Porphyromonas gingivalis lipopolysaccharide rapidly activates trigeminal sensory neurons and may contribute to pulpal pain

Aim To determine whether Porphyromonas gingivalis lipopolysaccharide (LPS) can directly activate trigeminal neurons, to identify which receptors are involved, and to establish whether activation leads to secretion of the neuropeptide calcitonin-gene related peptide (CGRP) and/or the translocation of NF- κ B. Methodology Mouse trigeminal ganglion (TG) cells were cultured in vitro for 2 days. The effect of P. gingivalis LPS (20 µg/mL) on calcium signalling was assessed (by calcium imaging using Cal-520 AM) in comparison to the transient receptor potential channel A1 (TRPA1) agonist cinnamaldehyde (CA; 100 µM), the TRP channel V1 (TRPV1) agonist capsaicin (CAP; 1 µM), and high potassium (60 mM KCl). TG cultures were pre-treated with either 1 µM CLI-095 to block Toll-like receptor 4 (TLR4) signalling or with 3 µM HC-030031 to block TRPA1 signalling. CGRP release was determined using ELISA, and nuclear translocation of NF- κ B was investigated using immunocytochemistry. Data were analysed by one-way analysis of variance, followed by Bonferroni’s post-hoc test as appropriate. Results P. gingivalis LPS directly exerted a rapid excitatory response on sensory neurons and non-neuronal cells (p<0.001 to p<0.05). The effects on neurons appear to be mediated via TLR4- and TRPA1-dependent pathways. The responses were accompanied by an increased release of CGRP (p<0.001) and by NF- k B nuclear translocation (p<0.01). Conclusions This study reports for the first time that P. gingivalis LPS directly activates trigeminal sensory neurons (via TLR4 and TRPA1 receptors) and non-neuronal cells, resulting in CGRP release and NF- k B nuclear translocation. This indicates that P. gingivalis can directly influence activity in trigeminal


Introduction
Infection of the dental pulp initiates a variety of pathophysiological changes including the generation of pain. While immune mechanisms are known to indirectly influence nociception, bacteria or their products have also been reported to activate sensory neurons directly. A range of oral bacteria are known to infect dental pulps, many of which are Gramnegative, and amongst which is Porphyromonas gingivalis (Rôças et al. 2002).
A major component of Gram-negative bacterial surfaces is lipopolysaccharide (LPS) and previous investigators have shown that LPS might directly activate sensory neurons (Diogenes et al. 2011, Tse et al. 2014. Escherichia coli LPS has been shown to increase neuronal excitability of dorsal root ganglia and trigeminal ganglion neurons (Meseguer et al. 2014). However, P. gingivalis LPS differs from that of E. coli and it is not known whether this and other non-enterobacterial LPS that would be found in pulpal infection have the same effects and, if so, which neuronal receptors are involved.
There are two categories of cell surface receptors that are thought to be key participants in a rapid nociceptor response to LPS and the initiation of neurogenic inflammation; Toll-like receptors (TLRs) and transient receptor potential channels (TRPs). TLRs are patternrecognition receptors highly conserved throughout evolution (Santoni et al. 2015), and it is known that they can be activated by some forms of LPS (Diogenes et al. 2011, Meseguer et al. 2014, Tse et al. 2014. Of the 11 TLRs identified in humans so far (Qi et al. 2011), TLR4 specifically recognizes LPS of Gram-negative bacteria (Coats et al. 2005) and previous workers have shown that TLR4 is expressed by many cell types within the dental pulp, including odontoblasts (Jiang et al. 2006), fibroblasts (Hirao et al. 2009) and nerve fibres (Wadachi & Hargreaves 2006). TRPs are a family of cation channels that respond to a range of thermal and chemical stimuli including inflammatory stimuli (Geppetti et al. 2008, Veldhuis & Bunnett 2013, Santoni et al. 2015. E. coli LPS is capable of sensitizing TRPV1 via a mechanism involving TLR4 (Diogenes et al. 2011). Also, the TRPA1 receptor can be activated by the anionic lipid A of LPS (Meseguer et al. 2014). Although TRPV1 and TLR4 co-expression has been demonstrated (Diogenes et al. 2011), a complete picture of the crosstalk with other receptors and the actual pathways that these receptors feed into has not been elucidated. Therefore, the aim of this study was to determine whether P. gingivalis LPS could induce pulpal pain by directly activating trigeminal neurons and lead to secretion of the neuropeptide CGRP and the translocation of NF-κB.

