Outer Membrane Vesicles Prime and Activate Macrophage Inflammasomes and Cytokine Secretion In Vitro and In Vivo

Outer membrane vesicles (OMVs) are proteoliposomes blebbed from the surface of Gram-negative bacteria. Chronic periodontitis is associated with an increase in subgingival plaque of Gram-negative bacteria, Porphyromonas gingivalis, Treponema denticola, and Tannerella forsythia. In this study, we investigated the immune-modulatory effects of P. gingivalis, T. denticola, and T. forsythia OMVs on monocytes and differentiated macrophages. All of the bacterial OMVs were phagocytosed by monocytes, M(naïve) and M(IFNγ) macrophages in a dose-dependent manner. They also induced NF-κB activation and increased TNFα, IL-8, and IL-1β cytokine secretion. P. gingivalis OMVs were also found to induce anti-inflammatory IL-10 secretion. Although unprimed monocytes and macrophages were resistant to OMV-induced cell death, lipopolysaccharide or OMV priming resulted in a significantly reduced cell viability. P. gingivalis, T. denticola, and T. forsythia OMVs all activated inflammasome complexes, as monitored by IL-1β secretion and ASC speck formation. ASC was critical for OMV-induced inflammasome formation, while AIM2−/− and Caspase-1−/− cells had significantly reduced inflammasome formation and NLRP3−/− cells exhibited a slight reduction. OMVs were also found to provide both priming and activation of the inflammasome complex. High-resolution microscopy and flow cytometry showed that P. gingivalis OMVs primed and activated macrophage inflammasomes in vivo with 80% of macrophages exhibiting inflammasome complex formation. In conclusion, periodontal pathogen OMVs were found to have significant immunomodulatory effects upon monocytes and macrophages and should therefore influence pro-inflammatory host responses associated with disease.

membrane vesicles (OMVs), are released from the subgingival plaque into the subjacent connective tissue where they induce a pro-inflammatory host response (4). Periodontal pathogen OMVs are closed proteoliposomes composed of lipopolysaccharide, lipoproteins, nucleic acids (DNA and RNA), peptidoglycan, porins, and receptors (5)(6)(7)(8)(9), which are known to disrupt tight junctions in epithelial monolayers, induce neutrophil and macrophage recruitment, and stimulate strong pro-inflammatory cytokine responses from various host cells (10)(11)(12). While inflammation is an important component of the host defense, persistent and dysregulated inflammation provides a nutritionally favorable environment for oral pathogenic bacteria adhered to the tooth root in a periodontal pocket and is largely responsible for the tissue and bone destruction that characterizes periodontitis (13).
Monocytes and macrophages are known to shape the host immune response to bacterial infection through phagocytosis, antigen presentation, and cytokine production. Gingival tissue biopsies from periodontitis patients have shown elevated numbers of macrophages and higher concentrations of nitric oxide synthase and pro-inflammatory cytokines IL-1β, TNFα, IL-8, IL-6, and MIP-1α, which serve to promote inflammation and recruit additional immune cells to the site of infection (14)(15)(16). IL-1 family cytokines are significant contributors to inflammation and bone loss during chronic periodontitis and have been correlated with the severity of disease (17,18). The maturation and secretion of IL-1β is mediated by powerful multiprotein complexes termed inflammasomes, which are found in the cytosol of myeloid cells (19). Inflammasome-induced IL-1β secretion requires two signaling events, an initial "cell priming" through NF-κB to mediate synthesis of pro-IL-1β and a second "triggering" event induced by cell surface or cytosolic receptor recognition of pathogen-or damage-associated molecular patterns (PAMPs/DAMPs) that initiate oligomerization of inflammasome components to form an enzymatic complex that results in the proteolytic maturation and secretion of IL-1β (20). Intriguingly, bacterial OMVs are known to bind to mammalian cells and through a number of mechanisms be rapidly internalized, thus OMVs would deliver PAMPs to both cell surface and cytosolic receptors (21) Several classes of inflammasome exist, including the NLR subsets NLRP1, NLRP3, and NLRC4, of which NLRP3 is the best studied. NLRP3 formation is known to be triggered by a wide range of external and internal stimuli, which prime and activate the inflammasome through signal transduction pathways (22,23). Direct cytosolic contact with bacterial PAMPs or other stimuli is not necessary to activate the NLRP3 inflammasome (22). The alternative AIM2 inflammasome is stimulated by cytosolic double-stranded DNA, which may be of viral or bacterial origin or resulting from disruption of the nuclear envelope (24). Inflammasome activation also triggers a form of inflammatory cell death, termed pyroptosis, which promotes the rapid release of cytosolic contents (including IL-1β) primarily due to Caspase-1-induced pores in the cell membrane (25), although other caspases are also known to perform this role (26). Gasdermin-D has recently been identified as a major pore-forming protein (27,28) and can be cleaved by Caspases 1,4,5, and 11 to mediate pyroptotic cell death (29). Pyroptosis is an antimicrobial response that not only eliminates intracellular niches for pathogens but can also cause tissue injury, accelerate bacterial dissemination, and inhibit bacterial clearance from tissues (30). Recently, inflammasome components Caspase-1, NLRP3, and AIM2 have been shown to be upregulated in the gingival tissue of periodontitis patients, suggesting that macrophage inflammasome activation may play a significant role in periodontal immune responses (31).
Circulating blood monocytes are differentiated into phenotypically diverse macrophage classes when recruited into periodontal tissues by the early inflammatory response (32). The classic inflammatory M[IFNγ + lipopolysaccharide (LPS)] macrophage, formerly known as M1, is differentiated by early IFNγ exposure followed by TLR ligation, while anti-inflammatory M(IL-4) macrophages, formerly known as M2, are differentiated by IL-4 or IL-13 cytokine exposure (32,33). This well adapted flexibility allows macrophages to promote, control, or resolve inflammation as required in host tissues. We have shown that M(IFNγ + LPS) macrophages are the dominant infiltrating macrophage in mouse periodontitis models followed by monocytes and undifferentiated M(naïve) class macrophages and are crucial for disease progression (34). Despite their pathogenic potential, few studies have explored the effects of periodontal pathogen OMVs on monocytes and macrophages. P. gingivalis OMVs are reported to induce nitric oxide production (35) and foam cell formation in macrophages (36). Yet gingipains on P. gingivalis OMVs are also reported to promote immune evasion by the proteolytic degradation of membrane-bound LPS receptor CD14 on human macrophages (37). T. forsythia OMVs are known to induce pro-inflammatory cytokine release from macrophages and periodontal fibroblast cell lines (38). While no study to date has explored the effects of T. denticola OMVs on macrophages or monocytes, outer membrane lipoproteins, and lipooligosaccharide, known to be present on T. denticola OMVs (39), have been shown to stimulate nitric oxide production and strong proinflammatory cytokine secretion (TNFα and IL-1β) in murine macrophages (40). To facilitate the development of new therapies for periodontal disease it is vital to understand how inflammation is initiated, controlled, and resolved by immune-modulatory cells in response to bacterial products. The aim of this study was to investigate the interactions between monocytes/macrophages and OMVs derived from oral pathogens.

