 492 nm; the error bars represent the standard deviation, and the p-values for 24, 48, and 72 h are p = .633, 0.916, and 0.19, respectively. (c) To determine the number of cells and the possible impact of IL-1β on cell proliferation, ten visual fields of Hoechst-stained cell nuclei of IL-1β-treated and untreated cells, respectively, were counted.

Moreover, the visualization of cell membranes of OKG4 cells by fluorescence-labeled lectin wheat germ agglutinin (WGA) revealed that IL-1β had no effect on cell morphology. This was also confirmed for primary oral keratinocytes (GECs) isolated from biopsies of human gingiva (Figure A2).

In the gingiva, epithelial cells form a physical barrier consisting of the stratum corneum and the TJs,¹⁵ which among other functions, hinder the penetration of pathogens. To assess whether the strength of the physical barrier is affected by IL-1β, we measured the transepithelial electrical resistance (TEER). The switch to a medium containing a high Ca²⁺ concentration of 1.4 mM led to a biphasic resistance curve over time in unstimulated GECs. Here, the cells exhibited a maximum of resistance of 706.50 ± 470.29  Ω x cm² at d  9 (Figure 3(a)). Subsequently, the TEER decreased but remained stable until d 14, with a mean value of 264.11 ± 39.85 Ω x cm². GECs cultured at a low Ca²⁺ concentration of 60 µM showed a consistent average TEER of 8.73 ± 2.54 Ω x cm² over 14 d.

Figure 3. IL-1β increases transepithelial resistance in a human 3D epithelial model.

(a) Development of TEER over time for GECs in response to an increase in the Ca²⁺ concentration from 60 µM to 1.4 mM (black line; n = 14) in comparison to cells cultivated in the presence of a low Ca²⁺ concentration (gray line; n = 3). GECs were obtained from three different donors. Three independent experiments were performed. Error bars represent the standard errors of the mean. (b) Representative HE-stained microsections of gingival epithelial 3D models derived from GECs and OKG4 cells. Cells were cultivated in the presence of 1.4 mM Ca²⁺ on Transwell® filters for 16 (GECs) or 15 d (OKG4 cells). Mucosal equivalents were then fixed and embedded in paraffin. Image processing was performed with AxioVision and Adobe Photoshop, scale bar, 50 µm. (c) The TEER of OKG4 cells (five independent experiments were performed; n = 7) and GECs (three independent experiments were performed; n = 7) was determined in the presence and absence of IL-1β. For measurements, confluent OKG4 cells and GECs were stimulated with 10 ng/ml IL-1β or were left untreated. To compare IL-1β-treated and untreated cells, the starting points of stimulation were set to 100%. Subsequently, the test groups were related to the controls at the respective time point. GECs from three different donors were used. Error bars represent the standard deviation, * indicates a significant difference between the IL-1β and control groups (OKG4 cells 72 h, p = .039).

The calcium switch induced differentiation in GECs and OKG4 cells, where the cells formed a multilayered tissue with signs of desquamation, as shown by HE-stained microsections (Figure  3(b)). Therefore, these constructs could be regarded as simplified 3D in vitro models of stratified squamous gingival epithelium.

Additional treatment of OKG4 cells with IL-1β for 24, 48 and 72 h led to an increase in resistance of between 20% and 30% compared to untreated cells. After 72 h, this change was found to be significant. A similar effect was observed in GECs, although the clear trend did not reach statistical significance (Figure  3(c)).

