OSCC-released BDNF contributes to oral cancer pain via peripheral TrkB activation

BACKGROUND

Quality of life of head and neck cancer (HNC) patients is notoriously poor due to the greater prevalence of pain compared to other cancer types [15; 43]. Pain is not only the top ranked symptom for these patients, but unachievable management of this co-morbidity is what drives patients to seek initial treatment [5; 16]. These patients have limited options due to adverse effects of opioids and have higher opioid reescalation index [34; 55]. Mechanisms by which HNC produce pain in not yet clearly understood; however, it is speculated that the greater prevalence and severity of pain is due to the heterogenous properties of HNC tumors and the physiology of orofacial tissues [21]. Oral squamous cell carcinomas (OSCC) located in the tongue or floor of the mouth, report the highest pain severity, originating from site of primary tumor development even when the tumor is small in size [8; 9; 15; 32]. Spontaneous pain, pain while chewing and swallowing, and referred pain across the orofacial area is frequently reported by these patients [8; 9].

We and others have shown that oral tumors release soluble factors, that directly regulate sensory neuronal activities. [41; 43; 49]. Neurotrophins such as the nerve growth factor (NGF) are one such class of growth factors that are expressed at elevated levels in oral tumors, mediate cancer-promoting effects as well as promote increased sensory neuronal transmission contributing to oral cancer pain [56]. Similarly, the expression of brain-derived neurotrophic factor (BDNF) and its receptor TrkB; has been reported to be significantly increased in HNC tumors and contribute to tumor progression by promoting epithelial to mesenchymal transition, preventing apoptosis, and inducing chemoresistance [24; 26; 60]. However, the role of BDNF signaling in oral cancer pain is not clearly defined. We previously reported that BDNF expression was increased in murine trigeminal neurons post tongue tumor growth and systemic inhibition of its receptor TrkB reversed pain behaviors induced by oral tumors [6], suggesting that BDNF signaling contributes to oral cancer pain. This report was in accordance with many studies demonstrating that BDNF is upregulated in sensory neurons and significantly contributes to several pain types including neuropathic, inflammatory and bone cancer pain [1–3; 39; 48].These studies have primarily focused on central mechanisms of BDNF which posts a therapeutic barrier due to its integral role in the development and homeostasis of central nervous system [2; 3].

Recently, it was revealed that BDNF signaling can also mediate pain through a peripheral mechanism in an osteoarthritis (OA) pain model [18]. Since clinical and preclinical models of oral tumors have shown to have elevated expression of BDNF and TrkB, the current study evaluated whether peripheral BDNF signaling expressed by the oral tumors could contribute to oral cancer pain. We assessed BDNF expression in tongue tumors, whether local blocking of BDNF or TrkB reversed oral tumor pain, identified lingual fibers regulated by TrkB signaling, characterized expression of TrkB isoforms in lingual sensory neurons and assessed the effect of TrkB knockdown in sensory neurons on tumor-induced pain behaviors.

MATERIALS AND METHODS

Animals

Six- to eight-week-old adult inbred Balb/c male and female athymic nude mice (Charles River, Wilmington, MA, USA) were used for all experiments. All animals were used at 20–22gms of body weight. Animals were housed at 4–5 per cage for at least four days prior to the start of any experiments. All animal experiments were approved by the UTHSCSA IACUC and conforms to IASP guidelines.

Cell Lines

The human OSCC cell lines, HSC2, HSC3, and HSC4 were purchased from Health Science Research Resources Bank, Japan. The cell lines were cultivated and maintained in Dulbecco’s minimal essential medium (DMEM, Life Technologies, Carlsbad, CA) supplemented with 1x glutamine, 1x penicillin/streptomycin, and 10% fetal bovine serum. Human normal oral keratinocyte (NOK) cell line, OKF6-TERT2, was kindly provided by Dr. Cara Gonzales of UTHSCSA and maintained in keratinocyte serum-free media (Life Technologies, Carlsbad, CA) supplemented with 25 μg/ml bovine pituitary extract (BPE), 0.2 ng/ml epidermal growth factor (EGF), and 0.4 μM calcium chloride and penicillin/streptomycin.

