DNA methylation differentially regulates cytokine secretion in gingival epithelia in response to bacterial challenges

Abstract

Epigenetic modifications are changes in gene expression without altering DNA sequence. We previously reported that bacteria-specific innate immune responses are regulated by epigenetic modifications. Our hypothesis is that DNA methylation affects gingival cytokine secretion in response to bacterial stimulation. Gingival epithelial cells (GECs) were treated with DNMT-1 inhibitors prior to Porphyromonas gingivalis (Pg) or Fusobacterium nucleatum (Fn) exposure. Protein secretion was assessed using ELISA. Gene expression was quantified using qRT-PCR. The ability of bacteria to invade inhibitor pretreated GECs was assessed utilizing flow cytometry. Changes were compared to unstimulated GECs. GEC upregulation of IL-6 and CXCL1 by Pg or Fn stimulation was significantly diminished by inhibitor pretreatment. Pg stimulated IL-1α secretion and inhibitor pretreatment significantly enhanced this upregulation, while Fn alone or with inhibitor pretreatment had no effect on IL-1α expression. GEC upregulation of human beta-definsin-2 in response to Pg and Fn exposure was enhanced following the inhibitor pretreatment. GEC susceptibility to bacterial invasion was unaltered. These results suggest that DNA methylation differentially affects gingival cytokine secretion in response to Pg or Fn. Our data provide basis for better understanding of how epigenetic modifications, brought on by exposure to oral bacteria, will subsequently affect host susceptibility to oral diseases.

This well written, succinct article establishes a basis of understanding on how epigenetic modifications, brought upon exposure to oral bacteria, impact host susceptibility to oral disease. It opens up new avenues of research on the impact of DNA methylation on the cytokine response in the oral cavity.

This well written, succinct article establishes a basis of understanding on how epigenetic modifications, brought upon exposure to oral bacteria, impact host susceptibility to oral disease. It opens up new avenues of research on the impact of DNA methylation on the cytokine response in the oral cavity.

INTRODUCTION

In oral immunity, gingival epithelial cells (GECs) are the first line of defense against oral pathogens (Dale 2002; Squier and Brogden 2011). GECs form a physical barrier, and they secrete anti-microbial factors such as small peptide defensins, cytokines and chemokines in response to microbial challenges (Dale 2002; Squier and Brogden 2011). In a healthy state, there is a base level of secretion in response to commensal bacteria and an acute immune response to pathogenic bacteria (Roberts and Darveau 2002; Chung and An 2012). However, in a disease state, such as periodontitis, this balance shifts from surveillance to inflammation (Seymour and Taylor 2004; Pihlstrom et al., 2005; Chung and An 2012; Hajishengallis 2014). There are numerous factors which contribute to this shift including the microorganisms present in biofilms, environmental exposures, pre-existing medical conditions, genetics and epigenetics (Seymour and Taylor 2004; Pihlstrom et al., 2005; Chung and An 2012; Hajishengallis 2014).

Epigenetics are heritable and reversible changes in gene expression without altering DNA sequence (Rodenhiser and Mann 2006; Lod et al., 2013). Epigenetics are implicated in control of embryotic development and in disease processes such as cancer and autoimmune disease. They are influenced by age, environmental factors such as smoking and diet and microbial exposures (Lod et al., 2013; Barros and Offenbacher 2014). The mechanisms by which epigenetics control changes in gene expression are numerous with the most common being histone acetylation, histone de-acetylation, histone methylation and DNA methylation (Rodenhiser and Mann 2006; Lod et al., 2013; Barros and Offenbacher 2014). Histone acetylation is associated with gene expression while methylation and de-acetylation are associated with gene silencing. Similarly, DNA methylation is associated with gene silencing and is the most studied of the epigenetic processes, especially in the field of cancer research (Subramaniam et al., 2014; Weisenberger 2014). However, relatively little is known of the role epigenetics plays in the progression and development of periodontal diseases (Lod et al., 2013; Barros and Offenbacher 2014).

DNA methylation occurs when hydrogen groups (-H) are replaced with methyl group (-CH₃) at 5′ cytosine residues within CpG-rich regions of DNA (Rodenhiser and Mann 2006). These CpG islands occur at a higher frequency in the promoter regions of DNA and are inaccessible to transcription factors when methylated. Methylation of these residues occurs via several DNA methyltransferase (DNMT) enzymes (Barros and Offenbacher 2014; Subramaniam et al., 2014). Of these, DNMT-1 is the most abundant and is involved in the maintenance of methylation (Barros and Offenbacher 2014; Subramaniam et al., 2014). Exposure to pathogens and environmental factors has been associated with changes in the expression of DNMT-1 (Bierne et al., 2012). In addition, DNA is methylated at different levels when comparing periodontitis patients with healthy subjects (Lod et al., 2013; Zhang et al., 2013). However, little is known if pathogenic alterations in DNMT-1 affect the development or degree of periodontitis.