MaTerials anD MeThODs study Design
The objective of this study was to observe the immunomodulatory effects of periodontal OMVs upon THP-1 cells (in vitro) and peritoneal macrophages extracted from C57BL/6 J mice (ex vivo and in vivo). To assess THP-1 monocyte/macrophage interactions we developed in vitro OMV binding, phagocytosis, cytokine secretion, and inflammasome activation assays. All in vitro experiments were performed at least three times with triplicate samples, data collection was stopped when comparable results were found throughout three experiments.

OMV Preparation and Enumeration Isolation and Enrichment of Highly Purified OMVs
Highly purified OMVs were isolated and enriched using ultracentrifugation, tangential flow filtration, and density gradient separation as previously described (39). Briefly, bacteria were grown to late exponential phase and removed from culture supernatant by centrifugation. The collected supernatant was filtered (0.22 µm) and then concentrated through a 100-kDa filter using tangential flow filtration. The collected concentrate was centrifuged at 100,000 × g for 2 h at 4°C to yield a crude OMV preparation. The crude OMVs were then separated from membrane fragments and other contaminates by a discontinuous OptiPrep™ (Sigma-Aldrich, NSW, Australia) gradient at 150,000 × g for 16 or 48 h. Gradient fractions containing the purified OMVs were pooled and washed with 0.01 M PBS (Sigma-Aldrich, NSW, Australia) and stored at 4°C for short-term storage (<14 days) and −80°C for long-term storage (>14 days).

Enumerating OMVs
Purified OMVs were counted using PKH-26 Red Fluorescent Cell Linker (Sigma-Aldrich, NSW, Australia) and an Apogee A50-Micro Flow Cytometer calibrated with Apogee Flow Systems Calibration Beads (1.6 µm for Red Laser, 1,030 eV/μL) as previously described (39). The Apogee A50-Micro Flow Cytometer was kindly provided by Prof Frank Caruso (University of Melbourne, Australia). Cells were seeded into 96-well flat-bottom tissue culture plates at a volume of 200 µL per well (1 × 10 5 cells per well). THP-1 cells were incubated with P. gingivalis, T. denticola, and T. forsythia OMVs labeled with lipid intercalating dye PKH-26 according to the manufacturer's instructions. Vesicles were added at OMV to cell ratios of 10:1, 50:1, and 100:1. Following incubation at 37°C for 60 min in a 5% CO2 incubator, cell media were removed, the cells were removed by trypsin EDTA solution (Sigma-Aldrich, NSW, Australia), washed and resuspended in PBS, and the level of binding determined using flow cytometry methods previously described (41).

OMV Phagocytosis Assays
THP-1 cells were seeded into 96-well flat-bottom tissue culture plates at a volume of 200 µL per well (1 × 10 5 cell per well). THP-1 cells were incubated with P. gingivalis, T. denticola, and T. forsythia OMVs labeled with pH-sensitive dye pHrodo Red, succinimidyl ester (Life Technologies, Australia) according to the manufacturer's instructions. Vesicles were added at OMV to cell ratios of 10:1, 50:1, and 100:1. Following incubation at 37°C for 60 min in a 5% CO2 incubator, cell medium was removed, the cells were released by trypsin EDTA solution (Sigma-Aldrich, NSW, Australia), washed and resuspended in PBS, and the level of phagocytosis determined using flow cytometry as above.