Animals
Healthy, 3-9-month-old C57BL/6 wild-type adult mice (30-35 g) were used. All animals were maintained on a 12-hour light/dark cycle in a temperature-controlled environment and given food ad libitum. All animal procedures were conducted under the Animal (Scientific Procedures) Act 1986, and approved by the UK Home Office.

Mouse trigeminal ganglia (TG) primary cultures
Mice were sacrificed by an overdose of isoflurane anaesthetic and cervical dislocation, and the TG were removed immediately. TG were placed in ice-cold, calcium-and magnesiumfree Hanks' balanced salt solution (HBSS; Invitrogen, Paisley, UK), and then quickly washed twice with HBSS supplemented with 1% penicillin/ streptomycin (PS) (Gibco, Paisley, UK) at 37°C in 5% CO2. Ganglia were then treated with 1.25% w/v type IV collagenase (Sigma, Gillingham, UK) once for 60 min and then again for 45 min, followed by 1.25% w/v trypsin (Sigma) for 15 min at 37°C. After digestion, the TG were homogenized mechanically by pipetting up and down, and centrifuged at 1500 g for 3 min, then re-suspended in supplemented medium (DMEM/F12 media, 1% w/v BSA, 1% N2 supplement, 1% PS; Bottenstein & Sato 1979). The mixtures were transferred to poly-l-lysine and laminin-coated coverslips that were placed at the bottom of 24-well tissue culture trays and cultured for 2-3 days with media change once after 24 h. All experiments on cultured cells were performed on day 2-3.

