Aggregatibacter actinomycetemcomitans Induces Autophagy in Human Junctional Epithelium Keratinocytes

The adverse environmental conditions found in the periodontium during periodontitis pathogenesis stimulate local autophagy responses, mainly due to a continuous inflammatory response against the dysbiotic subgingival microbiome. The junctional epithelium represents the main site of the initial interaction between the host and the dysbiotic biofilm. Here, we investigated the role of autophagy in junctional epithelium keratinocytes (JEKs) in response to Aggregatibacter actinomycetemcomitans or its purified lipopolysaccharides (LPS). Immunofluorescence confocal analysis revealed an extensive nuclear translocation of transcription factor EB (TFEB) and consequently, an increase in autophagy markers and LC3-turnover assessed by immunoblotting and qRT-PCR. Correspondingly, challenged JEKs showed a punctuate cytosolic profile of LC3 protein contrasting with the diffuse distribution observed in untreated controls. Three-dimensional reconstructions of confocal images displayed a close association between intracellular bacteria and LC3-positive vesicles. Similarly, a close association between autophagic vesicles and the protein p62 was observed in challenged JEKs, indicating that p62 is the main adapter protein recruited during A. actinomycetemcomitans infection. Finally, the pharmacological inhibition of autophagy significantly increased the number of bacteria-infected cells as well as their death, similar to treatment with LPS. Our results indicate that A. actinomycetemcomitans infection induces autophagy in JEKs, and this homeostatic process has a cytoprotective effect on the host cells during the early stages of infection.

serotype b, its purified LPS, and junctional epithelium keratinocytes (JEKs) to recreate the initial stage of the pathogenesis of periodontitis.

Bacterial Strain and LPS Purification
A. actinomycetemcomitans serotype b (ATCC ® 43718 ™ ) was used in this study. Bacteria were cultured in a capnophilic environment (8% O 2 and 5%-12% CO 2 ) at 37 • C in Brain Heart Infusion (BHI) medium (#CM1135B, Oxoid, Hampshire, UK) supplemented with 10% horse serum, as previously described [25,26]. To ensure viable bacteria expressing their full antigenic potential, all experiments were performed with bacteria harvested at the exponential growth phase. LPS from A. actinomycetemcomitans serotype b was purified by the Tri-reagent method and analyzed by 14% SDS-PAGE stained with periodic acid-silver, as previously described [27,28] (Supplementary Figure S1E).

In Vitro Infection Model
Cell infection assays were performed in 24-well plates (#174899, Thermo Fisher Scientific, Waltham, MA, USA) containing 13 mm diameter round glass coverslips at the bottom of each well. Coverslips were coated with 1.5 × 10 5 adherent OKF6/TERT-2 (70%-80% confluence) in supplemented keratinocyte serum-free medium. Individual wells were washed with sterile PBS to remove the antibiotics and then incubated with bacteria (MOI = 200) diluted in antibiotic-free keratinocyte medium. Plates were centrifuged at 1200 rpm for 5 min and then incubated for 3 h at 37 • C. At the end of the incubation period, cells were washed thrice with sterile PBS, and the slides were processed for downstream assays. The same above approach was employed for LPS stimulation.

Immunofluorescence Colocalization Analysis
Confocal images of untreated control cells or LPS treated and infected keratinocytes, immunostained with anti-TFEB (#ab122910), and DAPI were processed for colocalization analysis using the coloc tool of Imaris software, as previously reported [29]. Briefly, the software allowed the construction of additional fluorescent channels corresponding to colocalized voxels adjusted by fluorescent thresholds. TFEB fluorescent channel was colocalized with the DAPI channel, which was depicted as a new image of colocalized voxels. The software quantified the number of colocalized voxels between the channels (DAPI/TFEB) by applying the same thresholds for all experimental conditions.

Algorithm-Based Autophagosomes Detection
Detection was performed as previously reported [29], with some modifications. The series of z-stack images acquired at confocal microscopy were processed by Imaris software for three-dimensional (3D) reconstructions and detection of autophagosomes in x-y-z coordinates. Autophagosomes detection Cells 2020, 9, 1221 5 of 21 was performed as follows. Confocal images of OKF6/TERT-2 cells were immunostained with anti-LC3B (autophagosome), anti-LPS (bacteria), and DAPI (nucleus), and then processed by Imaris software, which allowed the construction of isospots from the fluorescence signals [30,31]. Isospots were constructed based on two classes of LC3B signal detection, colocalizing or not with intracellular bacteria; thus, generating two types of isospots (green and purple spheres) with different sizes.