Drugs

Recombinant Human IgG isotype control and TrkB Fc Chimera (R&D Systems, Minneapolis, MN) were diluted to 100ng/ul stock solutions in 0.1M phosphate-buffered saline (PBS) and further diluted to 400ng in 20ul of PBS for in vivo use. The TrkB receptor antagonist, ANA-12 was purchased from Sigma Aldrich (St. Louis, MO, USA) and diluted to 120ng/ul in 100% DMSO. This stock was further diluted 3ug per 20ul injection in 1% DMSO/0.6% Tween-20/PBS. Recombinant BDNF was purchased from PeproTech (Cranbury, NJ, USA, catalog Number 450–02) and diluted to 0.33ug/ul in sterile water.

In vivo orthotopic xenograft model

We have previously published our work with this model [6]. Briefly, OSCC cells were harvested in growth media and prepared at a concentration of 3.5 × 10^5 cells/50 μL. Mice were anesthetized using 2% isoflurane and 50 μl was injected into the ventral tongue on the anatomical left side, over a 30-s period using a 25-gauge needle. Control groups consisted of injection of NOK cells into the tongue at identical cell concentrations. Following injection, animals were immediately placed in warm bedding and allowed to recover. Prior to assigning animals to experimental groups, every animal was tested for observable tumor growth. To test this, animals were anesthetized with isoflurane inhalation and their mouths were opened to pull the tongue out with blunt forceps and quickly evaluate for pinkish-white lump on the injected side of the tongue. Our prior study depicts the gross appearance of HSC2 tumors in athymic mice as well as the histology of the tumor with H$E staining [6]. Till date, we have observed tumor-growth in 100% of HSC2-injected animals.

Tumor Volume Measurement

Mice were anesthetized using 2% isoflurane and caliper measurements for length, width and height of the tongue tumors were recorded and tumor volumes were determined using the formula π/6 × (length × width × height)[47; 52] and presented in mm³. All tumor measurements were made by blinded observers.

Western blot

HSC2, HSC3, HSC4 and NOK cells were seeded in a 6-well plate and cultured for 24 hrs. Media was then aspirated, washed with 0.1M PBS, and 1 ml of each cells line’s respective serum free media was added. Cells were harvested at 48 hours. At the time of collection, wells were placed on ice, washed with cold 0.1 M PBS, and lysed by adding 500ul of T-PER (ThermoFisher Scientific, Carlsbad, CA) with cOmplete mini protease inhibitor (Millipore Sigma, St. Louis, MO) for 30 minutes. Cells were then scraped, and samples were centrifuged at 8,000 × g for 5 minutes at 4C to pellet cell debris. Supernatants were collected in fresh tubes and stored at −80C until use. For cell lysates of tongue and TG tissues, tissues were lysed with 200ul of RIPA buffer containing the protease cocktail inhibitor and homogenized in a Qiagen Tissue Lyser (Biorad, Hercules, CA, USA) at 50 oscillations per sec for 1 min, incubated on ice for 15 mins and repeated the step once. Homogenates were incubated on ice for additional 30 mins and centrifuged at 14000g for 10mins at 4C. Supernatants were collected in fresh tubes and stored at −80C until further use. Total protein determination was quantified following company’s protocol for Pierce^™ BCA Protein Assay (ThermoFisher Scientific, Carlsbad, CA). Samples were then prepared and subjected to SDS-PAGE on 4–12% gradient Bis-Tris gels according to the NuPAGE protocol (Novex, San Diego, CA). Proteins were transferred to PVDF membranes via iBlot2 and prepared for immunoblotting according to company’s protocol (ThermoFisher Scientific, Carlsbad, CA). PVDF membranes were washed for 5 minutes in ddH2O and incubated with a 1:1 dilution of 0.1M Tris Buffered Saline (TBS) and Intercept Blocking Buffer (LI-COR, Lincoln, NE) for 90 minutes at RT. Primary antibodies diluted in blocking solution were added to PVDF membranes and rocked for 16hr at 4C. Anti-human BDNF(1: 1000, cat log # 28205–1-AP, Proteintech, IL, USA) Anti-mouse TrkB, (Cat log # 4607, Cell Signaling, MA, USA), anti-actin (1:25000, cat log# ab49900, Abcam, Cambridge, MA) and anti-GAPDH (1:1000, Cat log # 2118S, Cell Signaling) primary antibodies were used. After primary staining, PVDF membranes were rinsed 4 × 5 minutes with 0.1M Tris Buffered Saline + Tween (TBST) and incubated with secondary antibodies (1:5000, Amersham ECL HRP-linked IgG antibodies) diluted in blocking solution for 90 min at RT. Membranes were then washed 4 × 5 minutes 0.1M TBST and 2 × 5 minutes TBS, and air dried in the dark. Proteins were visualized using Clarity^™ Western ECL Substrate (Bio-Rad, Hercules, CA) per manufacturer’s instructions. Chemiluminescence was captured using Bio-Rad Chemi Doc.

