Temporomandibular joint pain: A critical role for Trpv4 in the trigeminal ganglion

2.1. Animals

The pan-null phenotype of Trpv4^−/− mice [29] relies on excision of the exon encoding transmembrane domains 5–6. Mice were outcrossed to C57BL/6J background and PCR-genotyped. Male WT and Trpv4^−/− mice, 3–4 months of age, were used for all experiments, and bite force was also recorded in female mice of the same age.

Male dominant-negative MEK transgenic mice [41], 3–4 months of age, were used. The neuron-specific and pan-neuronal Tα1 α-tubulin promoter was used to drive the transgene. We documented expression of dnMEK in sensory neurons of the trigeminal ganglion (Fig. 6C).

Animals were housed in climate-controlled rooms on a 12/12h light/dark cycle with water and standardized rodent diet available ad libitum. All animal protocols were approved by the Duke University IACUC in compliance with NIH guidelines.

2.2. Induction of TMJ inflammation, neural tracing, and chemical injections

All mice were briefly anesthetized with 2% isoflurane and bilaterally injected using a 30G needle on a 10μL Hamilton syringe. To induce chronic inflammation of the TMJ, 10 μL of complete Freund’s adjuvant (CFA, 5 mg/mL; Chondrex) were injected, targeting both TMJs. Since the bite performed by the mice was an incisor bite, also to reduce variability of reduction of bite force, both TMJs were injected. Control animals were injected with the same volume of incomplete Freund’s adjuvant (IFA, Chondrex). To track TMJ innervation, mice were injected with 1.5 μL of neural tracer Fast-Blue (FB, 2% aqueous solution, Polysciences) into TMJ 15 min before CFA or IFA administration.

Moistened food pellets were made available to animals after induction of TMJ inflammation to decrease food consistency. To investigate the effects of the non-steroidal anti-inflammatory drug Ketorolac and the specific TRPV4 inhibitor HC-067047 on CFA-induced attenuation of bite force, mice received a single intraperitoneal administration of Ketorolac (7.5 mg/kg, dissolved in normal saline; Sigma) or HC-067047 (10 mg/kg, dissolved in 6% DMSO; Sigma) on day 1 after CFA injection, and were measured for bite force changes at 1h, 3h and 5h following treatment. Control animals received the same volume of NS or 6% DMSO. For testing the effect of the compounds on bite force in the absence of induced TMJ inflammation, mice received the same dose of Ketorolac or HC-067047, without concomitant TMJ injection.

2.3. Bite force and bite rate analysis

Mice were acclimated to testing facility and handling for 2 days prior to behavioral testing.

Bite force was measured using a custom-made bite force transducer designed as described previously [47]. Briefly, the transducer consists of two aluminum beams, each instrumented with two single-element strain gauges (TML UFLA-1-350-11-3LT). The four strain gauges are connected in a Wheatstone bridge. Deformation of parallel-mounted beams results in a proportional change in resistance and subsequently in the voltage output to the Wheatstone bridge. One end of each beam served as bite plate and was covered with an acrylic coating (Micro-Measurements) to protect the animals’ teeth. Prior to each use, the bite force transducer was calibrated and checked for linearity by suspending a series of calibration weights ranging from 0.1 to 0.5 kg from the bite plates. The voltage output from each weight was regressed against force exerted by calibration weights. Output for each set of calibrations was both linear and reproducible, with correlation coefficients (R²) ranging from 0.9901–0.9997 for each calibration. The distance between the beams of the bite force transducer was adjusted to 4.5mm (~40% of maximal jaw gape) at which mice can produce the maximum bite forces [47]. Mice were placed in a cylindrical tube with an opening at one end for accommodation of the mouse’s head. When the bite transducer was moved towards the mouse at 0.5–1cm/sec, a bite was invariably elicited. The voltage output during each bite was recorded as a continuous wave at 500Hz using Labview8.0 (National Instruments). The peak voltage of each bite was determined and converted into force (N) based on the regression equation derived from calibration. Each animal was tested 5x/testing day. The interval between two trials was >1 min. The highest force the animal generated per trial was recorded as bite force, then averaged for all trials.