Cell Viability Analysis
OKF6/TERT-2 cells were incubated with 3 mM 3-Methyladenine (3-MA) (#ab12084, Abcam), an upstream autophagy inhibitor, for 3 h and then stimulated with A. actinomycetemcomitans (MOI = 200) or 1 µg/mL of its LPS for 3 h at 37 • C. Cells were washed and incubated for 20 min with Zombie Aqua™ Fixable Viability Kit (#423101, Biolegend ® , San Diego, CA, USA), an amine-reactive fluorescent dye that is non-permeant to live cells but permeant to cells with compromised membranes. Samples were washed, fixed in 4% PFA, and resuspended in PBS prior to analysis in CytoFLEX TM V3-B3-R3 flow cytometer (Beckman-Coulter Life Sciences). Fluorescence intensity and cell counts were determined using the pacific blue detector, acquiring a total of 10 5 events. Collected data were analyzed using FlowJo v10.0.8 software (Tree Star Inc, Ashland, OR, USA).
Cell viability was further analyzed by a second approach. Briefly, cells were treated and stimulated as aforementioned, followed by incubation for 15 min with the fluorescent DNA-binding dyes DAPI and 7-Aminoactinomycin D (7-AAD) (#A1310, Life Technologies), which are semipermeable and non-permeable to live cells, respectively. Following incubation, samples were washed, fixed with 4% PFA, mounted using Prolong Gold antifade (#P36930, Life Technologies,) and analyzed by confocal microscopy. The percentage of viable cells was calculated subtracting the number of dead cells (7-AAD positive) from the total number of cells (DAPI positive).

A. Actinomycetemcomitans Induces Autophagy in JEKs
A. actinomycetemcomitans has been strongly implicated in the development of rapidly progressing periodontal disease due to its ability to adhere, invade, and damage the junctional epithelium, adjacent to the tooth surface. [16][17][18]. To evaluate whether A. actinomycetemcomitans induces autophagy in this periodontal context, JEKs were incubated with A. actinomycetemcomitans ( Figure 1) and then processed using confocal immunofluorescence to visualize TFEB. Upon autophagy stimulation, TFEB translocates from the cytoplasm to the nucleus, where it induces the transcription of genes involved in autophagosomes biogenesis [34]. JEKs challenged with A. actinomycetemcomitans showed an extensive nuclear translocation of TFEB ( Figure 2A) as determined by the quantification of colocalized DAPI/TFEB voxels ( Figure 2B). Consistently, increased levels of TFEB transcripts were detected in challenged cells ( Figure 2C). Likewise, with the activation of TFEB, we observed an increase of Atg5 transcription ( Figure 2D) and Atg5-Atg12 protein complex expression in infected JEKs ( Figure 2E, and Supplementary Figure S1A). Microtubule-associated protein 1 light chain 3 (LC3B-I) is a soluble cytosolic protein that is converted into LC3B-II, on the autophagosome membrane, upon autophagy stimulation [5,35]. Although no differences in LC3B transcription levels were detected in OKF6/TERT-2 cells ( Figure 2F), increased JEKs challenged with A. actinomycetemcomitans showed an extensive nuclear translocation of TFEB ( Figure 2A) as determined by the quantification of colocalized DAPI/TFEB voxels ( Figure 2B). Consistently, increased levels of TFEB transcripts were detected in challenged cells ( Figure 2C). Likewise, with the activation of TFEB, we observed an increase of Atg5 transcription ( Figure 2D) and Atg5-Atg12 protein complex expression in infected JEKs ( Figure 2E, and Supplementary Figure S1A). Microtubule-associated protein 1 light chain 3 (LC3B-I) is a soluble cytosolic protein that is converted into LC3B-II, on the autophagosome membrane, upon autophagy stimulation [5,35]. Although no differences in LC3B transcription levels were detected in OKF6/TERT-2 cells ( Figure 2F), increased LC3B-II expression was observed after A. actinomycetemcomitans-challenge, suggesting a bacterial-induced LC3B turnover ( Figure 2G, Figure 5B, and Supplementary Figure S1B).
Confocal images revealed enhanced expression of LC3B protein in infected keratinocytes, indicating a bacteria-induced autophagosome formation ( Figure 3A, yellow arrowhead, and Figure 3B). Moreover, three-dimensional (3D) reconstructions of confocal sections [29,31] showed a close association between intracellular bacteria and LC3B-positives vesicles (inset of Figure 3A, Figure Figure 3E, purple spheres). We speculate that the vesicles of 1.2 µm could be induced by the ability of A. actinomycetemcomitans to secrete many LPS-coated outer membrane vesicles (OMVs) during infection, enhancing its virulence and exacerbating the host's inflammatory response [14,18]. Taken together, these results suggest that A. actinomycetemcomitans induces autophagy in JEKs.
Cells 2020, 9, x FOR PEER REVIEW 2 of 22 LC3B-II expression was observed after A. actinomycetemcomitans-challenge, suggesting a bacterialinduced LC3B turnover ( Figure 2G, Figure 5B, and Supplementary Figure S1B). Confocal images revealed enhanced expression of LC3B protein in infected keratinocytes, indicating a bacteria-induced autophagosome formation ( Figure 3A, yellow arrowhead, and Figure  3B). Moreover, three-dimensional (3D) reconstructions of confocal sections [29,31] showed a close association between intracellular bacteria and LC3B-positives vesicles (inset of Figure 3A, Figure 3B-D, and Supplementary Figure S1D), suggesting that A. actinomycetemcomitans exploits autophagy route during its intracellular life cycle. Two populations of LC3B-positive vesicles were detected, vesicles of approximately 2.5 µm containing bacteria ( Figure 3E,F, green spheres) and a population  of about 1.2 µm without bacteria ( Figure 3E, purple spheres). We speculate that the vesicles of 1.2 µm could be induced by the ability of A. actinomycetemcomitans to secrete many LPS-coated outer membrane vesicles (OMVs) during infection, enhancing its virulence and exacerbating the host's inflammatory response [14,18]. Taken together, these results suggest that A. actinomycetemcomitans induces autophagy in JEKs.