Intracellular calcium measurements
Cells were loaded for 60 min at 37°C with 5 µM Cal-520 AM (Abcam, Cambridge, UK) in DMEM/F12 medium after which coverslips were transferred into a chamber (Warner Instruments RC-25F; Harvard Apparatus Ltd, Edenbridge, UK ) mounted on the stage of a Zeiss Axiovert S100TV microscope (Zeiss, Jena, Germany) and continuously perfused with buffer (142 mM NaCl, 5 mM NaHCO3, 10 mM Hepes, 16 mM Glucose, 2 mM KCl, 2 CaCl2, 1 mM MgCl2, and 0.1% BSA, pH 7.3) at a flow rate of 5 mL/min. Cells were illuminated at 488 nm with a monochromator (TILL Photonics, Kaufbeuren, Germany), and viewed using a band-pass filter (510-540 nm) and 40 × objective (NA1.3; Zeiss). Images were acquired at 6second intervals using a Photometrics Cascade 512B CCD camera (Photometrics UK Ltd, Marlow, UK) and timelapse sequences recorded using WinFluor ® software(V3.8.7; Strathclyde Imaging Software, University of Strathclyde, Glasgow, UK). Following background subtraction, fluorescence values (measured for every image from regions of interest [ROIs] placed over the soma of cells) were exported and plotted using GraphPad Prism 6 (GraphPad Software, San Diego, CA, USA). Cells were classified as responsive to a particular stimulus when fluorescence, measured in an ROI, increased by more than five standard deviations over the baseline fluorescence. Baseline fluorescence was measured as the average of the last 10 frames before application of tested products or agonists. Data are presented as the changes in intensity of fluorescence between the baseline and peak of the activated curve. This calculation was used to account for variations in the starting fluorescence of individual cells.
The effect of P. gingivalis LPS (20 µg/mL) on calcium signalling was assessed in comparison to the TRPA1 agonist cinnamaldehyde (CA; 100 µM), the TRPV1 agonist capsaicin (CAP; 1 µM), and high potassium (60 mM KCl). To determine whether signalling was occurring via TLR4 or the TRPA1 receptor, we used specific blocking agents applied to the TG cultures prior to exposure to LPS. TG cultures were pre-treated with either 1 µM CLI-095 (Invivogen, UK) for 3 h to block TLR4 or for 1 minute with 3 µM HC-030031 (Tocris Bioscience, Bristol, UK) to block TRPA1 signalling (Kunkler et al. 2011, Meseguer et al. 2014, Tse et al. 2014) and the resultant calcium flux of cells was recorded as described above.
After two initial washes in HBSS, a 15 min baseline sample was collected. The cells were then exposed to either HBSS alone or HBSS containing P. gingivalis LPS (20 µg/mL) for 15 min. Where indicated, inhibitors of TLR4, TRPA1 and TRPV1 were administered at 1 µM CLI-095 for 3 h, 3 µM HC-030031 for 2 min, or 10 µM capsazepine for 2 min prior to the exposure to LPS, as described by previous investigators (Eid et al. 2008, Henrich & Buckler 2009, Tse et al. 2014. The supernatants were collected and assayed by a competitive enzyme-linked immunosorbent assay (ELISA) for calcitonin gene-related peptide (CGRP).
Briefly the ELISA was performed by coating microtitre plate wells with 1:1000 anti-CGRP mouse monoclonal capture antibody (Sigma) in 0.05 M carbonate buffer pH 9.5 overnight at 4°C to immobilize the capture antibody to the wells. Plates were then washed three times with 0.05% PBS-Tween and blocked with 5% BSA in 0.05% PBS-Tween (pH 7.4) for 3 hours at room temperature. After washing, 50 µL of standard CGRP or samples were added to the plate, together with 50 µL of 2 ng/mL biotin-conjugated CGRP (Biorbyt Ltd, Cambridge, UK), and incubated for 1 hour at 37°C. After washing six times, the plates were incubated with 1:400 streptavidin-horseradish peroxidase conjugate (strep-HRP) for 45 min.
Tetramethylbenzidine/hydrogen peroxide substrate solution (TMB; Sigma) was then added until colour developed (10-20 min), and the reaction was stopped by addition of 2M HCl.
Resulting colour was read using a spectrophotometer (Infinite 200 PRO; Tecan, Reading, UK) at 450 nm. Unknown CGRP concentrations were calculated by interpolation from a standard curve. NF-κB nuclear translocation was determined by immunohistochemistry and co-localising with DAPI and β-tubulin III, markers of DNA and neurons, respectively. The translocation of NF-kB could be determined by analysing the proportion of NF-kB fluorescent label in the nuclear and cytoplasmic compartments. Positive nuclear translocations were quantified by measuring the relative fluorescence intensity of NF-κB staining at each intracellular location.

NF-kB translocation
DAPI nuclear staining was used to confirm the location of the nuclei. The translocation of NF-kB was counted as positive when the nuc/cyt ratio was >1. Neurons and non-neuronal cells positive for NF-kB nuclear translocation were quantified.

P. gingivalis LPS can rapidly activate trigeminal ganglion neurons and non-neuronal cells
The responses of trigeminal neurons to P. gingivalis LPS, cinnamaldehyde (CA), capsaicin (CAP) and high potassium were assessed using the fluorescent calcium indicator, Cal-520 AM ( Figure 1). A total of 90 neurons were assessed from three to four coverslips per experiment and responses were examined from three independent experiments. P. gingivalis LPS was found to directly activate approximately 27% of the trigeminal neurons present (24/90). Our study also demonstrated that approximately 13%, and 33% of mouse TG neurons responded to CA and CAP, respectively. Application of high potassium was performed to assess neuron viability and to help distinguish neuronal from non-neuronal cells ( Figure 1G, H).
Neuronal cells show a very clear rapid rising and transient response to KCl resulting from depolarization and action potential-evoked calcium signals, the amplitude and duration of calcium transients being determined by the contribution of different ion channels contributing to the action potential (e.g. Mohammed et al. 2017); in contrast, non-neuronal cells do not fire action potentials and where calcium signals are observed, these are notably slower, and are likely related to transporters. As shown, neuronal cells also show clear rapid responses to cinnamaldehyde or capsaicin, whereas non-neuronal cells show no or only very weak responses to these compounds. As well as differences in the calcium signalling, the morphology of cells can also be used as a discriminator, as indicated in Figure 1A. Note the non-neuronal cells are much smaller than neurons and lack the extensive thin branching neurites characteristic of neurons.
Approximately 40% of the non-neuronal cells present, such as satellite glial cells and Schwann cells, also responded to P. gingivalis LPS (representative data are shown in Figure 1). This suggests that neuronal supporting cells can form part of the responsive population to P. gingivalis LPS in the peripheral nervous system (see Figure 1H).