IL-1β increases TJ formation and the expression of TJ proteins in gingival epithelial cells

TJs are part of the physical barrier, and the existence of TJs and the expression of tight junction proteins in gingival epithelial cells have been described.^33,39,40 Nevertheless, the exact distribution of the numerous different tight junction proteins in the healthy human gingiva is unknown. Thus, we initially studied the localization of claudin-1 (CLDN1), zonula occludens-1 (ZO-1), and occludin (OCLN) in healthy gingiva using indirect immunofluorescence analysis (Figure 4). This study revealed that all three proteins were expressed in mature human gingiva. For CLDN1 and ZO-1, distinct staining of intercellular contacts in the stratum granulosum was observed. Additional signals in lower layers appeared more diffuse. Both proteins were excluded from the stratum corneum and the stratum basale. OCLN was mainly expressed in the stratum granulosum (Figure 4). Costaining with antibodies directed against OCLN and ZO-1 revealed signs of colocalization only in superficial parts of the stratum granulosum (Figure A3).

Figure 4. Expression and localization of CLDN1, ZO-1 and OCLN in healthy human gingiva.

Fresh human adult gingiva was embedded in tissue freezing medium and frozen in liquid nitrogen. Microsections (7 µm) were prepared, permeabilized, blocked, and incubated with antibodies against CLDN1, ZO-1, and OCLN and the respective secondary antibodies. Dashed lines represent the border between the epithelium and lamina propria. Cell nuclei were visualized using Hoechst 33342 dye. CLDN1, ZO-1, and OCLN were expressed in mature human gingiva. The expression of OCLN was restricted to the stratum granulosum, whereas ZO-1 and CLDN-1 could be additionally detected in the deeper layers. Image processing was accomplished using ZEN and ImageJ. Scale bar, 100 µm

Since IL-1β increased the TEER in gingival epithelial cells, we investigated whether this increase is associated with altered expression or localization of the tight junction components CLDN-1, ZO-1, and OCLN in human gingival epithelial cells. In OKG4 cells and GECs cultured in the presence of 1.4 mM Ca²⁺, CLDN-1 was observed in intercellular contacts. Since Ca²⁺ is an important inducer of TJ formation,⁴¹ tight junctions were absent in cells that had been cultured in the presence of a low Ca²⁺ concentration of 60 µM (Figure A4).

A detailed fluorescence analysis revealed that Ca²⁺ induced the organization of defined strands containing ZO-1 and OCLN in GECs (Figure 5(a)) and in OKG4 cells (Figure A5) under the control conditions. Quantification of the length of ZO-1/OCLN-positive TJ strands revealed that the cytokine treatment significantly increased the quantity of TJ strands in both cell types compared to that in untreated cells, respectively (Figure 5(a–c)).

Figure 5. IL-1β increases the expression of OCLN and enhances the formation of TJ stands in gingival keratinocytes.

(a) A culture of 1.2 × 10⁴ primary GECs treated with IL-1β or left untreated for 24 h was fixed and stained for CLDN1, ZO-1, and OCLN. Image processing was performed with ZEN and ImageJ. Scale bar, 50 µm. The lower panel shows costaining for OCLN and cell nuclei using the fluorescent dye Hoechst 33342. Note that the OCLN-positive TJ strands are not only located at the cell-cell borders but also cross some cell nuclei (marked by arrows). (b,c) The length of OCLN-positive TJ strands of OKG4 cells and primary GECs were quantified using ZEN and ImageJ software. The lengths are given in µm per mm². For OKG4 cells, three different experiments were carried out, and three different visual fields were quantified. The GECs of two different donors were taken into consideration, and 11 and 8 visual fields were quantified. IL-1β significantly enhances the formation of TJ strands in (b) OKG4 cells (p = .032) and (c) GECs (p < .001) compared to untreated cells. Error bars represent the standard deviation. * indicates significant differences between the test and control groups (*0.01 ≤ p < .05; *** p < .001). (d, e) Protein expression of TJ proteins and CDH1 of OKG4 cells was analyzed 24 h and 72 h after IL-1β treatment by Western blotting (d) and quantified (e).