ELISA

All ELISA measurements were obtained using Human BDNF Elisa kit (RayBiotech, Peachtree Conters, GA). For ELISA of cell lines, cells were harvested as described above for western blot and cell lysates were subjected to ELISA per manufacturer’s protocol. For tumor cell lysates and superfusates, animals were injected with HSC2 or NOK cells and tongue tissue harvested at day 9 or day 14 post-cell inoculation. Tissues was weighed, cut, and placed in 500ul of serum free/colorless media for 2 hours at 37C. Superfusates and tissue were separated and stored at −80C until use. Total Protein in superfusates was determined using BCA kit and BDNF release in superfusate samples was quantified following company protocols for the ELISA kit. For tumor tissue lysates, the tissues were subjected to cell lysate preparation as described above for western blot and subsequently subjected to ELISA following company protocols.

Immunohistochemistry

Mice were anesthetized with an i.m. injection of ketamine (75 mg/kg) and dexmedotomidine (1 mg/kg), and transcardially perfused with 30 ml of fixative consisting of 4% paraformaldehyde (PFA) in 0.1 M Phosphate Buffer (PB). Tissues were dissected and post-fixed in 4% PFA for 20 min. washed 3 × 15 min in 0.1M PB, immersed in 10% sucrose at 4°C overnight and stored in 30% sucrose at −20°C. For cryo-sectioning, tissues were thawed and placed in Neg-50 (Richard Allan, Kalamazoo, MI, USA) prior to freezing on dry ice. Tissues were sectioned in the horizontal plane at 20um for trigeminal ganglia and 30um for tongue and placed onto Superfrost Plus slides (Fisher Scientific, Waltham, MA, USA), dried, and stored at −20C until stained. Immunostaining was performed as described previously[6; 40]. Briefly, tissue sections were washed with 0.1M PB 3 × 5 minutes and incubated with blocking solution consisting of 4% normal goat serum (Sigma, St. Louis, MO), 2% bovine gamma-globulin (Sigma-Aldrich, St. Louis, MO) and 0.3% Triton X-100 (Fisher Scientific) in 0.1M PBS for 90 min at RT. Following, tissue sections were incubated with primary antibodies diluted in blocking solution for 16 h at RT. 1:150 of anti-BDNF (Proteintech) and anti-TrkB (Cell signaling) primary antibodies were used. After primary staining, sections were rinsed 3 × 5 minutes with 0.1M PBS and incubated with secondary antibodies in blocking solution for 90 min at RT. Secondary antibodies were purchased from Molecular Probes, Eugene, OR, USA. Tissue sections were then washed 3 × 5 minutes in 0.1M PBS and 2 × 5 minutes in ddH2O, air-dried in the dark, and cover-slipped with Vectashield Antifade Mounting Medium with DAPI (Vectorlabs, Burlingame, CA, USA). Sections were evaluated and images obtained with a Nikon Eclipse 90i microscope equipped with a C1si laser scanning confocal imaging system. Multiple images were acquired of tongues and the V3 region of TGs with a 20× objective and identical laser gain settings across groups. Laser gain settings were determined such that no-primary control did not show any positive staining. Images were taken using fixed acquisition parameters across all groups and were unaltered from that initially taken.

Preparation of Conditioned Media

HSC2 cells were cultured in a 24-well plate at 8000 cells/well. Cells were plated in complete growth media for the first 24 hours. Next day, media was removed and cells washed with 0.1M PBS and incubated in colorless DMEM (Life Technologies, Carlsbad, CA) supplemented with 1x glutamine, 1x penicillin/streptomycin without serum for 72 hours following which conditioned media (CM) was collected and centrifuged at 2000 rpm for 5 minutes. Prior to use of CM for behavioral experiments, CM was sequentially filtered with centrifugal filter units of 100K and then at 10K (Amicon Ultra, 0.5mL, Ultracel-100K, 10K) to remove proteins over 100kDa and smaller than 10kDa.