Previous in vitro work from our group has demonstrated that DNMT-1 is down regulated in GECs exposed to the non-pathogenic bridging organism Fusobacterium nucleatum (Fn) or the periopathogen, Porphyromonas gingivalis (Pg) (Yin and Chung 2011). Also, the data suggested that human beta-defensin-2 (HBD-2), a small antimicrobial peptide secreted by epithelial cells and important in innate oral immunity, may be influenced by DNA methylation status in response to Fn but not Pg. Furthermore, DNA methylation status may influence GEC expression of the innate immune marker, CC Chemokine ligand 20 (CCL20), in response to Fn but not Pg. In a similar study, altering DNA methylation of THP-1 monocytes led to a change in TNF-α transcription following bacterial exposure (Zhang et al., 2013).

We hypothesize that DNA methylation plays a role in oral innate immune responses to bacteria. Our goal is to identify gingival epithelial inflammatory cytokines which are potentially under epigenetic control. This was completed by treating GECs with inhibitors of DNA methylation and then exposing those cells to oral bacteria. Cytokines known to contribute to the innate immune response were then assayed for changes in secretion levels. In addition, the ability of bacteria to invade GECs following inhibitor treatment was assessed utilizing flow cytometry in order to determine if changes in methylation status alter cell susceptibility.

MATERIALS AND METHODS

Cell culture

Human primary GECs were isolated from gingival biopsies from healthy patients undergoing third molar extraction as previously described (Yin and Chung 2011). Isolated GECs from a single donor were brought up from cryostorage and expanded in T75 flasks in keratinocyte basal media (KBM) supplemented with 0.03 mM CaCl₂ and BEGM SingleQuot supplements (Lonza, Walkersville, MD), minus retinoic acid. Experiments were completed in multi-well plates with cells cultured in KBM supplemented with 0.15 mM CaCl₂ and KGM SingleQuot supplements (Lonza). All cultures were maintained in a 5% CO₂, 100% humidity, 37°C incubator and used between passages 3 and 5.

Bacteria culture

Pg (ATCC: 33277) were cultured anaerobically at 37°C in trypticase soy broth (BD, Franklin Lakes, NJ), supplemented with 1 g L⁻¹ yeast (BD), 5 mg L⁻¹ hemin (Fluka of Sigma-Aldrich, St. Louis, MO) and 1 mg L⁻¹ vitamin K (Sigma-Aldrich). Fn (ATCC: 25586) were cultured anaerobically in Todd-Hewitt Broth (Fluka), supplemented with 10 g L⁻¹ yeast (BD). Cell number was estimated via turbidity measurements and added to cultures at a multiplicity of infection (MOI) of 1:100.

Inhibitor optimization via cell viability test

GECs were plated in 96-well plates at densities ranging from 10 000 to 30 000 cells cm⁻². The cells were allowed to attach for 24 h, and then either 5^′-azacytidine (AZA: Sigma-Aldrich) or 5-aza-2^′-deoxycytidine (Decitabine, DAC: Sigma-Aldrich) was added at 0, 0.1, 1, 5 or 10 μM concentrations for 24 h. CellTiter-Blue^® (Promega, Madison, WI) viability assay was used to determine an optimum cell density and maximum inhibitor concentration at which cell viability was maintained.

Cytokine secretion ELISA

GECs were plated in 6-well plates at a density of 20 000 cells cm⁻² and allowed to attach for 24 h. Cultures were then treated with either AZA (1 μM) or DAC (5 μM) for 4 h prior to 16 h exposure to either Pg or Fn. Supernatants were then collected and stored at −20°C until processing. Supernatants were tested using a MultiAnalyte ELISArray Custom Kit (Qiagen) for the following secreted cytokines and chemokines: TNF-α, IL-1α, IL-6, CXCL-1 (Gro-α), IL-12 and CCL-3 (MIP-1α).

qRT-PCR

Cell lysates were collected from the same wells as supernatants. From these lysates, mRNA were isolated (RNeasy Mini Kit: Qiagen, Valencia, CA), and cDNA were generated (High Capacity cDNA Reverse Transcription Kit: Applied Biosystems of Life Technologies, Carlsbad, CA). Using qRT-PCR (iQ SYBR Green Supermix; CFX96 Real Time System: Bio-Rad, Hercules, CA), HBD-2 expression was probed with GAPDH as a housekeeping gene as previously described (Yin and Chung 2011).