Calcium influx activated by LPS can be abolished by TRPA1 antagonist and modified by TLR4 antagonist
Whether P. gingivalis LPS triggered calcium influx via TLRs or the TRPs was then investigated. For this, TLR4 and TRPA1 were functionally blocked by pretreating TG neurons with either CLI-095 or HC-030031, respectively. Responses were examined from three independent experiments. A total of 72 neurons in the presence of HC-030031 and experiment. The TRPA1 antagonist (3 µM HC-030031) completely abolished the neuronal response to P. gingivalis LPS (0/72), whereas pre-treatment of neurons with a specific TLR-4 signalling inhibitor (1 µM CLI-095) reduced but did not completely inhibit the response. The percentage of neurons responding to LPS in the presence of CLI-095 was reduced to less than 5%, suggesting a contributory role for the TLR4 receptor. Representative, individual traces are shown in Figure 2.

Neuropeptide (CGRP) release by P. gingivalis LPS can be prevented by TLR4, TRPA1, and TRPV1 antagonists
LPS is thought to be a mediator of neurogenic inflammation (Hou et al. 2003, Diogenes et al. 2011, Ferraz et al. 2011) initiated by neuropeptide release. Whether P. gingivalis LPS could mediate neuropeptide release from the TG neurons in vitro was therefore determined.
TG cultures were either stimulated with vehicle, P. gingivalis LPS (20 µg/mL) or capsaicin (1 µM) for 15 minutes at 37°C and the resulting levels of CGRP released into the culture medium determined by ELISA. P. gingivalis LPS significantly increased CGRP release from TG neurons (Figure 3, p<0.001) as did treatment with the positive control, 1 µM capsaicin ( Figure 3).
Since the calcium influx in response to LPS could be abolished by the TRPA1 blocker, and partially removed by TLR4 signalling inhibition, whether TLR4, TRPA1 and TRPV1 inhibitors could prevent CGRP secretion in response to LPS was explored. CGRP release experiments were again performed by exposing TG cultures to LPS following pre-treatment with 3 µM HC-030031, 1 µM CLI-095 or 1 µM capsazepine. CGRP release from TG cultures in the presence of P. gingivalis LPS (20 µg/mL) was inhibited by all of three blocking reagents to the level obtained in vehicle-only stimulated samples (Figure 4, p<0.001).

P. gingivalis LPS induces NF-kB nuclear translocation in TG neurons and can be modified by TLR4, and TRPA1 antagonists.
P. gingivalis LPS is known to activate nuclear translocation of NF-kB in a range of cell types (Hashimoto et al. 2004, Herath et al. 2013. Consequently, whether LPS induces NF-kB nuclear translocation in neurons and/or the support cells present in the TG cultures was determined. Nuclear translocation of NF-kB was considered positive when the nuclear/cytoplasm ratio (nuc/cyt ratio) of NF-kB in neurons was >1, calculated using the methods of Hunot et al. (1997) and Kuiken et al. (2012). Immunofluorescence revealed that P. gingivalis LPS stimulation resulted in positive nuclear translocation of NF-kB in TG neurons and nonneuronal cells over a period of 30 minutes ( Figure 5). The percentage of neuronal cells positive for NF-kB translocation was significantly greater in LPS-treated cultures compared with control cultures (62% vs 18%, respectively; p<0.01). The percentage of NF-kB translocation-positive non-neuronal cells was also significantly greater following LPS treatment (67% vs 37% in stimulated and control cultures, respectively; p<0.05).
Further work showed that both neuronal and non-neuronal nuclear translocation of NF-kB in response to LPS could be inhibited by blocking TLR4 signalling (1 µM CLI-095), but not by blocking TRPA1 (3 µM HC-030031) or TRPV1 signalling (10 µM capsazepine). The percentage of neurons showing nuclear translocation of NF-kB in the presence of 1 µM CLI-095 inhibitor along with LPS decreased significantly from 65% to approximately 30%, and in non-neuronal cells it decreased from 65% to 35% (Figure 6; p<0.05).