Bite rate analyses were performed as previously described [39]. The biting rate was calculated as peak bite force magnitude divided by duration of biting. Two biting rates, i.e. loading (5% of peak force to peak; rate 1) and unloading (peak to 5% of peak force, rate 2), were determined.

2.4. Jaw morphometry

Considering bite force can be influenced by jaw-muscle anatomy [47], maximum gapes were determined immediately post-mortem by opening the jaws with a forceps. Care was taken not to avulse any of the soft and hard tissues of the TMJ. Using a digital caliper, we measured the maximum linear distance between the upper and lower incisor tips to the nearest 0.1 mm. After dissection, in addition, jaw length was determined as the maximum linear distance from the caudal surface of the mandibular condyle to the tip of the lower incisor.

2.5. Immunohistohemistry and morphometry

Routine procedures were followed [29]. Briefly, mice were perfused transcardially with 0.01M PBS followed by ice-cold 4% PFA at the experimental time-point under study. Their TGs were dissected and post-fixed in 4% PFA overnight, cryoprotected in 20% sucrose (48h) and sectioned on a cryostat at 12 μm. Brainstem sections were processed similarly and sectioned at 30 μm. Sections were blocked with 5% normal goat serum (Jackson), and incubated overnight with primary antibodies: rabbit anti-TRPV4 (1:300, Abcam), mouse anti-NF200 (1:10000, Sigma), mouse anti-peripherin (1:200, Millipore), guinea pig anti-CGRP (1:2000, Bachem), anti-IB4 (2μg/mL, biotinylated, Vector-Labs), rabbit anti-HA (1:500, Santa Cruz), rabbit anti-pERK (1:250, Cell Signaling Technology), mouse anti-pERK (1:600, Cell Signaling Technology; for double-labeling with TRPV4 in TG), and mouse anti-Brn-3a (1:200, Millipore; for brainstem sections only). Immunodetection was accomplished with secondary antibodies (AlexaFluor594-conjugated goat anti-rabbit; AlexaFluor488-conjugated goat anti-mouse or -guinea pig; all 1:500; Invitrogen) for 2h, and cover-slipped with Vectashield (Vector). Digital micrographs were acquired using a BX61 Olympus upright microscope equipped with high-resolution CCD camera and with constant acquisition/exposure settings, using ISEE software (ISEE Imaging Systems, Raleigh NC). TGs from Trpv4^−/− mice were also stained to validate specificity of TRPV4 immunolabeling.

For every animal under study, 4–6 TG sections were analyzed per mouse; neurons were identified by morphology. The cutoff density threshold was determined by averaging the density of three neurons per section that were judged to be minimally positive, using ImageJ software. All neurons for which the mean density exceeded the threshold >25% were counted as positive, and the positive cells were expressed as a percentage of total counted TG neurons.

2.6. Trpv4-copGFP transgenic mice

Generation and characterization of this mouse line have been reported previously [40]. Briefly, in a BAC transgene, the murine full-length Trpv4 promoter drives copepod-GFP. Deletion of exons 10–14 assures that a functional ion channel cannot arise. TMJ inflammation was induced in male transgenic mice, 3–4 months of age, and TGs were dissected on d3 post-injection. Native GFP-fluorescence was assessed in 12μm fresh-frozen sections, unfixed and embedded using fluoromount, imaged with constant acquisition/exposure parameters, from both CFA and IFA-injected mice. ≥25 TG neurons per animal were assessed by densitometry (ImageJ). For this, only neurons above a cutoff density (3x the density of the minimally fluorescent neurons in the respective animal) were evaluated. Customized regions-of-interest that did not include the nucleus were used. Mean densities were background-subtracted, and the average value per animal (from all neurons assessed) was determined to yield the group average.

2.7. Western blot

Routine procedures were followed [29]. Briefly, snap-frozen TGs were protein-extracted in CHAPS, and electroblotted to nitrocellulose membranes after gel separation of proteins in a 10% polyacrylamide gel. Membranes were blocked with 5% dry-milk, TRPV4 was specifically detected with primary antibody (rabbit anti-TRPV4, 1:1000; Alomone), secondary antibody (anti-rabbit peroxidase-conjugated, 1:5000; Jackson ImmunoResearch) and chemoluminescence substrate (ECL-Advance, GE Healthcare). Abundance was quantified using ImagePro Plus software. Specificity of the antibody was confirmed using Trpv4^−/− TG. Neuronal isoform of β-tubulin was detected with a mouse monoclonal anti-β-tub_III antibody (1:5000, Iowa Hybridoma Bank).