Purified LPS from A. Actinomycetemcomitans Induces TFEB Nuclear Translocation and Biogenesis of LC3-Positive Vesicles in JEKs
To elucidate whether the LPS released from A. actinomycetemcomitans during the infection is able to induce autophagy in the host cell, JEKs were incubated with purified LPS and then analyzed by confocal microscopy. Extensive nuclear translocation of TFEB, independent of LPS concentration, was observed ( Figure 4A,B). Quantification of colocalized DAPI/TFEB voxels in LPS kinetic-stimulation assays confirmed that LPS-induced TFEB activation was time-dependent ( Figure 4C (0.5 µg/mL) and 4D (1 µg/mL)). Nevertheless, TFEB and LC3B transcriptional levels were increased only in cells stimulated with 1 µg/mL of purified LPS (Figures 4E and 5A). Consistent with TFEB activation, A. actinomycetemcomitans' LPS induced a strong turnover of LC3B protein ( Figure 5B). Densitometric measurements showed a significant increase in LC3B-II expression at 3 h after LPS-challenge, an effect that was inhibited in keratinocytes pretreated with 3-MA, an inhibitor of autophagosomes biogenesis ( Figure 5C,D). In the same context, JEKs were incubated for 3 h with 1 µg/mL of purified LPS and then subjected to confocal immunofluorescence analysis for detection and visualization of autophagosomes and endocytosed LPS. Confocal images showed a punctuate cytosolic profile of LC3B protein in LPS-challenged cells, in contrast to the diffuse distribution observed in untreated controls ( Figure 5E). Similarly, the vesicular profile of LC3 (autophagosomes) was intimately associated with the cytosolic presence of LPS, as shown in Figure 5E (LC3B/LPS colocalization channel), suggesting that LC3B-vesicular profile (LC3-II) was induced by endocytosed-LPS. Overall, these results indicate that LPS from A. actinomycetemcomitans induces autophagy in JEKs.