Bacterial invasion

Flow cytometry was used to assess the ability of bacteria to invade GECs. Briefly, GECs were plated at 80% confluency and allowed to adhere for 24 h, and then treated with AZA (1 μM) or DAC (5 μM) for 4 h. Concurrent with inhibitor treatment, 1 × 10⁹ Pg or Fn were labeled with the fluorescent probe, 5-(6)-carboxyfluoresceinsuccinimidylester (CFSE; Invitrogen of Life Technologies) according to previously published results (Pils et al., 2006). GECs were then exposed to labeled bacteria (MOI = 100) for 1.5 h (Pg), 4 h (Fn) or 16 h (Pg or Fn). Following inhibitor and bacterial exposure, GECs were detached from the culture plate with Detachin (Genlantis, San Diego, CA), rinsed with PBS (Invitrogen), and suspended in PBS. Just prior to flow analysis, an equal volume of 0.4% Trypan Blue solution (final concentration 0.2%; Invitrogen) was added to the cell suspension to extinguish signal from the extracellular bacteria. Cells were evaluated using an LSRII flow cytometer, gating on eukaryotic cells based on forward and side scatter and detecting CFSE fluorescence (EX492 nm : EM517 nm). An uptake index was computed as the product of the percentage of CFSE-positive cells multiplied by the mean fluorescence intensity of those cells, for 10 000 cells.

Data analysis

All samples were completed in a minimum of triplicates. Data from ELISA were normalized to untreated samples and represented as fold changes over untreated for each cytokine. A two-sided student t-test was utilized to determined statistical significance as defined by a p-value < 0.05.

RESULTS

DNMT-1 inhibitor concentration optimization

Experimental optimization was completed to ensure cell viability and health. Fig. 1 shows the results of GEC exposure to a range of concentration of AZA and DAC for an optimal seeding density of 20 000 cells cm⁻² (n ≥ 8 for each condition). The arrow indicates the maximum inhibitor concentration in which viability was statistically identical to untreated cultures: 1 μM for AZA and 5 μM for DAC (p > 0.05). These inhibitor concentrations were then used in all subsequent experiments.

Figure 1.

GEC viability following inhibitor treatment. GECs were exposed to a range of DNMT-1 inhibitor (AZA or DAC) concentrations for 24 h to determine the maximum inhibitor concentration that could be tolerated without a significant decrease in cell viability. Arrows indicate the maximum concentration of AZA (1 μM) and DAC (5 μM) chosen for all subsequent experiments. The * indicates inhibitor concentrations which significantly decreased GEC viability (p < 0.05) compared to untreated GECs.

AZA and DAC pretreatment alter HBD-2 expression

GEC expression of HBD-2 was assessed to determine if DNMT-1 inhibitors were putting added stress on the cells and to test the effect of methylation status on the responsiveness of GECs to bacterial challenges. Treatment with either AZA or DAC alone did not change HBD-2 expression compared to untreated GECs (Fig. 2). Exposure of GECs to Pg (Fig. 2a) and Fn (Fig. 2b) resulted in a 2–20-fold upregulation of HBD-2 expression but only Pg-induced HBD-2 upregulation was statistically significant (p < 0.05) compared to unstimulated control. The expression of HBD-2 in response to Pg was unchanged by AZA pretreatment, but significantly increased (40-fold over unstimulated control) by DAC pretreatment. For Fn, DNMT-1 inhibitor pretreatment resulted in significant increases in HBD-2 expression compared to unstimulated GECs (7-fold increase), but only DAC pretreatment resulted in a significant (p < 0.05) increase in HBD-2 expression compared to Fn exposure alone.

Figure 2.

GEC expression of HBD-2. HBD-2 mRNA expression by GECs normalized to the housekeeping gene GAPDH in response to Pg or Fn. Pg exposure significantly upregulated HBD-2 expression by GECs and DAC pretreatment further enhanced this response. Fn exposure slightly upregulated HBD-2 expression and both AZA and DAC pretreatment further enhanced this response. The * denotes p < 0.05 compared to unstimulated GECs. The ** denotes p < 0.05 compared to bacterial stimulation only.

AZA and DAC pretreatment alter cytokine secretion

Pg stimulated the production of four of six cytokines assayed, TNF-α, IL-1α, IL-6 and CXCL1 (Figs 3a and 4a). In contrast, Fn stimulated IL-6 and CXCL-1 (Figs 3b and 4b). AZA pretreatment in the absence of bacterial stimulation decreased the secretion of CXCL1 compared to unstimulated controls but had little effect on any other cytokines (Fig. 3). Conversely, DAC pretreatment in the absence of bacterial exposure increased both IL-6 and CXCL-1 secretion compared to unstimulated controls while having no effect on any other cytokines tested (Fig. 4).