2.8. Quantitative real-time PCR

Total RNA was isolated from mouse TG using TRIzol reagent (Invitrogen). After purification, 5μg of total RNA were DNase I-treated using RNase-free DNase (Qiagen) followed by RNeasy Cleanup (Qiagen). 1μg of total RNA was then reverse transcribed using oligo primers (dT) and SuperScriptIII first-strand synthesis kit (Invitrogen). Gene expression was assessed by quantitative real-time PCR using 2× SYBR Green Master Mix (Applied Biosystems) and a two-step cycling protocol (anneal/elongate at 60°C, denature at 95°C). Specificity of primers was verified by dissociation/melting curve for the amplicons when using SYBR Green as a detector. All reactions were performed in triplicate. The amount of target messenger RNA (mRNA) in the experimental group relative to that in the control was determined from the resulting fluorescence and threshold values (Ct) using the ΔΔCt method. Primer sequences for “pain-TRPs” are shown in table below. β_III-tubulin was used as housekeeping gene.

2.9. Micro-CT scanning and histopathological analyses of TMJ

TMJ with soft connective tissues was carefully dissected and post-fixed in 4% PFA for 48h, and then placed in 70% Ethanol at 4°C until used. For quantitative 3-D evaluation of the mandibular condyle, the right TMJ was scanned by a micro-CT system (micro-CT 40, Scanco Medical AG). Samples were scanned at 55 kV with a slice increment of 16 μm. Calcified tissues were segmented from soft tissues using a threshold of 360. The head of the mandibular condyle was defined as the “cap-shaped” region at the end of the condyle and included approximately 90 slices. A 3-D reconstruction was generated of this region and analyzed using the internal software of the micro-CT system to determine bone volume (mm³), total volume (mm³) and calcified bone fraction (bone volume/total volume) for both WT and Trpv4 ^−/− naïve, day 13 IFA injected, and day 13 CFA injected mice.

To evaluate synovial inflammation, histological sections were prepared of the TMJs from samples that were scanned on the micro-CT. The joint samples were decalcified for 6 days at 4°C in Cal-Ex (Fisher-Scientific). Next, the TMJs were dehydrated and paraffin embedded in a Leica ASP300 tissue processor (Leica Microsystems Inc) followed by vacuum infiltration with paraffin for generation of 8 μm sections throughout the entire joint (sagittal plane). Sections from the central region of the joint were stained with Harris Hematoxylin and Eosin Phloxine (Poly Scientific) and then graded using a modified Gynther grading system [17], including synovial hyperplasia (0 to 4) for each synovial lining layer in the TMJ (for a maximum score of 16 for total synovial hyperplasia) and vascularity (0–4). Total hyperplasia was reported as the sum of the histological grading score for all four synovial membranes per TMJ. The synovitis score was determined as the sum of the total hyperplasia score and vascularity score for each TMJ. For all parameters, 0 is normal and the higher the score, the more severe the joint changes. Scores were determined by averaging the values from three blinded graders.

2.10. Statistical analysis

All data are expressed as mean ± SEM. Two-tail t-tests or one-way ANOVA followed by Tukey post-hoc test were used for group comparison. The histological scores of TMJ were not normally distributed, thus the non-parametric Kruskall-Wallis test was performed to determine significant differences. P<0.05 indicated statistically significant differences.

3.1. TMJ inflammation attenuates bite force in mice

We were able to accurately determine bite force in mice, using a custom-built force transducer fitted to their gape (Fig. 1A–B, Suppl. Fig. 1) [39; 47]. Mice bit with forces of 14–15N. The evoked bite appeared aggressive in nature, in response to the slowly approaching transducer. We believe the provoked aggression that prompted the bite to be a valid method to assess bite force because recorded responses were immediate, robust and reproducible.