A. Actinomycetemcomitans Induces Selective Autophagy Mediated by p62/SQSTM1
Ubiquitin is a crucial molecule in xenophagy that labels substrates that will undergo selective degradation. Autophagy adaptors, such as p62/SQSTM162 (p62), have ubiquitin-binding domains and LC3-interacting regions that recognize ubiquitinated substrates and deliver them for lysosomal

A. Actinomycetemcomitans Induces Selective Autophagy Mediated by p62/SQSTM1
Ubiquitin is a crucial molecule in xenophagy that labels substrates that will undergo selective degradation. Autophagy adaptors, such as p62/SQSTM162 (p62), have ubiquitin-binding domains and LC3-interacting regions that recognize ubiquitinated substrates and deliver them for lysosomal degradation. P62 is an adapter protein that has been associated with anti-bacterial and selective autophagy induced by LPS [4,36]. Thus, we investigated the influence of bacteria or purified LPS on p62-recruitment in JEKs. Although infected cells did not show significant differences in p62-expression as compared with the non-infected control ( Figure 6A,B and Supplementary Figure S1C), an intimate association between p62-labeling in LC3B-positive vesicles containing bacteria was observed ( Figure 6C, white arrowheads). Approximately 62% of the LC3B-positive bacteria also exhibited p62-labeling ( Figure 6C graph). Likewise, JEKs incubated with LPS (1 µg/mL) exhibited a significant increase in the transcription and expression of p62 ( Figure 6D,E). Confocal microscopy images corroborated the colocalization between p62 and endocytosed-LPS (3 h after challenge) ( Figure 6F). Taken together, these results indicate that p62 is the main adapter protein recruited during A. actinomycetemcomitans infection. degradation. P62 is an adapter protein that has been associated with anti-bacterial and selective autophagy induced by LPS [4,36]. Thus, we investigated the influence of bacteria or purified LPS on p62-recruitment in JEKs. Although infected cells did not show significant differences in p62expression as compared with the non-infected control ( Figure 6A,B and Supplementary Figure S1C), an intimate association between p62-labeling in LC3B-positive vesicles containing bacteria was observed ( Figure 6C, white arrowheads). Approximately 62% of the LC3B-positive bacteria also exhibited p62-labeling ( Figure 6C graph). Likewise, JEKs incubated with LPS (1 µg/mL) exhibited a significant increase in the transcription and expression of p62 ( Figure 6D,E). Confocal microscopy images corroborated the colocalization between p62 and endocytosed-LPS (3 h after challenge) ( Figure 6F). Taken together, these results indicate that p62 is the main adapter protein recruited during A. actinomycetemcomitans infection.

The Pharmacological Inhibition of Autophagy Increases the Infected Cell Number and the Cell-Death of A. Actinomycetemcomitans-Challenged JEKs
Autophagy is a sequential dynamic process that involves first the biogenesis and acidification of vesicles, followed by the degradation of vesicular contents after their fusion with lysosomes, which is the reason why this process can be pharmacologically inhibited in different stages [1,5]. To assess the role of JEKs-autophagy during A. actinomycetemcomitans infection, we inhibited autophagic flux in early and late stages. First, OKF6/TERT-2 cells were challenged for 3 h with A. actinomycetemcomitans and subsequently treated with 200 nM bafilomycin A1, 10 µM chloroquine, or 20 mM NH 4 Cl ( Figure 7A). These compounds alkalinize the host's lysosomes (Supplementary Figure S1G,H), inhibiting the maturation of autophagosomes to autolysosomes, a fundamental step in the degradation of autophagic cargoes [5]. Downstream-autophagy inhibition significantly increased the number of infected cells ( Figure 7B, central column, and Figure 7C). In addition, many intracellular bacteria were observed closely colocalizing with the host-lysosomes in cells treated with autophagic inhibitors ( Figure 7B, yellow arrowheads in the left column). Consistent with the increase of infected cells, treated-JEKs exhibited several bacteria colocalizing with intercellular actin protrusions ( Figure 7D white arrowheads). These suggest that the arrest of autolysosomal degradation increased intracellular-bacterial accumulation and favored the transcellular spread. In the same context, autophagy can play a dual role in periodontitis, either promoting or blocking cell-death, depending on the host cell type [21].
We also evaluated the effect of upstream autophagy pharmacological inhibition on the viability of infected or LPS-treated keratinocytes. JEKs were pretreated with 3-MA (3 h) and then challenged with A. actinomycetemcomitans or their purified-LPS (3 h). First, we checked that the pretreatment with the drug does not affect the number of infected JEKs by the bacteria (Supplementary Figure S1I). 3-MA blocks the formation of autophagosomes by inhibiting the class III phosphatidylinositol 3-kinases (PI3K) [35]. Cell viability was assessed by two different methodologies: confocal microscopy ( Figure 8A) and flow cytometry ( Figure 8B), as specified in the methods section. Both approaches revealed a significant increase in JEKs death when upstream autophagy was suppressed prior to challenge with A. actinomycetemcomitans or its purified LPS ( Figure 8A,B graph), suggesting that autophagy has a protective effect against cell-death during A. actinomycetemcomitans infection. Collectively, our results indicate that A. actinomycetemcomitans and its LPS induce autophagy in JEKs, a molecular process that has a protective effect on host cells in the early stage of the infection by this periodontal pathogen. Cells 2020, 9, x FOR PEER REVIEW 9 of 22 We also evaluated the effect of upstream autophagy pharmacological inhibition on the viability of infected or LPS-treated keratinocytes. JEKs were pretreated with 3-MA (3 h) and then challenged 8A) and flow cytometry ( Figure 8B), as specified in the methods section. Both approaches revealed a significant increase in JEKs death when upstream autophagy was suppressed prior to challenge with A. actinomycetemcomitans or its purified LPS ( Figure 8A,B graph), suggesting that autophagy has a protective effect against cell-death during A. actinomycetemcomitans infection. Collectively, our results indicate that A. actinomycetemcomitans and its LPS induce autophagy in JEKs, a molecular process that has a protective effect on host cells in the early stage of the infection by this periodontal pathogen.