NF-κB Activation Assays
THP-Blue cells (Invivogen, USA) are characterized by the stable integration of an NF-κB-inducible SEAP reporter gene that induces the secretion of alkaline phosphatase following cell stimulation by pattern recognition receptor (PRR) agonists. THP-Blue cells were grown at 37°C in an anaerobic chamber in complete DMEM (10% v/v fetal calf serum, 3.5% v/v glucose, 1% v/v Pen/ Strep, and 1% v/v l-Glut). Cells were removed from culture flasks (Corning, VIC, Australia) by gentle tapping and counted using a

Cytokine Production Assays
THP-1 cells [monocyte, M(naïve), and M(IFNγ)] were incubated with OMVs for 60 min as per OMV to cell binding assays (above). Following the 60 min incubation, culture fluid was removed, cell monolayers were washed with media, and incubated for a further 24 h in the absence of OMVs as described previously (4). After the second incubation period, supernatant was collected and centrifuged at 800 × g for 5 min at room temperature to remove remaining cells. Cytokine/chemokine secretion was measured using human TNF-α, IL-1β, IL-8, and IL-10 ELISA Kits (Jomar Life Research, VIC, Australia) according to the manufacturer's instructions.

THP-1 and OMV Cytotoxicity Assays
THP-1 cells (both unprimed and primed with P. gingivalis OMVs at 100 ng protein/mL for 4 h) were treated with unlabeled OMVs in cell suspension at the cell concentration and OMV to cell ratios indicated above. OMV-induced cytotoxicity was determined following 4 h of OMV incubation by trypan blue exclusion using a Z1 Coulter Particle Counter (Beckman Coulter, NSW, Australia) as previously described (42).
Stained cells were assessed by flow cytometry using an appropriate laser for the excitation and filter sets for emission of Alexa Fluor 488. Data were analyzed first by creating a scatter gate around the main cell population (SSC-area versus FSC-area dot plot) to exclude debris and outlying events. Scatter gate events were then singlet gated to exclude cell doublets (FSC-width versus FSC-area dot plot). Singlet gate events were plotted by whole cell fluorescence (using empty channel BV 421-A) and ASC (FITC)area, to observe increased in ASC fluorescence.
To assess fluorescent ASC accumulation by microscopy, fixed and ASC antibody-labeled cells were further stained with DAPI as previously described (45) to visualize cell DNA. Stained cells were stored in diamond antifade (Life Technologies) overnight and imaged on an OMX V4 Blaze super-resolution microscope (Deltavision) to observe fluorescent ASC specks.
, and the pH adjusted with the addition of 10 µl of 1.5 M Tris/HCl, pH 8.0, and then heated for 5 min at 100°C, prior to loading on to precast 12% v/v acrylamide gels. Western blots were performed as previously described (47). Briefly, the primary antibody Anti-IL-1β Armenian Hamster IgG (eBioScience) was used at a 1:1,000 dilution, secondary antibody Anti-Arm Hamster IgG Biotin (eBioScience) was used at a 1:2,000 dilution, and Avidin-HRP (Thermo Fisher Scientific, SA, Australia) was used at a 1:2,000 dilution. The proteins determined by Western blot analysis were compared to purified IL-1β (Jomar Life Research, VIC, Australia) controls included on each gel and identified on the basis of MW. protein/mL). Fifteen minutes post injection mice were killed and intraperitoneal cells harvested with 5 mL PBS. Intraperitoneal cells were collected by centrifugation (600 × g × 10 min at room temperature) and resuspended in 1 mL of RPMI 1640 medium without HI-FCS.

IL-1β Secretion by In Vitro Primed and Ex Vivo Activated Intraperitoneal Macrophages
Following centrifugation and resuspension in complete DMEM, intraperitoneal samples were incubated for 6 h at 37°C. Intraperitoneal cells from naive mice (unprimed and unactivated) were also cultured overnight at 37°C in a 5% CO2 incubator on 96-well tissue culture plates to allow the adherence of macrophages. Unbound cells were removed and the adhered macrophages primed for 4 h with P. gingivalis OMVs (100 ng protein/mL) then stimulated with nigericin, silica, P. gingivalis, T. denticola, or T. forsythia OMVs at OMV:cell ratios of 10:1, 50:1, and 100:1 for 6 h. Cellular supernatants of intraperitoneal samples and cultured macrophages were collected and analyzed for IL-1β secretion by a human IL-1β ELISA Kit (Jomar Life Research, VIC, Australia) according to the manufacturer's instructions.

In Vivo Intraperitoneal Cell Inflammasome Detection by ASC Antibody Flow Cytometry and Microscopy
Inflammasome formation was confirmed by flow cytometry using ASC-specific antibody ASC (N-15)-R: sc-22514-R as above for THP-1 cells (see THP-1 ASC Antibody Flow Cytometry and Imaging) and described previously (43,44). Washes were additionally stained with F4/80 and CD11b to correlate ASC inflammasome specks with macrophage populations.

statistical analysis
The abovementioned methods were statistically analyzed using Student's t-test with a minimum size of three biological replicates with three technical replicates. Significant differences were determined as p < 0.05. Error bars represent the SEM of each data subset.