TMJ-inflammation was elicited by microinjecting complete Freund’s adjuvant (CFA) (Fig. 1A–B), and control injections were with incomplete Freund’s adjuvant (IFA). With this protocol, we could observe a prolonged and significant attenuation of bite force over 9 days (Fig. 1B). Bite force returned to >80% of baseline within 2 weeks. To assess bite force dynamics, we determined bite rates, defined as increase/decrease in force over time, leading to loading and unloading rate (Suppl. Fig. 2) [8; 39]. Bite rate time-courses paralleled the time-course of maximum force because bite duration was not significantly changed (Suppl. Fig. 2). Body weight was reduced ≤10% (not shown). The reduction of bite force in the absence of a robust reduction in body weight suggests induction of masticatory sensitization rather than general inflammatory and anorexic effects of inflammation. Pain behavior in our TMJ-inflammatory model was responsive to systemic application of the non-steroidal-anti-inflammatory, ketorolac, confirming the validity of the model (Fig. 1C). Next, we assessed skeletal changes of TMJ-inflammation by micro-CT of the mandibular condyle and TMJ histology [9; 14]. The bone matrix was severely affected by TMJ-inflammation, evidenced by significant reduction of calcified bone fraction (from 0.83±0.01 to 0.58±0.04 for 13d IFA vs. 13d CFA; Fig. 1D). Joint histology showed increased synovial hyperplasia- (from 1.83±0.75 to 4±1.97) and synovitis (from 2.83±1.10 to 5.06±2.29) indices on d13 for IFA vs. CFA injections (Fig. 1E). These radiologic and histopathologic findings provide further evidence of the direct targeting of the TMJ with CFA injections, rather than surrounding masticatory muscles, because a CFA-induced myositis of masticatory muscles would not induce TMJ arthritic changes and mandibular condyle bone loss to the degree we have observed.

3.2. Trpv4 is necessary for attenuation of bite force in response to TMJ inflammation

We took advantage of this methodology to study the role of Trpv4 in TMJ inflammatory pain by comparing Trpv4^−/− pan-null mice [29] with their WT controls. The rationale for this is based on (1) the extraordinary level of mechanical “masticatory” allodynia present in TMJD; (2) the evidence for TRPV4’s involvement in mechanical signal transduction; (3) the record of TRPV4 in inflammatory pain conditions. First, we confirmed that TRPV4 is expressed in the TG (34.3% of total TG neurons), using specific antibody, revealing an expression pattern similar to that reported previously [28; 29; 46] (Fig. 2A–B and Fig 4B). We also conducted co-labeling experiments that revealed TRPV4-expressing TG neurons to be 27.3% peptidergic, 11.2% non-peptidergic, to express NF200 (marker of TG neurons with myelinated axons) in 67.2% and peripherin (marker of small-size TG neurons) in 22.1% (Fig. 2C). We then performed neural tracing with Fast-Blue (FB) injected into the TMJ (Fig. 2A). This resulted in labeling of 6.5% of all TG neurons, and ≈50% of those expressing TRPV4 (=3% TRPV4+/FB+). Thus, the TMJ is innervated by TRPV4-expressing TG sensory neurons. When we re-assessed the lineages in TRPV4+/FB+ TG neurons, we found 50.4% to express CGRP (peptidergic), 36.7% to express IB4 (non-peptidergic), 47.0% to express peripherin, and 56.0% to express NF200 (Fig. 2C). These data suggest that the TMJ is innervated robustly by TRPV4-expressing TG neurons, and that a substantial number of TRPV4-expressing, TMJ-innervating TG neurons might function as nociceptors. We also studied TRPV4 expression in sensory neurons of the proprioceptive mesencephalic trigeminal nucleus. We were unable to identify TRPV4-expressing neurons here (Suppl. Fig. 3).

Overall, these TMJ tracing- and TRPV4 expression studies provide a sound rationale for comparing Trpv4^−/− mice with WT controls in our inflammatory TMJD model. This approach will inform us whether Trpv4 contributes towards bite physiology as well as masticatory sensitization. Whereas, in the non-inflamed state, we were unable to identify differences between genotypes in TMJ anatomy (Suppl. Fig. 4), nor at the functional level for bite force and bite rates, striking differences became apparent in response to TMJ inflammation. This could be observed at similar levels and time-dynamics for both sexes (Fig. 3A). Attenuation of bite force was conspicuously reduced in Trpv4^−/− vs. WT, indicating that Trpv4 is necessary for development and maintenance of masticatory sensitization. We speculate that residual reduction of bite force in response to TMJ inflammation in Trpv4^−/− mice is sustained by other mechanisms. However, in a focused screen for “pain-TRPs”, we could not find an upregulation of any of the assessed genes in Trpv4^−/− mice (see below). Compensatory gene expression in the trigeminal system in Trpv4^−/− mice awaits in-depth future studies with genome-wide scope.