Discussion
The defense mechanism provided by the junctional epithelium is key for protection against the onset of periodontitis and the formation of the periodontal pocket. Indeed, structural changes are pivotal in the transformation of the junctional epithelium to a pocket epithelium and determine the formation of the periodontal pocket [37]. In this context, protective mechanisms displayed by JEKs, such as autophagy, are critical for periodontal protection against bacteria and their products. Autophagy can protect cells from apoptosis; however, excessive autophagy can destroy essential cellular components and lead to cell death [22]. In the present study, we demonstrated for the first time that A. actinomycetemcomitans infection induces autophagy in human JEKs, a cellular homeostatic process with a cytoprotective effect on this cell type, in the early stages of infection (Figure 9).
events showing the live and dead (inside the black box) cell populations. The bar graph below represents the percentage of cell death obtained by flow cytometry of five independent assays. *** p < 0.001.

Discussion
The defense mechanism provided by the junctional epithelium is key for protection against the onset of periodontitis and the formation of the periodontal pocket. Indeed, structural changes are pivotal in the transformation of the junctional epithelium to a pocket epithelium and determine the formation of the periodontal pocket [37]. In this context, protective mechanisms displayed by JEKs, such as autophagy, are critical for periodontal protection against bacteria and their products. Autophagy can protect cells from apoptosis; however, excessive autophagy can destroy essential cellular components and lead to cell death [22]. In the present study, we demonstrated for the first time that A. actinomycetemcomitans infection induces autophagy in human JEKs, a cellular homeostatic process with a cytoprotective effect on this cell type, in the early stages of infection (Figure 9). During periodontitis pathogenesis, the junctional epithelium represents the first natural barrier that biofilm-forming bacteria must undermine to reach deeper periodontal tissues. As an interface During periodontitis pathogenesis, the junctional epithelium represents the first natural barrier that biofilm-forming bacteria must undermine to reach deeper periodontal tissues. As an interface between the gingival sulcus and connective tissues, the junctional epithelium controls the constant microbiological challenge, protecting the tooth-supporting periodontium [19,38]. Our analysis revealed that the challenge with bacteria or its purified LPS induced autophagy in OKF6/TERT-2 cells during the first hours after stimulation. Several lines of evidence suggest that the activation of TFEB early during infection is evolutionarily conserved, positioning this molecule as a key component of host defense [39]. Immune signaling elicits inflammation by promoting cellular damage, and TFEB has a recognized anti-inflammatory protective activity against microbial stimuli or its secreted products, including LPS. In this context, TFEB suppresses inflammation in a manner dependent or independent of the Nuclear Factor-kappa B (NF-kB) modulation, a key signaling pathway for the maintenance of immune homeostasis of the epithelia. [39][40][41][42]. Phosphorylated TFEB is localized on the lysosomes in the cytosol; however, when it is dephosphorylated, it translocates to the nucleus and stimulates the expression of genes involved in lysosomal biogenesis and autophagy induction [34]. In this study, bacterial and LPS challenge induced an increase in transcription of TFEB mRNA and extensive nuclear translocation of TFEB protein to the nucleus of JEKs (Figure 2A,B and Figure 4), where it acts as the primary regulator of the autophagosomes biogenesis [34]. Consistently, stimulated cells also exhibited a significant increase in essential autophagic markers involved in the early stages of autophagosomes formation. Bacteria and its purified LPS induced a significant LC3-protein turnover, a widely used parameter for evaluation/confirmation of autophagic flux ( Figure 2D-G and Figure 5A-D) [5,35]. Confocal microscopy analysis confirmed a vesicular profile of LC3 protein in challenged cells, in contrast to the diffuse pattern exhibited by untreated cells (Figures 3A and 5E). 