resUlTs host Monocytes and Macrophages Bind and Phagocytose Periodontal OMVs
To investigate early OMV interactions with THP-1 monocytes, M (naïve), and differentiated M (IFNγ) macrophages, OMVs from P. gingivalis, T. denticola, and T. forsythia were labeled with lipid fluorescent dye PKH26 and binding determined by  flow cytometry. Periodontal bacteria OMVs were found to bind to monocytes, PMA-treated M(naïve), and cytokine-treated M(IFNγ) macrophages in a dose-dependent manner (Figure 1).
Phagocytosis of OMVs was determined by labeling OMVs with the pH-sensitive dye pHrodo, which increases fluorescent intensity in the low pH conditions within a phagolysosome. OMVs from all periodontal pathogens were phagocytosed in a dose-dependent manner both in percentage cell binding (Figures 2A,C,E) and MFI (Figures 2B,D,F). While percentage phagocytosis was equal among cell types, MFI revealed M(naïve) and M(IFNγ) were more proficient at phagocytosis than monocytes; M(IFNγ) cells produced the highest MFI (p < 0.01), followed by M(naïve) cells (p < 0.01) (Figures 2B,D,E). T. forsythia OMVs were phagocytosed to the greatest degree (p < 0.001), as observed by both percentage phagocytosis and MFI, followed by P. gingivalis (p < 0.001) and T. denticola OMVs (p < 0.01) (Figure 2).

Periodontal OMVs induce Proinflammatory cytokine Production in host Macrophages
The NF-κB pathway has long been considered a key signaling pathway for the secretion of many cytokines and chemokines. We, therefore, examined how NF-κB activation by OMVs observed in monocytes, M(naïve), and M(IFNγ) cells translated to cytokine secretion. Undifferentiated monocytes, M(naïve), and M(IFNγ) cells were incubated with P. gingivalis, T. denticola, or T. forsythia OMVs and cytokine secretion determined by ELISA assays. Pro-inflammatory cytokines TNFα, IL-1β, and chemokine IL-8 were produced by monocytes, M(naïve), and M(IFNγ) cells after stimulation with all periodontal bacteria OMVs (p < 0.05) (Figures 4A-I). In all cases, M(naïve) and M(IFNγ) macrophages secreted the highest cytokine concentrations in response to OMVs; IL-8 and TNFα were produced in the highest concentrations followed by IL-1β secretion (Figure 4). In monocytes, T. forsythia OMVs induced the greatest TNFα secretion (p < 0.001) followed by T. denticola (p < 0.005) and P. gingivalis (p < 0.05) (Figure 4A). In M(naive) cells, TNFα responses to T. denticola and T. forsythia OMVs were very similar, while responses to P. gingivalis OMVs were significantly less at the 50:1 and 100:1 OMV to cell ratios (p < 0.05) (Figure 4B). M(IFNγ) cells incubated with T. forsythia OMVs secreted the greatest amount of TNFα, followed by T. denticola and P. gin givalis ( Figure 4C). Interestingly, increasing concentrations of OMVs were observed to induce less TNFα secretion in all THP-1 cell subsets to all OMV stimuli (Figures 4A-C).
Porphyromonas gingivalis OMVs were capable of inducing anti-inflammatory cytokine IL-10 in a dose-dependent manner (p < 0.05) at all OMV to cell ratios for M(naïve) and M(IFNγ) macrophages, but only the 50:1 and 100:1 ratios for monocytes (Figures 4J-L). T. denticola and T. forsythia OMVs did not induce IL-10 secretion at any OMV to cell ratio from any THP-1 subtype (Figures 4J-L).

OMV-induced cytotoxicity Differs Between Unprimed and Primed ThP-1 cells
Outer membrane vesicle-induced reduction of cell viability of THP-1 macrophages was observed by the exclusion of trypan blue and viable cell counts using a hemocytometer. Viable cell counts decreased in a dose-dependent manner when THP-1 cell suspensions [monocytes, M(naïve) and M(IFNγ)] were exposed  to T. denticola OMVs (at 50:1 and 100:1 OMV to cell ratios) and T. forsythia OMVs (at the 100:1 ratio) (p < 0.05), while P. gin givalis OMVs were observed to have no cytotoxic effects under these conditions (Figures 5A,C,E).
We further investigated OMV-induced reduction of cell viability of primed THP-1 cell subsets. Upon incubation with P. gingivalis, T. denticola, and T. forsythia OMVs the primed THP-1 subsets had significantly (p < 0.05) less viable cells as compared to control and unprimed THP-1 cells (Figures 5B,D,F). In primed monocytes, all periodontal OMVs induced significant and comparable reductions in viable cells at the 10:1 and 50:1 ratios, while P. gingivalis OMVs induced the greatest viable cell reduction at 100:1 (p < 0.05) (Figure 5B). In primed M(naïve) macrophages, T. forsythia OMVs induced the greatest reduction in viable cells at all OMV to cell ratios, followed by P. gingivalis and T. denticola OMVs (Figure 5D). In primed M(IFNγ) macrophages, P. gingivalis OMVs stimulated a significant reduction in viable cells at all ratios; whereas T. denticola and T. forsythia OMVs stimulated significant viable cell decreases (p < 0.05) at 100:1 and 50:1 ( Figure 5F). Intriguingly, there was no increase in apoptosis or necrosis in the unprimed or primed THP-1 cells upon incubation with any of the periodontal pathogen OMVs.