This IL-1β-dependent enhanced formation of TJ strands was accompanied by increased expression of ZO-1 and OCLN, as shown by Western blot analysis using OKG4 cells (Figure 5(d)). The upregulation of ZO-1 and OCLN after 24 h was found to be significant (Figure 5(e)), while expression of CLDN1 and E-cadherin (CDH1), a cell adhesion molecule localized to adherens junctions, remained unaffected by IL-1β. Quantification of Western blot signals revealed that the amount of OCLN in IL-1β-treated cells was still significantly increased after 72 h compared to that in control cells (Figure 5(e)).

The counterstaining of cell nuclei with Hoechst 33342 showed that the OCLN-positive TJ strands partially crossed the cell nuclei (Figure 5(a), lower panel, marked with arrows). Therefore, we determined the distribution of OCLN and ZO-1 in IL-1β-treated and untreated cells more precisely and acquired z-stacks of TJ strands of OCLN/ZO-1 overlays (Figure 6). The reconstructed orthogonal projections revealed that untreated OKG4 cells were primarily organized in flat monolayers, with most cells being positive for ZO-1 (xz1). In areas of visible TJ strands, some cells and cell nuclei were located above the basal layer. In these areas, we observed OCLN in colocalization with ZO-1 but its distribution also extended above the colocalization area without ZO-1. Additional treatment with IL-1β increased the signals of both colocalized OCLN and ZO-1 and subapical localized OCLN (Figure 6(a)).

Figure 6. IL-1β enhances the subapical localization of OCLN in TJs.

OKG4 cells (a) and GECs (b) were cultured (analogous to Figure 5) in the presence and absence of IL-1β. Subsequently, costaining of OCLN (red) and ZO-1 (green) was performed. Three different orthogonal (xz) projections (marked with white dashed lines) generated by CLSM and ZEN software are shown corresponding to the xy projection above. Here, serial sections were collected from the apical surface, with a step increment of 0.99 μm in the z-axis. Below the xz projection, selected magnifications from the respective areas marked with dashed boxes are shown. The arrows indicate subapical-localized OCLN, which is enhanced in IL-1β-treated OKG4 cells and GECs compared to control cells. Images were processed with ZEN software. Scale bar, 50 µm.

These data were confirmed for primary GECs (Figure 6(b)), with distinct staining positive for OCLN and ZO-1, with spots positive for only OCLN in the subapical region of the cells. In GECs, these effects were also enhanced by IL-1β.

Inhibition of NF κB activation prevents the IL-1β-induced upregulation of OCLN

Since NF-κB is a key transcription factor that regulates proinflammatory responses in various cells, including epithelial cells,⁴² we investigated whether NF-κB activation contributes to the IL-1β-induced upregulation of OCLN. For this purpose, OKG4 cells were stimulated with BAY 11–7085, an inhibitor of IκB α phosphorylation that prevents NF-κB activation, in the presence and absence of IL-1β and analyzed occludin expression using Western blotting. As shown in Figure 7, the IL-1β-induced increase in occludin was prevented by the inhibitor. A slightly decrease in occludin expression caused by BAY 11–7085 was also observed under control conditions.

Figure 7. Inhibition of NF-κB activation prevents IL-1β-induced upregulation of OCLN.

Protein extracts were prepared and subjected to Western blotting using an antibody raised against OCLN. β-actin was used as a loading control. IL-1β upregulates OCLN protein expression. Upregulation of OCLN expression was not observed when the cells were pretreated for 1 h with the NF-κB inhibitor BAY 11–7085.

Discussion

The oral mucosa presents a stratified squamous epithelium that forms important physical and immunological barriers that prevent pathogens from invading the tissue. Impairment of the barrier facilitates penetration of plaque-associated bacteria, which may cause periodontal inflammation.⁴³

During inflammatory processes in the oral cavity, the concentration of IL-1 increases in the gingival crevicular fluid⁴⁴ and in the inflamed epithelium.¹² Periodontitis-relevant pathogens such as P. gingivalis and F. nucleatum also provoke a high IL-1 release,⁴⁵ and the biological effects of IL-1 are directly associated with periodontal inflammatory processes, including collagenase activation, bone resorption, and the induction of other inflammatory mediators such as IL-6.^8,46 Inhibition of IL-1 activity in a nonhuman primate model led to reduction of the inflammation, attachment loss, and bone resorption induced by periodontal pathogens.⁴⁷ Thus, IL-1 is regarded as a key cytokine of periodontitis and was therefore selected in this study to generate an inflammatory oral mucosa model derived from gingival epithelial cells.