3D reconstructions of infected JEKs showed two populations of LC3-positive vesicles with different sizes, i.e., vesicles of 2.5 µm containing bacteria and smaller LPS-positive vesicles without bacteria inside. This suggests that both, intracellular bacteria and endocytosed-LPS released during infection activated the autophagic pathway. Nevertheless, because the molecular processes that induce dephosphorylation and activation of TFEB are quite complex, the mechanisms by which A. actinomycetemcomitans and its LPS dephosphorylate TFEB and induce its migration to the nucleus needs to be fully clarified. Xenophagy and LPS-induced autophagy are selective processes. Adaptor-mediated LC3 recruitment is a widely accepted model for explaining the selective recognition of ubiquitinated substrates by the autophagic machinery [4]. Our study confirmed the role of p62 as the main autophagy-adapter protein during the early stage of A. actinomycetemcomitans infection. The human p62 protein contains an N-terminal LIR (LC3-interacting region) motif and a C-terminal UBA (Ubiquitin-associated) domain [36]. In accordance with this, we observed that approximately 62% of the LC3-positive bacteria also exhibited p62-labeling, which suggests that p62 links the bacteria to autophagosomes via LC3 ( Figure 6). Collectively, these data suggest that the selective autophagy stimulated in gingival keratinocytes was in response to the initial interaction with A. actinomycetemcomitans and their products.
Although the processes of adhesion/invasion, capture, and microbial degradation have been well characterized, the host mechanisms that recognize intracellular pathogens to induce xenophagy remain unclear [4,8]. Based on our results, we visualized a plausible scenario that was schematically represented in Figure 9 (Upper panel). A. actinomycetemcomitans adheres and invades epithelial cells through a process that involves actin polymerization and subsequent receptor-mediated endocytosis [16]. In parallel, LPS-coated OMVs are actively released by A. actinomycetemcomitans in vivo and can be internalized into cultured epithelial cells mainly via clathrin-dependent endocytosis or fusion with host cell membranes in a cholesterol-dependent manner [18]. According to our proposal, the bacteria, its purified LPS, or OMVs lead to the activation of NF-kB when recognized by the surface Toll-like receptors [17,43] or cytosolic NOD-like receptors [44,45]. Thus, the activated NF-kB pathway would trigger the dephosphorylation and activation of TFEB, which in turn induces its migration to the nucleus and the upregulation of autophagy-related genes. Once internalized, this bacterium has several virulence factors (such as hemolytic factors or phospholipase C) that allow it to transgress the endosomal membrane, and be released into the nutrient-rich cytosol, to begin its intracellular replication [16]. As a consequence, the damage to the endosomal membrane results in its ubiquitination and recruitment of p62, possibly through his C-terminal UBA domain. Once the vesicular cargo has been recognized, p62 interacts with the lipidized form of LC3 on the surface of the forming autophagosome (LC3-II) and directs A. actinomycetemcomitans and its LPS to lysosomal degradation through the autophagic pathway. Given the importance of these processes in antibacterial cell clearance, the mechanisms that trigger the ubiquitination of A. actinomycetemcomitans and its LPS, and the subsequent recruitment of p62, should be extensively addressed and clarified by future studies.
Evidence indicates that A. actinomycetemcomitans traverses the gingival epithelium using pro-apoptotic mechanisms triggered by the host's enhanced pro-inflammatory response [17]. In the present study, the up and downstream pharmacological inhibition of autophagy (Figures 7 and 8) significantly increased the infected-cell percentage and the cell death of A. actinomycetemcomitans-challenged or LPS-treated JEKs (Figure 9, lower panel). These findings confirm the pro-survival protective role of autophagy in periodontal tissues during the pathogenesis of periodontitis, which has been widely reported [23]. For instance, a recent study reported that macrophages infected with A. actinomycetemcomitans when treated with trans-cinnamic aldehyde significantly diminished the inflammatory response and induced bacterial-intracellular killing by autophagy activation [46]. Accordingly, the anti-inflammatory drug meresin inhibited the LPS-induced apoptotic cell-death of periodontal ligament cells via autophagy stimulation [47]. Exfoliative sulcular keratinocytes and human cultured keratinocyte cells (HaCaT) showed increased autophagy after P. gingivalis-LPS treatment, via increased-ROS levels [48]. A similar effect was observed in human periodontal ligament cells, in which increased autophagy was observed in response to hypoxia-induced apoptosis [22].