Feeding Behavior

We previously have published the feeding model as a way to measure pain behaviors for tongue cancer where we showed that food intake decreases with time as the tumor progresses[6]. Furthermore, this reduction in food intake is reversed by administration of analgesics such as NSAIDs and opioids suggesting that the reduction in food intake is due to tumor-induced nociceptive behaviors. The observed feeding behavior is not due to the tumor-induced obstruction of the oral cavity for three reasons: 1) The fact that we observe increase in food intake post-acute drug administration (NSAIDs and Opioids) suggests that the feeding behavior is not due to tumor obstruction; 2) injection of colon cancer cells that also produces a tumor in the tongue, does not show the same feeding behavior as OSCC cells suggesting that having a progressive mass within the oral cavity is not enough to reduce food intake. These findings are reported in our prior publication [6] and 3) In our lab, we have observed that HSC2 injected animals are able to intake soft diet ( i.e. crushed pellets softened with water) significantly more at times when they aren’t able to eat very much solid diet (usually by day 13–14 post cell-inoculation), indicating that the obstruction of tumor does not hinder their ability to feed themselves. This observation was made very early on while establishing the model in our lab. Taken together, our observations and findings indicate that reduction in feeding behavior by OSCC tongue tumor is at least in part due to pain-like symptoms in mice.

Briefly, mice were placed in individual cages with bedding and food deprived for 18 hours. Each mouse was then placed in an individual clean cage without bedding and allowed to freely feed for 1 h on a pre-measured amount of standard chow. At the end of an hour, the amount of food intake was determined by calculating the difference (in g) from the initial amount and the remaining amount of food. Baseline measurements were obtained prior to randomizing the animals to experimental groups. All experiments were randomized using stratified randomization protocol to obtain similar average baseline values between groups and performed by blinded observers.

von Frey thresholds in the vibrissal Pad

von Frey filaments were applied to the V2 trigeminal division (i.e., vibrissal pad) and head withdrawal, face swipe, or grimace were considered a positive response as previously described [6]. Briefly, animals were first acclimatized to the von Frey filaments by exposing each animal to the lowest force for 5 times on one day prior to experimentation. Baseline readings and post-experimentation testing was performed by applying a sequential series of calibrated von Frey filaments (bending forces ranging from 0.008 to 6g of force) to the vibrissal pad. Each force was applied for 5 times at the interval of 30–60 seconds and percentage of responses recorded. Post baseline readings, animals were randomized to experimentation groups using stratified randomization protocol. Observers were blinded to all treatment paradigms.

Single Fiber Tongue-Nerve Electrophysiology

The tongue-lingual nerve preparation was readied identical to the approach previously described[19]. The dissected tongue was cut at midline on the ventral side and the entire preparation was immediately placed in a perfusion chamber subjected to constant flow of homeostatic medium consisting of 123 mM NaCl, 9.5 mM Na-gluconate, 5.5 mM glucose, 7.5 mM Sucrose, 10 mM HEPES, 3.5 mM KCl, 2 mM CaCl2, 0.7 mM MgSO4, 1.7 mM NaH2PO4, pH 7.4, held at 95% oxygen and 5% carbon dioxide saturation and a temperature of 32C. The tongue was opened up from the midline to expose the nerve endings and pinned down in the silicone chamber with the dorsal side facing down. The lingual nerve was then passed into the smaller recording chamber containing the recording electrodes. The lingual nerve was desheathed and teased into fine filaments for extracellular recordings using silver-wire recording electrodes (World Precision Instruments, Sarasota, Florida, USA). Receptive fields for mechanically sensitive fibers were identified by probing the tongue with a blunt glass rod. Latencies of action potentials were determined using electrical stimulus as described previously[19]. To submerge the tongue in ANA12 or Veh, the buffer pump was paused, and the immediate buffer solution was drained from the tongue chamber and replaced with 0.06ug/ul of 1.5–2ml of drug solution. After 5 minutes, the drug solution was removed and superfusion was re-established with the tongue chamber. Following this, von Frey threshold of the fiber using calibrated filaments and mechanical impulses to increasing force using the automated step and hold ramp (10–200mN) was determined, as described previously[19]. Conduction velocity (CV) of each fiber was determined by dividing the distance between the receptive field and the recording electrode by the electrical latency of the action potential. Fibers were classified by their conduction velocities as followed: fibers conducting slower than 1 m/s were identified as C-fibers and those conducting between 1 and 3.0 m/s were A-slow. High Threshold Mechanoreceptors were identified as those that produced increased number of impulses with increase in force.