Periodontal OMVs induce inflammasome activation in ThP-1 Monocytes/ Macrophages
The capacity of periodontal OMVs to induce NF-κB activation, IL-1β secretion, and priming-dependent cell death in cytotoxicity assays was suggestive of inflammasome activation, we therefore explored the ability of OMVs to induce inflammasome activation in THP-1 cell subsets. Several NF-κB stimulating moieties were trialed for their capacity to prime THP-1 cell subsets prior to inflammasome activation. Priming THP-1 cells with E. coli LPS, P. gingivalis LPS, or P. gingivalis OMVs for 4 h was found to significantly increase (p < 0.05) nigericin-, silica-, and Poly(dA:dT)induced IL-1β secretion from monocytes, M(naïve), and M(IFNγ) macrophages compared to unprimed cells (Figure 6). Priming with T. denticola and T. forsythia LPS and OMVs did not induce IL-1β secretion above unprimed cells (Figure 6). Poly(dA:dT) was found to be only an effective inflammasome activator for primed THP-1 monocytes, M(naïve) and M(IFNγ) cells (Figure 6). LPS although a strong inflammasome priming agent was found to be a weak activator for inflammasomes, with only high concentrations of LPS E. coli and P. gingivalis inducing inflammasome activation in primed M(IFNγ) cells M(naive) cells (Figures 6B,C). LPS extracted from T. denticola and T. forsythia was found to not be able to induce inflammasome activation (Figure 6).
As pro-IL-1β can be released by dying cells in the absence of inflammasome activation and be detected as a false-positive signal by ELISA, IL-1β Western blots were performed to distinguish between pro-IL-1β (32 kDa) and mature IL-1β (17 kDa). Pro-IL-1β and mature IL-1β were found in the cell lysate of primed M (naive) cells and increased following nigericin, P. gingivalis, T. denticola, and T. forsythia OMV treatment ( Figure 7D). Mature IL-1β was found in much higher concentrations, predominately in the cell supernatant, and was greatly increased following all OMV treatments (Figure 7D).
Outer membrane vesicle-induced inflammasome formation was further confirmed by utilizing a newly developed flow cytometry method to detect ASC speck formation in macrophage cells (44). A gating strategy was designed to capture whole cells undergoing ASC speck formation and to exclude pyroptotic bodies and non-cellular fluorescent debris. This gating strategy was adopted as it provided consistent results when applied to both in vitro THP-1 cells and in vivo peritoneal macrophages. Fixed and ASC antibody-stained THP-1 monocytes, M(naive), and M(IFNγ) cells were gated by cell size, cell singlets, and finally ASC antibody fluorescence (Whole Cell Fluorescence vs. ASC) to observe increases in the fluorescent signal indicative of ASC speck formation. Background ASC fluorescence [in M(naïve) cells] was recorded at approximately 5.55% and increased to 17.47% with the addition of positive control nigericin ( Figure 7E). Increases in ASC fluorescence of 31.66, 12.22 and 9.49% were observed in monocytes stimulated with P. gingivalis, T. denticola, and T. forsythia OMVs, respectively ( Figure 7E). P. gingivalis OMVs induced less ASC fluorescence in M(naïve) (20.61%) and M(IFNγ) (23.33%) macrophages compared with monocytes, whereas T. denticola and T. forsythia OMVs induced similar levels of speck formation in all cell types (Figure 7E).
Immortalized bone marrow-derived macrophages generated from individual inflammasome gene-deficient mice (48) were used to identify the inflammasome activation pathways stimulated by periodontal OMVs. Wild-type IBMDMs were significantly   , and M(IFNγ) (c) macrophages were primed for 3 h with Escherichia coli, P. gingivalis, Treponema denticola, and Tannerella forsythia LPS and OMVs or left unprimed. Cells were then stimulated with positive controls nigericin, silica, and Poly(dA:dT) or E. coli, P. gingivalis, T. denticola, and T. forsythia LPS as required to match the priming material. Cellular supernatants were collected and IL-1β secretion detected by ELISA. Data are represented as mean ± SEM of three replicates. *represents a significant (p < 0.05) increase from the IL-1β secretion of untreated cells. stimulated (as determined by IL-1β secretion) by nigericin, silica, Poly(dA:dT), P. gingivalis, and T. forsythia OMVs at all OMV to cell ratios and T. denticola OMVs at the higher 50: 1 and 100: 1 ratios (p < 0.05) (Figures 7F and 8A). NLRP3−/− IBMDM cells were not activated by nigericin or silica controls but were activated by Poly(dA:dT), while P. gingivalis, T. denticola, and T. forsythia OMVs induced inflammasome activation similar to wild-type IBMDMs (Figures 7F and 8B). AIM2−/− IBMDMs Results are presented as a qualitative indication (+ to ++++) of IL-1β secretion significantly greater (p < 0.05) than that of unstimulated IBMDM cells and a percentage decrease in IL-1β secretion as compared to secretion from wild-type IBMDM (see Figure 8). did not respond to Poly(dA:dT) and were significantly less (p < 0.05) activated by P. gingivalis, T. denticola, and T. forsythia OMVs, nigericin and silica as compared to wild-type IBMDMs (Figures 7F and 8C). Importantly, ASC−/− IBMDMs were not activated by P. gingivalis, T. denticola, and T. forsythia OMVs or controls nigericin, silica, and Poly(dA:dT) (Figures 7F and 8D).