Figure 3.

GEC cytokine secretion following AZA pretreatment. Cytokine secretion by GECs following pretreatment with AZA and exposure to Pg or Fn. Three of six assayed cytokines showed significant alterations in secretion levels in response to AZA pretreatment and Pg exposure: IL-1α secretion was significantly enhanced compared to Pg exposure while IL-6 and CXCL-1 secretions were significantly decreased. AZA pretreatment significantly decreased GEC secretions of IL-6 and CXCL-1 in response to Fn exposure, but had no effect on the other cytokines assayed. The * denotes p < 0.05 compared to unstimulated GECs. The ** denotes p < 0.05 compared to bacterial stimulation only.

Figure 4.

GEC cytokine secretion following DAC pretreatment. Cytokine secretion by GECs following pretreatment with DAC and exposure to Pg or Fn. Four of six assayed cytokines showed significant alterations in secretion levels following DAC pretreatment and Pg exposure: TNF-α and IL-1α secretions were significantly enhanced compared to Pg exposure while IL-6 and CXCL-1 secretions were significantly decreased. DAC pretreatment had no significant effects on GEC secretions of the probed cytokines in response to Fn exposure. The * denotes p < 0.05 compared to unstimulated GECs. The ** denotes p < 0.05 compared to bacterial stimulation only.

When GECs were pretreated with AZA and then exposed to Pg, IL-1α secretion was significantly (p < 0.05) enhanced while both IL-6 and CXCL1 secretion were decreased (Fig. 3a). DAC pretreatment followed by Pg stimulation resulted in a similar pattern of expression for IL-1α, IL-6 and CXCL-1. In addition, TNF-α was significantly (p < 0.05) enhanced when GECs were pretreated with DAC and exposed to Pg compared to Pg exposure alone. AZA pretreatment followed by Fn exposure resulted in a significant decrease in secreted IL-6 and CXCL-1 compared to Fn exposure alone (Fig. 3b). However, DAC pretreatment had no effect on the response of GECs to Fn exposure for any secreted cytokines tested.

AZA and DAC's effect on bacterial invasion

Fig. 5 shows the uptake index for GECs exposed to fluorescently labeled Fn or Pg at both an initial invasion time point and a later internalization time point. Here, the uptake index is a relative measure of internalized bacteria; GECs not exposed to bacteria have an uptake index of essentially 0. The amount of internalized bacteria for both Pg and Fn increased with longer bacteria exposure. However, pretreatment with either AZA or DAC did not significantly alter GEC internalization of Pg or Fn at either the initial exposure times or over the 16 h time course of these experiments.

Figure 5.

GEC susceptibility to bacterial invasion following inhibitor treatment. Uptake index associated with GEC exposure to fluorescently labeled Pg or Fn as a function of both time and DNMT-1 inhibitor exposure. Neither AZA nor DAC significantly altered the initial invasion (1.5 h for Pg, 4 h for Fn) or late invasion/internalization (16 h) of Pg or Fn into treated versus untreated GECs.

DISCUSSION

The purpose of this study was to identify periodontal innate immunity cytokines and chemokines that may be under epigenetic control. Our approach involved treating GECs with known inhibitors of methylation and then exposing both treated and untreated cells to bridging or periopathogenic bacteria. Differences in response indicate the potential for epigenetic control. During the course of these experiments, we also completed invasion assays to ensure that the results observed were not due to these inhibitors altering the susceptibility of GECs to bacterial invasion.

As was demonstrated in Fig. 5, there was no biologically significant difference in the susceptibility of GECs to the invasion of Pg or Fn following treatment with either inhibitor of DNA methylation. These data were consistent at both initial, early exposure times and later time points. The initial invasion time point for Pg was taken at 90 min based on previous studies (Lamont et al., 1995; Chung et al., 2001), while the initial invasion time point for Fn was taken at 4 h both because the fluorescent signal was minimal at earlier time points and due to published studies citing a 4 h invasion time for Fn (Han et al., 2000; Ji et al., 2010). The later time point was explored to determine if inhibitor pretreatment had a long-term effect on the susceptibility of GECs to bacterial invasion and internalization. It was specifically chosen as 16 h because this was the longest time period GECs were exposed to Pg or Fn in these experiments due to cell stress and death at longer exposure times. Overall, the flow cytometry data suggest that any differences in cytokine secretion in response to bacterial challenges may be attributed to potential methylation effects within the cells rather than alterations in cell susceptibility to bacterial invasion resulting from treatment with DNA methylation inhibitors.