Retrograde labeling

Mice were briefly anesthetized by isoflurane inhalation and tongue innervating lingual neurons were labeled by bilateral 5ul injections of 1% WGA-488 (ThermoFisher Scientific, Carlsbad, CA) in naive mice as described previously[40]. To maximize labeling of lingual neurons, WGA was injected twice on each side of the tongue; once superficially and once deep in the tissue. The two injections were spaced at a 4h interval. Mice were used for experiments two days later.

Single Cell RTPCR

Experiments were performed based on protocols described previously [40]. Animals were retrogradely labeled with WGA-488 and 2 days later, TG tissues were dissected and dissociated in a sterile petri dish. Neuronal cells of three sizes (small for <15 microns, medium for <25 microns and large for >25 microns) were manually picked using a Narishige micromanipulator and FERTY Syringe Plus Microinjector under an inverted EVOS FL digital fluorescence microscope (AMG, Bothell, WA). Single cells were placed into 0.2ml RNase free PCR tubes with 4μl of 2X reaction mix from CellsDirectTM One-Step qRT-PCR Kit (ThermoFisher Scientific Carsbad, CA) subjected to RT-PCR as previously described[40] for CGRP, NFH, Full-length TrkB and truncated TrkB (Table 1).Ct value per gene per cell was normalized to its internal control gene, UBB. Relative gene expression levels were obtained using 2ˆ (-delta Ct) formula.

Table 1:

Genes used for Single-Cell RTPCR

Gene	Catalog Number
CGRP	Mm00801462_m1
NFH	Mm01191455_m1
Full Length TrkB	Mm01341761_m1
Truncated TrkB	Mm01341751_m1

Intra-ganglionic Injections

All intra-ganglionic injections followed previously established protocols [50; 54]. Mice were anesthetized with an i.m. injection of ketamine (75 mg/kg) and dexmedotomidine (1 mg/kg). Animals were placed in the Model 940 Small Animal Stereotactic Apparatus (Kopf, Tugunga, CA). The scalp was disinfected with 10% betadine and allowed to dry. A 1cm medial incision was made and the exposed skull was treated with hydrogen peroxide. Bregma was located and the Neuros syringe was placed over X (coordinates of 2.20mm for the left side and 2.01mm for the right side) and Y(coordinates of −0.740 for the left side an 0.740 for the right side) coordinates and lowered until near contact was made with the skull. Where the needle nearly touched, a 3mm² burr hole was made without puncturing blood-brain barrier. The Neuros Syringe was then slowly lowered to Z(coordinates of −6.8 mm for both sides) and 5ul of siRNA (5ug) was delivered manually at a rate of 1ul/minute. This was repeated in both TGs. The wound was closed using 5–0 perma-hand silk sutures (Ethicon, Somerville, NJ) and covered in topical bacitracin ointment. Animals were put back in individual cages on heating pad with food and water to fully recover.

siRNA knockdown of TrkB

Knockdown of all TrkB isoforms was performed using intraganglionic injections of Accell SMARTpool small interfering RNA (siRNA) (Horizon Discovery, Lafayette, CO). Control groups received non-targeting scrambled siRNA. Lyophilized siRNA was resuspended in RNase free water and 5ug of TrkB or scrambled siRNA was injected bilaterally into each TG in 5uls volume. Tissue harvestation or experimental testing was performed 5 days post injection.