P. gingivalis OMVs induce inflammasome activation In Vivo
Periodontal OMVs were capable of activating inflammasomes in THP-1 monocytes/macrophages, this virulence characteristic was further confirmed in primary murine macrophages extracted from intraperitoneal washes. Initially unprimed peritoneal macrophages were cultured overnight and primed ex vivo with P. gingivalis, T. denticola, and T. forsythia OMVs. Primary macrophages were found to have strong responses (determined by IL-1β secretion) after activation with nigericin and silica controls (p < 0.001) ( Figure 9A). Activation with P. gingivalis, T. denticola, and T. forsythia OMVs all induced significant (p < 0.05) and comparable dose-dependent IL-1β secretion indicative of inflammasome activation (Figure 9A). We further investigated the capacity of P. gingivalis OMVs to induce inflammasome activation in vivo by injecting P. gingivalis OMVs, E. coli LPS, or PBS into the peritoneal cavity of mice. Three days post injection the peritoneal cell infiltrate was harvested and characterized. P. gingivalis OMV-injected (primed) mice had significantly increased cellular infiltrate (cells/mL) compared to E. coli LPS-primed or PBS-injected mice (Figure 10A). Both P. gingivalis OMVs and E. coli LPS resulted in a similar cell population phenotype dominated by macrophages, which was significantly higher than PBS-injected mice (naive) (Figure 10B).
To confirm in vivo inflammasome activation by P. gingivalis OMVs, peritoneal cells were activated in vivo as above, immediately fixed and stained with ASC-specific antibody. ASC-speck formation was visualized by high-resolution fluorescent microscopy ( Figure 9C). Only peritoneal cells primed and activated with P. gingivalis OMVs displayed distinct ASC speck formation, whereas a diffuse ASC fluorescence was observed in PBS-injected controls, indicative of no activation (Figure 9C and Video S1 in Supplementary Material). To quantify in vivo inflammasome activation, fixed and ASC antibody-stained peritoneal macrophages (F4/80+ and CD11b+) were analyzed for ASC-speck formation by flow cytometry (Figure 9D). Peritoneal cells were gated by F4/80 and CD11b fluorescence to identify macrophage populations, and Figures 10C-H describe the gating strategy to identify ASC speck-positive cells. P. gingivalis OMVs were found to induce ASC-speck formation in a significantly higher (p < 0.01) percentage of E. coli LPS (78.86% of cells) and P. gingivalis OMV (80.42% of cells) primed mice compared to PBS-, nigericin-, and silica-injected mice (ranging from 8.47 to 20.17% of cells) (Figure 9D).