Considering the GEC cytokine secretion data, the pattern of secretion depended not only on the bacterial species but also on the specific inhibitor of DNA methylation. Much of the current data regarding these inhibitors come from cancer research (Christman 2002; Hollenbach et al., 2010; Nguyen et al., 2010). In this context, differential de-methylation of genes has been seen when the same cells are treated with AZA versus DAC, in vitro (Hollenbach et al., 2010; Nguyen et al., 2010). This may be because AZA is a ribonucleoside (incorporated into both DNA and RNA) while DAC is a deoxyribonucleoside (incorporated into only DNA). Thus, although AZA and DAC are structurally and functionally similar (both lead to depletion of DNMT, hypomethylation and DNA damage), they have distinctly different paths by which they reach this point (Christman 2002; Hollenbach et al., 2010; Nguyen et al., 2010). In addition, because AZA is also incorporated into RNA, there are many additional effects on cellular function beyond those only related to DNA methylation.

For the current data, the six cytokines (TNF-α, IL-1α, IL-6, CXCL-1, IL-12 and CCL-3) were chosen because of their importance in initiating and modulating the gingival innate immune response to microbial challenges (Feliciani et al., 1996; Silva et al., 2007; Squier and Brogden 2011). Pretreatment of GECs with AZA or DAC did not alter secretion of four of these cytokines. The two cytokines directly affected by inhibitor treatment were CXCL-1 which decreased after AZA pretreatment and increased after DAC pretreatment, and IL-6 which also increased with DAC pretreatment. For CXCL-1, one difference between this cytokine and the other tested is that CXCL-1 has a large CpG island within the gene itself (NCBI-Epigenomics 2014). Additionally, CXCL-1 is structurally similar to IL-8 and utilizes the same receptors (Geiseris et al., 1993). Previous research has suggested that IL-8 expression is not under epigenetic control via DNA methylation (Yin and Chung 2011) while these data suggest that CXCL-1 secretion is potentially under epigenetic control. Taken together, these data suggest a complex mechanism by which innate immune homeostasis is regulated and also suggest CXCL-1 as a potential cytokine of interest to further study both in vitro and in vivo.

In the case of IL-6, evidence from the cancer literature suggests that in addition to IL-6 functioning as an inflammatory cytokine in the immune system, it also has an epigenetic effect, locating DNMT-1 to the nucleus and ultimately resulting in increased CpG methylation and decreased gene expression (Hodge et al., 2007). In addition, upregulation of IL-6 was observed in vitro in lung epithelial cells treated with DAC (Tang et al., 2011). Thus it is possible that treatment of GECs with inhibitors of DNMT-1 may stimulate the production of IL-6. A previous research on thymic cells suggests a role of IL-6 in regulating CXCL-1 (Tseng et al., 2010), although the possible regulation of CXCL-1 by IL-6 in GECs is unclear in oral innate immunity.

When GECs were treated with AZA or DAC and then exposed to Pg, most patterns of cytokine secretion were similar with the exception of TNF-α which was altered by DAC pretreatment while unchanged by AZA pretreatment. Differences in GEC response following DAC and AZA pretreatment were also found for HBD-2 expression following Pg stimulation. As discussed above, these inconsistencies are most likely due to the differences in the two inhibitors themselves with AZA also being incorporated into RNA and subsequently affecting protein synthesis (Veselý and Čihák 1978; Christman 2002), potentially negating the changes in GEC response due to DNA methylation status alone. Recently published data suggest that TNF-α may be under epigenetic control via methylation of its promoter CpG island (Zhang et al., 2013). Regardless, the data with both AZA and DAC pretreatment suggest that the secretion of IL-1α, IL-6 and CXCL-1 in response to Pg may all be under epigenetic control via methylation of DNA and may contribute to the development, severity or treatment responsiveness of periodontal diseases.

As for the association of DNA methylation status with innate immune responses to Fn, our data suggest a role of DNA methylation for the regulation of HBD-2 in response to Fn. This response was consistent regardless of the type of inhibitor to which the cells were pre-exposed. The data on the modulation of cytokine secretion suggest that both IL-6 and CXCL-1 may be under some epigenetic regulation in response to Fn exposure.

Overall, DNA methylation differentially affects cytokine secretion from epithelial cells in response to Pg or Fn. Our data provide a basis for better understanding of how epigenetic modifications, brought on by exposure to oral bacteria, will subsequently affect host susceptibility to oral diseases and offers potential avenues for future treatment strategies.

FUNDING

Supported by NIH/NIDCR Grants R01DE19632.

Conflict of interest statement. None declared.