In agreement with other models of infection, our results indicate that JEKs autophagy is an important cell homeostatic mechanism of epithelial antibacterial defense that is activated by invasive bacteria and their products, in order to avoid the microbial dissemination and favor bacterial clearance [49][50][51]. In this context, LAMP-2-deficient mice, a key protein for the autolysosome's formation, developed severe periodontitis in the early stages of life. These animals exhibited exacerbated accumulation of bacterial biofilm, gingival inflammation, alveolar bone resorption, loss of periodontal attachment, and apical migration of junctional epithelium, reinforcing the importance of bacterial-autophagic clearance of our results [52]. Contrary to our results with A. actinomycetemcomitans, P. gingivalis-induced autophagy has pro-bacterial effects. This important periodontal pathogen stimulates their entry into the autophagic pathway and, thus, avoids or delays their lysosomal degradation, increasing its colonization and penetration into host periodontal tissues, critical events for the spreading of the infection [53][54][55]. If the activation and differential modulation of the autophagic pathway by periodontal pathogens is important for the establishment and the severity of periodontitis, this would be an important topic that needs further clarification. In this regard, a recent report shows increased levels of autophagy-related proteins in gingival tissues of aggressive periodontitis patients, suggesting a potential distinctive role of autophagy among aggressive and chronic forms of the disease [56]. The authors also described increased autophagy activity in A. actinomycetemcomitans-infected THP-1-derived macrophages, generating anti-inflammatory consequences linked to ROS modulation [56]. Nevertheless, it is important to point out that these results were obtained with a cell type that requires external stimuli for its in vitro differentiation, which could generate some artifacts in the induction of the autophagy process. Given that the here presented research was carried out using a human periodontal keratinocyte cell line, our results can be interpreted closer to the periodontal physiological reality.
During infection, host cells respond to bacterial pathogens through the upregulation of degradative processes, including autophagy [57]. This enhancement was demonstrated here in A. actinomycetemcomitans-induced JEKs, suggesting a protective role of periodontium against infection, at least against this bacterial species. As the junctional epithelium consists of 1-3 cell layers in thickness at its apical termination [38], the antibacterial and pro-survival effect mediated by autophagy herein described, could provide valuable insight into the pathogenesis of periodontitis, periodontal health, and healing. In this context, the stimulation of autophagy could increase antibacterial resistance and tissue tolerance mechanisms, reducing the effects of the constant challenge of biofilm bacteria and their products on the junctional epithelium; thereby, preventing bacterial invasion, transcellular dissemination, and subsequent destructive immunoinflammatory response displayed during the onset and progression of periodontitis pathogenesis. In summary, our data highlight the critical relationships between periodontal infection and autophagy in gingival keratinocytes and reveal putative mechanisms implied in gingival protection and periodontitis prevention.

Conclusions
Autophagy is a key modulator of several molecular processes involved in inflammatory diseases, including periodontitis. These findings allow us to conclude that autophagy modulation and targeting within the epithelial gingival barriers could represent a novel therapeutic strategy for prevention of the early stages of periodontitis, possibly stopping its progression and promoting periodontal homeostasis.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4409/9/5/1221/s1. Figure S1: Effect of A. actinomycetemcomitans on autophagy markers activation. S1A-C) Full-length immunoblottings presented in the main manuscript. Protein extracts from OKF6/TERT-2 cells incubated with several serotypes of A. actinomycetemcomitans were incubated with antibodies directed against the specified autophagy markers. The black arrowhead indicates the non-adjacent lanes displayed in Figure 2E-G and Figure 6B