DiscUssiOn
Chronic inflammation plays a pivotal role in the progression of periodontal disease, yet the role of periodontal pathogen OMVs in inflammatory immune cells, such as macrophages, is understudied. Due to their nanoparticle size, adhesive, and proteolytic properties, OMVs are capable of migrating through host tissues, disrupting epithelial tight junctions, and delivering bacterial virulence factors to immune cells in underlying tissues (11). In this study, periodontal OMVs were found to interact strongly with monocytes and macrophages, inducing phagocytosis, NF-κB activation, cellular priming, and strong pro-inflammatory responses, including IL-1β secretion and inflammatory cell death via inflammasome activation both in vitro and in vivo.
Porphyromonas gingivalis, T. denticola, and T. forsythia OMVs were all found to induce significant and comparable inflammasome activation in primed, differentiated M(naïve), and M(IFNγ) macrophages (as observed by IL-1β secretion) while the activation  Although IL-1β ELISA assays are standard in quantifying the inflammasome activity of macrophages, not all IL-1β secretion can be ascribed purely to inflammasome activity, as pro-IL-1β can be released during apoptosis and detected extracellularly. IL-1β western blots confirmed the presence of mature IL-1β in the cell supernatant following OMV treatment and to a lesser extent pro-IL-1β in the cell lysate. Inflammasome activation was also FigUre 9 | Continued Porphyromonas gingivalis outer membrane vesicles (OMVs) prime and activate inflammasome formation in vivo. (a) Intraperitoneal cells from naive mice (unprimed and unactivated) were cultured overnight and stimulated ex vivo with nigericin, silica, P. gingivalis, Treponema denticola, or Tannerella forsythia OMVs, IL-1β secretion was determined by ELISA. (B) For in vivo inflammasome activation C57BL/6 J mice received intraperitoneal injections of phosphate-buffered saline (PBS, naïve), Escherichia coli lipopolysaccharide, or P. gingivalis OMVs 72 h prior to harvest to recruit immune cells to the peritoneal cavity. A second intraperitoneal injection of PBS, silica, nigericin, or P. gingivalis OMVs was administered 15 min prior to killing to active inflammasomes in peritoneal macrophages, IL-1β secretion was determined by ELISA. Data are represented as mean ± SEM of three replicates. *represents a significant (p < 0.05) increase from the IL-1β secretion of naive cells. (c) Microscopy was used to visually observe the dense accumulation of the inflammasome component ASC (green) in DAPI stained (blue) peritoneal macrophages primed and treated with P. gingivalis OMVs. (D) Inflammasome formation was confirmed by flow cytometry using ASC-specific antibody and macrophage markers (F4/80 and CD11b). For full gating strategy see Figure 10. Flow cytometry data display ASC-positive events plotted on a FITC area versus BV421 area (whole cell autofluorescence channel) dot plot. confirmed visually by the localization of ASC fluorescence into the characteristic "speck" formation and via flow cytometry in macrophages. A combination of these techniques confirmed that P. gingivalis, T. denticola, and T. forsythia OMVs activate inflammasome responses in THP-1 monocytes, M(naive), and M(IFNγ) macrophages. We found that particular care must be taken with THP-1 cells during ASC antibody labeling, including the gentle, mechanical detachment of cells (as chemical detachment by trypsin-EDTA severely inhibited speck detection) and the use of FCS-free media during resuspension was critical. Immortalized bone marrow-derived macrophage-knockout cell lines demonstrated that OMV-induced IL-1β secretion is dependent on the inflammasome component ASC but was only attenuated by the absence of Caspase 1, which may be attributed to its reported functional redundancy with Caspase 8 (26), or the non-canonical activation of Caspase 11 by intracellular LPS (49). E. coli OMVs have recently been shown to act as a delivery system for cytosolic LPS, which binds and activates cytosolic Caspase 11 (as well as Caspase 4 and 5) to trigger Caspase-1-independent pyroptosis through the cleavage of pore forming gasdermin D (50,51). It is possible that Caspase 1−/− IBMDM cells utilize Caspase 11 to secrete IL-1β during OMV-induced inflammasome activation. However, the OMV/Caspase 11 activation pathway suggested by Vanaja et al. (51) is dependent upon the biological activity of OMV-associated LPS, as OMVs derived from an E. coli mutant lacking hexa-acylated lipid A (MKV15) induced greatly attenuated IL-1β secretion. P. gingivalis, T. denticola, and T. for sythia are known to possess atypical LPS which lacks functional hexa-acylated lipid A and therefore exhibits greatly reduced biological activity compared with E. coli LPS (52)(53)(54). We suggest that non-canonical activation of Caspase 11 by P. gingivalis, T. denticola, and T. forsythia LPS is unlikely to be the principle method of OMV-induced inflammasome activation, but is rather one of the many functionally redundant mechanisms through which periodontal pathogen OMVs may induce pyroptotic cell death and IL-1β secretion. AIM2 and (to a far lesser extent) NLRP3 knockout cell lines produced attenuated IL-1β secretion, which indicates that both inflammasome pathways contribute to OMV-induced inflammasome responses. NLRP3 and AIM2 have previously been suggested as the principle inflammasomes involved in P. gingivalis-induced inflammasome activation (31); however, it is possible that alternative inflammasome pathways, such as NLRP1 and NLRC4, may also play a role in OMV-induced IL-1β secretion in macrophages. Importantly, the absence of priming or key inflammasome machinery, such as ASC, abrogated IL-1β secretion in macrophage cells (IBMDMs), indicating that inflammasomes are principally responsible for IL-1β secretion in response to OMVs from each periodontal pathogen.
In investigating the ability of P. gingivalis OMVs to prime and activate inflammasomes in vivo, we observed that P. gingi valis OMVs were highly efficient at recruiting immune cells to the site of injection with a threefold increase over baseline cell numbers, compared to a twofold increase induced by E. coli LPS. Flow cytometry revealed that the phenotypic composition of intraperitoneal cells was predominately macrophages and B cells, most likely to be antibody-producing B1 cells due to their peritoneal location (55,56). While B cells are capable of immune modulatory cytokine secretion, including IL-2, IL-4, IL-6, IL-10, IFNγ, and TNFα, they are reported to not possess inflammasome machinery and thus are likely minimal contributors to the IL-1β secretion observed from peritoneal cells (57). This was confirmed by flow cytometry where ASC speck formation occurred exclusively in the macrophage populations. P. gingivalis OMV-induced inflammasome activation was confirmed both in vitro and in vivo by ASC speck formation and strong IL-1β secretion from peritoneal macrophages. To the best of our knowledge, this is the first example of inflammasome activation by vesicles and clearly demonstrates the pro-inflammatory potential of P. gingivalis OMVs.