Both components of masticatory sensitization, reduction of bite force and -rate, were less prominent in Trpv4^−/− mice (Fig. 3A, Suppl. Fig. 5), suggesting a role for TRPV4 channels in the nocifensive response to TMJ inflammation. Confirming and extending this concept, we observed that a single systemic application of TRPV4-specific inhibitor, HC-067047 (as characterized for systemic use in mice before [13]), led to a similar resistance to reduction of bite force and -rate as genetic deletion of Trpv4 (Fig. 3B, Suppl. Fig. 5). Interestingly, HC-067047, applied 24h past CFA-injection of the TMJs, lasted over several hours to restore bite force (as expected from its half-life). Furthermore, it did not have an effect on the residual attenuation of bite force of Trpv4^−/− mice, indicating absence of off-target effects and confirming the compound’s specificity. Thus, both genetic deletion of Trpv4 as well as its temporary inhibition with a specific compound exert similar effects on pain behavior evoked by TMJ-inflammation. However, both methods do not specifically target the nervous system, a mandatory approach for future studies, given the present results. We did, on the other hand, examine carefully skeletal pathology by micro-CT and histopathology in Trpv4^−/− mice for comparison to WT (Fig. 1D–E, Suppl. Fig. 6). Surprisingly, skeletal changes of the mandibular condyle and joint histopathology were similar between genotypes (Trpv4^+/+ vs. Trpv4^−/−: 0.58±0.04 vs. 0.66±0.07 for bone fraction; 4±1.97 vs. 4.83±1.44 for total hyperplasia; 5.06±2.29 vs.6.41±1.46 for synovitis), showing signs of bone loss and synovial inflammation in Trpv4^−/− mice to similar degree as in WT. This result indicates that TMJ-injection with CFA leads to joint inflammation and profound skeletal changes irrespective of Trpv4 genotype, a remarkable “negative” finding in view of documented TRPV4 expression in synoviocytes, chondrocytes, osteoblasts, macrophages and endothelial cells [9; 10; 15; 19; 22; 35]. Consequently, this result prompted us to focus on TRPV4’s role in the TG.

3.3. Regulatory mechanisms in the trigeminal ganglion in response to TMJ inflammation

To better understand the relationship between bite force and Trpv4 we explored the trigeminal sensory innervation of the inflamed TMJ, and the dynamic TRPV4 expression in innervating TG neurons. After TMJ inflammation, TRPV4 was more abundant in TG homogenates, using Western blotting, peaking at d3 after induction of inflammation (2.25-fold increase), a remarkable coincidence with the time-course of bite force reduction (Figs. 1B, 4A). This time-course was confirmed via TRPV4-specific immunolabeling, which showed that protein up-regulation was strictly neuronal (Fig. 4B). This regulation was confirmed and extended in a Trpv4 reporter mouse [40], showing increased reporter-gene expression following TMJ inflammation, suggesting increased Trpv4 transcription as contributory (Fig. 4C). Immunostaining experiments demonstrated that the percentage of TRPV4-expressing TG neurons that innervate the TMJ approximately doubled in mice treated with CFA compared to control mice treated with IFA (2.95 vs. 5.5%, p<0.05; Fig. 4D), indicating that TMJ inflammation up-regulates TRPV4 in these neurons, which we found were medium- or small-sized (Fig. 4E). However, with TMJ inflammation, TRPV4-expressing sensory neurons that innervate the TMJ did not become more prevalent than in the TG overall (53.7% of TMJ-innervating neurons vs. 50.9% of all neurons). Moreover, TRPV4 up-regulation, from 34.3% to 50.9% (TRPV4+ of all TG neurons; p<0.001), was not restricted to TMJ-innervating TG sensory neurons. This pattern suggests a possible cross-sensitization between trigeminal branches. These findings suggest a delimited feed-forward mechanism whereby TRPV4-expression in TG sensory neurons is increased by TMJ inflammation in both TMJ-innervating neurons as well as others. These latter neurons might function in neurogenic inflammation (see Fig. 2A&C, CGRP-expressing neurons), while providing sensory innervation to neighboring structures, which might become hypersensitive/allodynic.