Statistics

GraphPad Prism 8.0 (GraphPad, La Jolla, CA) was used for statistical analyses. Data are presented as mean ± standard error of mean (SEM). Differences between groups were assessed by unpaired Student’s t-test, one-way or two-way ANOVA with Bonferroni’s correction. Statistical significance was established based on a two-sided alpha of 0.05 for all tests. Sample sizes were designed using G* power and calculated based on effect size and standard deviation obtained in our initial preliminary data, to generate an 80% power at a two-sided tail with α error probability of 0.05.

RESULTS

BDNF is expressed in human OSCC cells

To determine the role of BDNF in oral cancer pain, we first tested whether OSCC cells and tumors expressed BDNF. As seen in Fig 1A inlet, cell lysates of three different OSCC cells showed detectable expression of BDNF as tested by western blot and this was confirmed quantitatively by ELISA. Moreover, BDNF expressed in all three cell lines was significantly higher compared to NOK cells (4.35-fold higher in HSC2, 4.92-fold higher in HSC3 and 2.84-fold higher in HSC4) (Fig 1A). Given the expression of BDNF in all three cell lines, we selected HSC2 cells for the rest of the experiments for the study. Using HSC2 in vivo orthotopic tongue cancer model as described previously[6], we visualized BDNF expression in HSC2 tumors in vivo using immunohistochemistry. Our data showed that BDNF was indeed expressed in HSC2 tumor cells within tongue tissue and these cells were in close proximity to sensory nerve endings innervating the tumor as observed by NFH staining of lingual nerve terminals (Fig 1B). We then asked if BDNF expression within the tumors increased with tumor progression. As shown in Fig 1C, cell lysates of tumor biopsies taken at day 14 post HSC2 cell-inoculation had significantly higher expression compared to Day 9 post-cell inoculation (2.65-fold higher at day 14, 1-way ANOVA, p= 0.0177). Normal tongue had minimal level of BDNF expression. Lastly, we determined whether BDNF is released from tumor biopsies by measuring BDNF levels in superfusates from normal tongue and tumor-bearing tongue harvested at day 9 and at day 14 post cell-inoculation and our data revealed that similar to expression in whole cell lysates, BDNF was expressed at detectable levels from superfusates of tumor-tongue tissues compared to normal tongue and that BDNF levels were released at higher levels with tumor progression (3.08-fold higher at day 14 compared to day 9, 1-way ANOVA, p=0.0343) (Fig 1D). Taken together, our data demonstrate that BDNF is expressed in several human tongue cancer cell lines as well in tongue tumors in vivo, is increased with tumor progression and may be released in the extracellular space within the tumors.

Fig 1. BDNF Expression in human OSCC cells.

A (inset). HSC2, HSC3 and HSC4 cells were grown in serum free media for 48 hours following which cells were harvested for basal expression of BDNF by western blot. A. HSC2, HSC3 and HSC4 and NOK cells were grown in serum free media and harvested at 48 hours. Whole cell protein lysates prepared and subjected to ELISA for human BDNF. n=3 samples per group. B HSC2 cells were injected in athymic mice and at day 9 post cell inoculation, animals were perfused with 4% PFA and tongue tissue sectioned for immunohistochemistry. Sections were stained for BDNF, Neurofilament heavy chain (NFH) and DAPI and imaged at 20X magnification. Image shown is within tumor of the tongue tissue. n=3 was used. Animals were injected with HSC2 cells or NOK cells and tongue tissue harvested at day 9 and at day 14 post- cell inoculation for C. whole cell lysate preparation or D. superfusate preparation. Lysates and superfusates were subjected to ELISA for human BDNF. n=3 per group. Data are represented as mean±SEM. Data analyzed with one-way ANOVA with Bonferroni post hoc test at p<0.05. Horizontal dotted line in C and D indicate levels of BDNF in NOK-injected tongue cell lysate and superfusate respectively.

Blocking BDNF in the tongue reverses tongue tumor-induced pain behaviors in vivo.