Discussion
This study reports for the first time that P. gingivalis LPS directly activates TLR4 and TRPA1 receptors in trigeminal sensory neurons and non-neuronal support cells. The study also shows that this activation leads to release of the neuropeptide calcitonin-gene related peptide (CGRP) and activation of the NF-kB signalling pathway.
The rationale for the study comes from evidence that some bacterial pathogens can directly stimulate a nociceptor neuronal response (Chiu et al. 2013). There is relatively little information, however, about the neuronal response to oral Gram-negative bacteria and since P. gingivalis is commonly reported in association with symptomatic pulpitis (Rôças et al. 2002, Jacinto et al. 2006 (Rang et al. 1991, Chen et al. 2011, Chung et al. 2011, Tóth et al. 2011 Qi et al. 2011) and in the present study ultrapure P. gingivalis LPS directly activated more than 20% of TG neurons in the cultures, with a resultant increase in intracellular calcium. This is the first time that P. gingivalis LPS has been shown to directly activate TG neurons in vitro. Also, the neuronal response could be partially blocked by the specific TLR4 inhibitor, CLI-095, with the proportion of responding cells reducing from around 20% to only 5%.
LPS of Gram-negative oral anaerobic bacteria is known to elicit different host responses from those observed with the classic enterobacterial endotoxin (Darveau & Hancock 1983, Dixon relatively low biological activity because of its low phosphate content, altered length and positions of fatty acids in the lipid A moiety, compared with E. coli LPS (Takada & Kotani 1989). Also, Ogawa & Uchida (1996) have described differential cytokine production by mononuclear cells in response to P. gingivalis LPS. Notwithstanding these differences, P. gingivalis LPS was found to induce inflammatory responses in LPS non-responder mice (TLR4 knockout mice) (Kirikae et al. 1999), which suggests a different molecular recognition pathway is involved from that for E. coli LPS.
The other category of potential receptors is the transient receptor potential channels (TRPs).
Numerous studies have shown that TRP channels are a broad group of ion channels located on nociceptors; TRPA1 and TRPV1 are thought to be involved in responses to bacterial infection (Chung et al. 2011, Gibbs et al. 2011, Huang et al. 2012. TRPV1 is expressed in small and medium diameter nerve fibres throughout the human tooth pulp, which comprise about 20-35% of the TG neurons (Chung et al. 2011, Gibbs et al. 2011. The data in the present study support this, with 33% of TG neurons functionally responding to the TRPV1 agonist, capsaicin. As well as TRPV1, the receptor TRPA1 could also act as a sensor since it is known to respond to a broad spectrum of endogenous compounds and irritants. However, there has been a report that TRPA1 expression in the trigeminal ganglion is distributed in only a small population of TG neurons (6-10%; Huang et al. 2012). The present findings concur with this as it was noted the proportion of neurons in the cultures responding to the TRPA1 agonist CA was less than 15%.
The present study demonstrates that P. gingivalis LPS activates trigeminal neurons through TRPA1-dependent pathways, since the response was completely blocked by the TRPA1 inhibitor, HC-030031. This is consistent with the finding of Meseguer et al. (2014) (Coats et al. 2005, Herath et al. 2013. It is also possible that activation of TLR4 sensitises Since TRPA1 has been shown to control vascularity in neurogenic inflammation (Kunkler et al. 2011), levels of CGRP release in response to P. gingivalis LPS were measured. The present study demonstrated that a relatively high concentration of P. gingivalis LPS (20 µg/mL) could enhance neuronal activation and produce a significant increase in CGRP release (p<0.001). In contrast, Ferraz et al. (2011) reported that a low concentration (2 µg/mL) of P. gingivalis LPS failed to induce CGRP release. However, since only supraphysiological