Beyond upregulation of TRPV4 expression in TGs by TMJ inflammation, we asked whether expression of other pain-related Trp genes is regulated by TMJ inflammation or Trpv4 deletion, using qRT-PCR: Trpv1, Trpv2, Trpv3, Trpa1, Trpm2, Trpm3 and Trpm8 [6; 26; 33]. For WT, not Trpv4^−/−, there was a significant increase with inflammation for TRPV1 (1.88-fold increase) and TRPA1 (1.52-fold increase) (Fig. 5). For TRPM2 and TRPM3, there was significant down-regulation in Trpv4^−/− vs. WT with inflammation (4.35- and 1.54-fold decrease for TRPM2 and TRPM3, respectively). Thus, the regulated expression of these “pain-TRPs” in the TG after TMJ inflammation was dependent on Trpv4. For TRPM3, this was also the case in non-inflamed controls. These findings indicate that TRPV4 regulates gene expression of other “pain-TRPs” in the TG after TMJ inflammation.

Overall, our findings imply that Trpv4 is critical for masticatory sensitization in TMJ inflammation, that a key locale is the TG where TMJ-innervating sensory neurons express TRPV4, which is upregulated by TMJ-inflammation, and that Trpv4 controls gene-expression of other “pain-TRPs” in the TG.

3.4. Critical role of MEK-ERK MAP-kinase activation and its TRPV4-dependence in the trigeminal ganglion in response to TMJ inflammation

Given the known role of extracellular-signal-regulated-kinase (ERK) signaling in nociceptive sensitization [20] and the documented role of MEK-ERK phosphorylation down-stream of TRPV4 in epithelial cells via Ca⁺⁺ influx and RAS-RAF signaling, and also in sensory neurons that innervate the gastrointestinal tract [25; 27], we asked whether ERK activation (phosphorylation, 4.21±0.18% of total TG neurons for control) was increased in TG neurons following TMJ inflammation. Phosphorylated-ERK (pERK) labeling was increased early and robustly (8.93±0.37%, 7.92±0.59% and 7.38±0.92% at 3h, 1d and 3d CFA, respectively) after injection of CFA, and expressed in TG sensory neurons (Fig. 6A). TRPV4 was co-expressed in pERK-expressing TG neurons that innervate the TMJ (Fig. 6B), for example co-expression increased significantly after TMJ-inflammation (on d3, 9% for IFA-injection, 21% for CFA-injection). This finding suggests pERK-mediated re-programming of TG sensory neurons in TMJ inflammation that involves TRPV4-signaling [4]. To test this concept in-vivo, we subjected dn-MEK-transgenic mice, which are incapacitated to phosphorylate ERK exclusively in their neurons, to TMJ inflammation [41]. Remarkably, their bite force reduction over time resembled that of Trpv4^−/− mice (Fig. 6C). Since the dnMEK-transgene is driven by an exclusively neuronal promoter, and since we verified transgene expression in the mouse TG (Fig. 6C), our findings suggest that a critical signaling mechanism for nocifensive behavior is operative in TG sensory neurons, not further up-stream in the pain pathway, not in astrocytes (where Trpv4 is also expressed [5]) or activated microglia in pain transmission. Importantly, the TG is a critical locale, rather than its peripheral targets because the dnMEK-transgene is only expressed in neurons, meaning both peripheral sensory neurons and within the CNS. We therefore focused on the TG and examined pERK-expression over time after TMJ-inflammation in WT, Trpv4^−/− and dnMEK-transgenic mice. We found more than doubling of pERK expression in WT 3h after CFA injection, which remained elevated until d3. In contrast, both Trpv4^−/− and dnMEK-transgenic mice failed to up-regulate pERK, suggesting that TRPV4 functions upstream of ERK phosphorylation in TG sensory neurons. No significant differences in basal levels of pERK in TG sensory neurons were detected between WT, Trpv4^−/− and dn-MEK-transgenic mice (data not shown)