Characterization of the influence of IL-1β on OKG4 cells revealed that this cytokine induced the secretion of the inflammatory markers IL-6 and IL-8. This is consistent with findings of other groups, which showed that gingival epithelial cells produce IL-6 and IL-8 in response to IL-1;^38,48 these factors are also involved in the course of periodontitis.⁴⁹ Therefore, our established epithelial system, in which cells undergo a transition from cubic cells on the basal side to squamous cells on the surface, can be regarded as an adequate inflammation model. Here, IL-1β neither impacts proliferation nor influences metabolic activity compared to untreated control cells.

To investigate whether the physical barrier, which is the first line of defense and must resist pathogens, is affected by this inflammatory process, we determined the TEER, which serves as an indicator of epithelial integrity.⁵⁰ As shown here, GECs formed a tight barrier during differentiation. The TEER values of keratinocytes cultivated in the presence of 1.4 mM Ca²⁺ were significantly higher than the TEER values of undifferentiated cells, cultivated at 60 µM Ca²⁺, but were lower than the TEER values measured at commercially available full-thickness equivalents, which also contain connective tissue beneath the epithelium.⁵¹

Meyle et al.⁴⁰ first described the course of resistance of cultured gingival keratinocytes. In their study, the TEER of primary epithelial cells reached its maximum 2 d after the Ca2+ switch to 1.4 mM Ca2+. In our system, the TEER reached its maximum a few days later. It then decreased but remained stable at approximately 250 Ω x cm2. The TEER values obtained with other immortalized cell lines derived from the oral epithelium under similar conditions were all lower than those obtained in our system.⁵²

However, the stability of these systems in culture was limited. Frequently, cell detachment and loss of viability were observed after 7 d.⁴⁰ In our system, the cell viability and barrier function of gingival epithelial cells was sustained for several weeks.

Treatment of cells with IL-1β strengthened the physical barrier, as TEER values increased over time after cytokine treatment. This was a surprising result since it is known from other tissues, such as the blood-brain barrier,^53-55 the intestine,⁵⁶ and stratified epithelia, including the skin and gingiva,^29,43 that inflammatory processes lead to the impairment of barrier function. Thus, a long-term treatment of skin keratinocytes with IL-1β reduced the TEER and the expression of occludin and ZO-1.²² Consistently, a longer infection of gingival keratinocytes with P. gingivalis decreased the TEER and changed the expression pattern of the tight junction components claudin-1, −2, and occludin.²⁹

However, there are also reports that describe a strengthening of the barrier at the early stages of inflammation. Treatment of cultured skin keratinocytes with psoriasis-relevant IL-1β resulted in an immediate increase in barrier resistance.²² It has been further reported that at early stages of P. gingivalis-induced inflammation, the barrier function was increased in oral keratinocytes.^29,39

In our models, a significant IL-1β-induced increase in TEER could only be determined for OKG4 cells. In GECs, the resistance was also increased by this cytokine, but this change was not found to be significant. This discrepancy may be explained by the variability among the GEC donors. Cells from donors with a very high TEER per se acted differently in the presence of IL-1β than cells from donors that developed a lower TEER under the control conditions. This differential influence has been already observed by others. Brandner and Schulzke⁵⁷ described a donor-dependent decrease or increase in TEER after IL-4 addition to primary skin keratinocytes. They hypothesized that genetic differences as well as a local and unnoticed inflammation before the preparation of the biopsies may be responsible for the discrepancies.