levels of LPS (here 20 µg/mL) triggered a neuronal response, it may indicate a different processing mechanism from that of the classical LPS receptor complex of TLR4 with MD2 and CD14 co-factors. With supraphysiological LPS concentrations, LPS signalling could occur via CD14-independent mechanisms (Ulevitch & Tobias 1995) and so the mechanisms revealed in this study may not completely represent the canonical pathway of TLR4-LPS interaction. Moreover, the effect of LPS on the release of CGRP could be reversed by not only TRPA1 or TLR4 signalling blockers, but also by a TRPV1 blocker. This finding may support the notion that P. gingivalis LPS somehow induces a TRPV1-dependent release of CGRP, which would be consistent with the findings of Diogenes et al. (2011).
These workers showed that TLR4 stimulation could enhance TRPV1 activation, leading to the release of CGRP. Even though there has been very limited evidence to show how LPS modulates peripheral sensitization, it has been thought that Ca 2+ permeation through TRPV1 leads to exocytosis, which underlies CGRP and SP release (Henrich & Buckler 2009) and which can account for a rapid neuronal response. However, it is also possible that CGRP release may also be influenced by the triggering of PKC/PLC pathways (Vellani et al. 2010), as a latent consequence. Taken together, the present data cannot rule out the possibility that LPS sensitivity of TG neurons requires TRPV1-mediated release of CGRP.
In order to have more prolonged effects on neurons, LPS would be expected to result in activation of the NF-kB pathway because this is the major transcription factor responsible for modulating gene expression in response to LPS (Arias-Salvatierra et al. 2011). Also, both TRPA1 and TLR4 mediate the production of inflammatory mediators through the NF-kB pathway and this pathway is thought to be related to voltage-gated channel over-expression (e.g. Ca 2+ , Na + channels) and possibly inducing allodynia, thermal hyperplasia, or chronic pain (Huang et al. 2006). Thus, the observed secretion of CGRP may relate to involvement of both these receptors (TRPA1 and TLR4), though CGRP secretion itself appears to involve the TRPV1-dependent pathway. These data lead us to propose a possible model for LPS stimulation of TG cells as shown in Figure 7.
Several previous reports have suggested that satellite glial cells could be a possible therapeutic target for modulation of pain (Watkins & Maier 2002, Takeda et al. 2009). This is because neuronal excitability could be developed and maintained through the sensory neuron along with the surrounding satellite glial cells acting as a functional unit (Takeda et al. 2009, Villa et al. 2010, Kushnir et al. 2011, Poulsen et al. 2014, Costa & Moreira Neto 2015, Hanani 2015. In the present study, TG neuronal supporting cells were directly activated by P. gingivalis LPS (e.g. using calcium influx and NF-kB nuclear translocation measurements). This is consistent with the finding of Yoon et al. (2012) who reported that E. coli LPS could increase expression of satellite cell markers in DRG. Taken together, the activation of neuronal supporting cells, such as satellite glial cells and/or Schwann cells, may be a crucial element in generation and maintenance of orofacial pain, however studies are currently being carried out to further characterize their contribution.

Conclusion
This study sheds light on the mechanisms involved in the pain pathways that can be directly activated by bacterial infection of the pulp. Contact between oral Gram-negative bacteria, such as P. gingivalis, and TG neuronal and non-neuronal cells can involve their direct activation through a combination of TRPA1, TLR4 and TRPV1 receptors. This results in NF-kB nuclear translocation and release of the neuropeptide CGRP.     Figure 1G, and is included here for ease of comparison with panels B and C. Figure   2D shows the percentage of LPS-responding neurons in three groups; no block, TLR4 block, and TRPA1 block.