In addition to inflammasome induced IL-1β, periodontal pathogen OMVs were found to induce strong NF-κB activation and stimulate pro-inflammatory cytokine responses through secretion of TNFα and IL-8. PMA-differentiated M(naïve) and M(IFNγ) cells exhibited significantly higher NF-κB activation and cytokine secretion than THP-1 monocytes. This finding agrees with previous studies that have found M(naïve) and M(IFNγ) macrophages to have higher gene expression of NF-κB, TLR2, TLR7, and TLR8 as well as produce higher concentrations of proinflammatory (IL-1β, IL-6, and TNFα) and anti-inflammatory (IL-10) cytokines (58, 59) than monocytes. Gingival tissues of chronic periodontitis patients exhibit significantly higher levels of pro-inflammatory cytokines IL-1β, IL-6, IL-8, and TNFα than those of periodontally healthy subjects (60,61), hence it is possible that periodontal pathogen OMVs are contributing to this response. The inflammatory effects of these cytokines have been extensively reviewed and include the activation of neutrophils, T and B lymphocytes, macrophages, natural killer cells, and osteoclasts to promote connective tissue destruction and alveolar bone resorption, the clinical hallmarks of chronic periodontitis (62)(63)(64). P. gingivalis OMVs were also found to induce the secretion of IL-10, a potent anti-inflammatory cytokine that is OMVs Activate Macrophage Inflammasome Frontiers in Immunology | www.frontiersin.org August 2017 | Volume 8 | Article 1017 also significantly upregulated in the gingival crevicular fluid of periodontitis patients (65) and is known to suppress the synthesis of pro-inflammatory cytokines from various cell types (66,67 than either T. denticola or T. forsythia OMVs] and the declining responses were not associated with an increase in cell death, this phenomenon may be an example of inflammatory anergy. Inflammatory anergy is a decreased or inhibited inflammatory response in monocytes and macrophages and is thought to be a protective mechanism to prevent detrimental inflammation in the presence of excessive LPS stimulation (68,69). Also known as LPS tolerance, anergy is stimulated in monocytes by secondary contact with endotoxin and is generally associated with the upregulation of IL-10 synthesis (69), as observed for P. gingivalis OMVs in this study. Several studies have reported P. gingivalis OMV-mediated tolerance in monocyte/macrophage cell lines, in which pro-inflammatory responses (TNFα and IL-1β) to E. coli LPS or live P. gingivalis were greatly inhibited by previous exposure to OMVs (37,70). Such differential responses benefit the host by minimizing the inflammatory damage induced by high OMV/bacterial concentrations and prolonged or repeated exposure, but may also benefit bacterial persistence by inhibiting bacterial clearance.
The prolonged presence of inflammatory cells may be a significant factor in chronic periodontal tissue destruction, as such several studies have examined the role of host cell death and its inhibition in periodontal disease (71). Four hours of stimulation with P. gingivalis OMVs induced no significant reduction in cell viability of THP-1 cells while T. denticola and T. forsythia OMVs induced minimal cell death of unprimed THP-1 populations. Under these conditions inflammasomes were not activated, hence low IL-1β secretion and minimal pyroptosis. OMV cytotoxicity was greatly increased by priming THP-1 cells with a low concentration of P. gingivalis OMVs. This is likely to be attributable to inflammasome-meditated pyroptosis, in which primed cells are stimulated by a second exposure to form caspase-dependent pores in the cellular membrane, which encourage inflammatory cell death. While previous studies have traditionally utilized E. coli LPS as a priming agent in inflammasome activation experiments, P. gingivalis LPS was found to be just as effective, despite its atypical LPS structure, while T. denticola and T. forsythia counterparts were found not to induce inflammasome priming. Interestingly, while E. coli and P. gingivalis LPS were adequate priming materials, they lacked the ability to provide a strong second signal and fully stimulate inflammasome activation, even when used at high concentrations. P. gingivalis OMVs were more effective than either LPS at priming THP-1 cells, and in addition were also capable of providing the second inflammasome activation signal. We suggest that either the particulate nature of OMVs is critical to inflammasome activation or the stimulation of multiple PRRs (39) has a synergistic effect greater than that of TLR4 stimulation alone. This finding has interesting implications for the role of OMVs in periodontal disease. The nanoparticlelike aspects of OMVs allow them to penetrate and disseminate through tissues, therefore, during infection an OMV concentration gradient will likely occur from the site of bacterial infection in the polymicrobial biofilm adhered to the tooth root. At distal areas from the initial site of infection OMV concentrations will be low enough to prime but not necessarily activate host cells, as disease bloom tissues will be primed by previous exposure to OMVs and the resulting immune responses will consequently be stronger.
All of these pro-inflammatory interactions are dependent upon initial OMV interactions with macrophages. OMVs derived from P. gingivalis, T. denticola, and T. forsythia were found to bind and be phagocytosed by M(naive)-and M(IFNγ)-differentiated THP-1 cells to a greater extent than monocytes; this increased capacity for phagocytosis has been observed in previous studies using whole bacterial cells (45,72). A comparison of periodontal pathogen OMVs revealed that P. gingivalis OMVs bound all cell types with the greatest affinity, possibly due to the selective enrichment of gingipains on the vesicular surface (9). P. gingivalis gingipains contain adhesin domains known to aid coaggregation and cell binding (73,74). However, T. forsythia OMVs were more readily phagocytosed than P. gingivalis or T. denticola OMVs in pHrodo phagocytosis assays. This may be attributable to the dependence of pHrodo fluorescence on acidification of the phagolysosome. P. gingivalis whole cells are known to evade phagosomal killing through several pathways including persisting in the cytoplasm and prevention of phagosome-lysosome fusion (75)(76)(77). If these mechanisms are shared by P. gingivalis OMVs it would render them partially undetectable by pHrodo but not by PKH-26 fluorescence.
Understanding the process of inflammation, and particularly the activity of key innate immune cells to bacterial virulence factors, is critical to understanding and modulating the progression of chronic periodontitis. This study observed the pathogenic potential of periodontal OMVs to both stimulate and inhibit macrophage pro-inflammatory responses in vitro and in vivo. While T. denticola and T. forsythia OMVs induced predominately pro-inflammatory responses, including TNFα, IL-1β, and IL-8 secretion in unprimed cells and strong inflammasome activation in primed THP-1 macrophages, P. gingivalis OMVs induced variable responses dependent on OMV concentration. Low concentrations of P. gingivalis OMV induced TNFα, IL-1β, and IL-8 secretion in unprimed cells, recruited inflammatory cells (in vivo), and primed macrophages to produce stronger immune responses, inducing both IL-1β secretion and pyroptosis upon second exposure. High concentrations of P. gingivalis OMVs were less inflammatory in unprimed cells due to LPS tolerance and inflammatory anergy but induced strong inflammasome activation upon second exposure. In the pathogenesis of periodontitis, P. gingivalis is generally regarded as the principal or keystone pathogen while T. denticola and T. forsythia are accessory pathogens playing a synergistic or supporting role (78)(79)(80)(81)(82). The potent but flexible immune stimulatory effects observed for P. gingivalis OMVs in this study are likely to assist whole cell P. gingivalis in manipulating and dysregulating the host immune response thereby initiating disease, while the pro-inflammatory effects of T. denticola and T. forsythia OMVs are likely to promote disease progression.