The IL-1β-dependent increase in resistance, observed in OKG4 cells and primary cells from donors with a lower resistance, was accompanied by an increase in the expression of the TJ proteins ZO-1 and OCLN after 24 h in OKG4 cells. An increase in OCLN was still detected after 72 h of IL-1β treatment in these cells. ZO-1, OCLN, and CLDN1 were shown to be expressed in healthy human gingiva. Here, CLDN1 and ZO-1 were observed in suprabasal layers (except the stratum corneum), whereas OCLN was found predominantly in the uppermost layers of the stratum granulosum.

Double-immunolabeling of the gingiva for ZO-1 and OCLN revealed that both proteins were colocalized in few areas of the upper granular cell layer. This specific localization of TJ components has already been described for the mature and developing human skin⁵⁸ as well as for bovine gingiva.²³

In skin, the upregulation of ZO-1 and OCLN in response to IL-1β has also been described.²² The injection of IL-1β in ex vivo porcine skin and human skin models led to a broadened localization of OCLN and ZO-1 within 24 h. Furthermore, Western blot analysis showed an upregulation of ZO-1, whereas the amount of CLDN1 remained unchanged after 24 h. At later stages of psoriasis, downregulation of TJ proteins was observed.²²

For gingiva, it has been shown that P. gingivalis is able to elicit a strong IL-1 response.⁴⁵ It is accepted that P. gingivalis compromises the physical barrier of the oral mucosa by infecting the oral mucosal tissue.^29,43 However, it has been found that at very early stages of P. gingivalis-induced inflammation, the expression of TJ proteins is enhanced, leading to an increase in barrier function.^29,39 In the skin, pathogenic and nonpathogenic bacteria also cause an upregulation of OCLN and ZO-1 at the beginning of colonization or infection. It was therefore suggested that upregulation of TJ proteins and the associated strengthening of the barrier could be a rescue mechanism after contact between keratinocytes and inflammation-inducing stimuli,⁵⁹ which is also supported by our data. In particular, OCLN may have an important role in strengthening the barrier, as this protein is stably upregulated after IL-1β treatment. The increase in OCLN expression seems to be controlled by the NF-κB pathway since inhibition of NF-κB activation prevents the upregulation of OCLN. A role for NF-κB for both maintaining normal barrier function and for controlling the inflammatory responses has recently been suggested for the alveolar epithelial barrier.⁶⁰

For the first time, this study revealed a new perspective on the physical barrier of the gingival epithelium. Even the healthy gingival epithelium is constantly exposed to a variety of commensal and pathogenic bacteria. Thus, a permanent responsiveness is required in order to maintain a healthy homeostasis. Here, the early expression of antimicrobial peptides and pro-inflammatory mediators appear to play a special role.⁶¹ This study highlighted a new aspect of how pro-inflammatory mediators may alter resistance to the microbial environment. It was a surprising finding that IL-1β indeed strengthened the physical barrier by increased expression of tight junction proteins, and therewith, increased resistance. Although a significant increase in resistance was only observed in OKG4 cells and not in primary GEC, a significant increase of tight junction strands and a similar change in the distributions of ZO-1 and OCLN after IL-1β treatment were recorded in both cell types. Therefore, we suggest that the strengthening of the physical barrier is an important process in the initial defense mechanisms of gingival epithelial cells.

These data provide insights into host responses at the early stages of inflammation as well as TJ formation in human gingival epithelial cells and their regulation under inflammatory conditions. In particular, ZO-1 and OCLN could be identified as important components of gingival epithelial tight junctions. As mentioned previously, gingival keratinocytes express additional tight junction proteins, which may contribute to barrier function and therefore need to be characterized to achieve comprehensive knowledge of the physical barrier in health and during inflammation. This will help, inter alia, in the design of drug/carrier systems that effectively target the inflamed barrier to improve the bioavailability of drugs for the treatment of periodontitis.

Disclosure of potential conflicts of interest

The authors report no conflicts of interest.

Acknowledgments

The authors thank Prof. Dr. Jörg-Dieter Schulzke and PD Dr. Susanne Krug (Department of Gastroenterology, Infectiology and Rheumatology, Charité) for providing valuable help for the visualization of tight junctions. We thank Gudrun Mrawietz for excellent technical assistance and Jasmin Berger for quantification of the TJ strands.

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.

Additional information

Funding

This work was supported by the Sonnenfeld-Stiftung (Antje Bürgel grant to Kim N. Stolte) and the Deutsche Forschungsgemeinschaft under grant DA310/8-1 and grant DO 1375/2-1.

Appendix A.

Table A1. Primary antibodies used in this study.

Antibody raised against/clone	purchased from	source	dilution	Permeabilization and/or blocking solutions
Claudin-1/2H10D10	Thermo Fisher Scientific Inc.	mouse	IF 1:100 WB 1:167	IF: 0.5% Triton-X-100 (v/v) in PBS^+/+/5% BSA (m/v) in PBS^+/+ WB: 5% NFDM/TBST
E-Cadherin/clone 36	BD Biosciences	mouse	WB 1:5000	5% BSA/TBST
Occludin/OC-3F10	Thermo Fisher Scientific Inc.	mouse	IF 1:200	0,25% Triton-X (v/v) in PBS^+/+/5% BSA (m/v) in PBS^+/+
Occludin/clone 19	BD Biosciences	mouse	WB 1:250	5% BSA/TBST
ZO-1, 61–7300	Thermo Fisher Scientific Inc.	rabbit	IF 1:150 WB 1:200	IF: 0,1% Triton-X (v/v) in PBS^+/+/3% BSA (m/v) in PBS^+/+ WB: 5% BSA/TBST
ZO-1/clone 1	BD Biosciences	mouse	WB 1:250	5% BSA/TBST
β-Actin/clone AC-74	Sigma-Aldrich	mouse	WB 1:5000	5% BSA/TBST

IF – Immunofluorescence; WB – Western blot; NFDM – nonfat dry milk; TBST – 0,1% Tween® 20 in TBS

Figure A1. Secretion of proinflammatory cytokines in the presence and absence of IL-1β. IL-8 signals obtained by the cytokine array (Figure 1(a)) were quantified. Error bars represent the standard deviation and * indicates significant difference between the test and control group (p = .04).

Figure A2. Influence of IL-1β on the morphology of OKG4 cells. Visualization of fluorescence-labeled wheat germ agglutinin (WGA) in OKG4 cells (above) and GECs (below) in control and IL-1β-treated cells. 1.2 × 10⁴ OKG4 cells and GECs were cultured in 8-well chamber slides and stimulated with IL-1β or left untreated for 24 h, respectively. Subsequently, cells were fixed. Membranes were visualized with WGA and cell nuclei were stained with Hoechst 33342. Image processing was performed with ZEN and ImageJ, scale bar, 20 µm.

Figure A3. Costaining of ZO-1 and occludin in adult human gingiva.

The analysis was performed as described in the legend to Figure 3; white-lined boxes indicating different regions of the stratum spinosum and the stratum granulosum are shown at a higher magnification. Arrows point to colocalized ZO-1 and occludin (yellow). Scale bar, 100 µm; scale bar in magnifications, 50 µm.

Figure A4. TJ strands are not formed in oral keratinocytes in the presence of a low Ca²⁺ concentration. 1.2 × 10⁴ cells cultivated in the presence of 60 µM Ca²⁺ were fixed and stained for claudin-1, ZO-1 and occludin. Image processing was performed with ZEN and ImageJ. Scale bar, 50 µm.

Figure A5. IL-1β increases the expression of occludin and enhances the formation of TJ strands in OKG4 cells. 1.2 × 10⁴ OKG4 cells treated with IL-1β or left untreated for 24 h were fixed and stained for claudin-1, ZO-1 and occludin. Image processing was performed with ZEN and ImageJ. Scale bar, 50 µm.