To determine whether expressed BDNF within tongue tumors contribute to OSCC-induced pain, we used our previously published feeding behavior[6] and neutralized BDNF using local injection of TrkB-Chimera Ab that binds to BDNF in extracellular space and prevents it from binding to its receptor TrkB. Injection of NOK cells or control IgG in the tongue had no effect on feeding behavior at day 9 post-cell inoculation compared to baseline (BL) feeding as previously reported (0.475g ±0.048 IgG vs 0.500g±0.041 BL, 2-way ANOVA p=0.915) (Fig 2A). Also, Injection of the chimera Ab in the tongue (400ng/20ul injection) of NOK-injected animals at day 9 post-cell inoculation, had no effect on food intake (0.440g±0.021 TrkB Chimera vs 0.56g±0.024 BL, 2-way ANOVA, p=0.1445) (Fig 2A). However, animals bearing a tongue tumor and injected with control IgG produced significant reduction in food intake compared to baseline (0.281g ±0.019 IgG vs 0.463 ±0018 BL, 2-way ANOVA, p<0.0001) (Fig 2B) as reported previously and injection of the chimera Ab (400ng/20ul injection) significantly reversed this tumor induced reduction in feeding behavior (Fig 2B) compared to injection of control Ab (0.411g±0.0132 TrkB Chimera versus 0.28, g± 0.019 IgG Control, 2-way ANOVA, p=0.0020). TrkB chimera Ab had no effect on tumor size or body weights tested at day 9, 11 and 14 post cell inoculation compared to control IgG (Fig S1A and B). These data suggest that BDNF in the extracellular space of tongue tumors contribute to cancer-induced pain. Notably, injection of recombinant BDNF in the tongue had no effect on feeding behavior in naïve mice (Fig. S2).

Fig 2. Effect of BDNF neutralization on OSCC-induced pain behaviors.

Following baseline (BL) feeding measurements, athymic male mice were injected with A. NOK cells or B. HSC2 cells and at day 9 post cell inoculation, animals were injected with 400ng/20ul injection, of control IgG or TrkB chimera Ab in the tongue and 1 hour following injection, food intake was measured for an hour. n=6–10 per group. Data are represented as mean±SEM. Data analyzed with one-way ANOVA with Bonferroni post hoc test at p<0.05.

Peripheral TrkB receptor contributes to oral cancer pain behaviors in vivo.

Because TrkB is a specific receptor for BDNF signaling, we asked whether inhibiting TrkB reverses oral cancer pain similar to the observed effect with neutralizing BDNF within oral tumors. We antagonized the receptor within the tongue using a specific antagonist ANA12. Injection of ANA12 (3ug/20ul injection) within the NOK-injected tongue had no effect on food intake at day 9 post-cell inoculation (0.420g ± 0.086 ANA12 vs 0.417g ±0.083 Veh, 2-way ANOVA, p=0.9991) (Fig 3A); however, ANA12 significantly reversed tumor-induced reduction in food intake compared to the vehicle injected group at day 9 post-cell inoculation (0.405g±0.048 ANA12 vs 0.260g Veh ±0.026, 2-way ANOVA, p=0.0089) (Fig 3B). To further evaluate whether OSCC-released BDNF contributes to tumor-induced pain behaviors, we injected conditioned media (CM) from HSC2 cells into naïve animals in the tongue to test for feeding behavior as an affective assay or in the vibrissal pad to test for von Frey thresholds in the face as a reflexive nociceptive behavior assay. As seen in Fig 3C, CM injection in the tongue produced reduction in food intake compared to baseline (BL) (0.243g ±0.057 CM vs 0.543 ± 0.035g BL, 2-way ANOVA, p<0.0001). Moreover, local pretreatment of ANA12 (3ug/20ul injection) prior to CM injection, prevented CM-induced reduction in food intake (0.411g ± 0.052 post CM vs 0.483g ±0.022 BL, 2-way ANOVA, p=0.2570) (Fig 3C). Similarly, CM injection into the vibrissal pad induced mechanical sensitivity in the face compared to baseline (BL) responses (EC50 0.16mN Veh/CM vs 0.47mN BL) that was also inhibited by pre-injection of ANA12 in the face (EC50 0.23mN ANA12/CM vs 0.16mN Veh/CM) (Fig 3D).

Fig 3. Effect of ANA12 on OSCC-induced pain behaviors.

Following baseline (BL) feeding measurements, athymic mice were injected with A. NOK cells or B. HSC2 cells. At day 9 post cell inoculation, NOK and HSC2-injected animals were injected with either vehicle or 3ug/20ul injection of ANA12 in the tongue and 1 hour following injection, food intake was measured for an hour. n=6–10 per group. C. HSC2 cells were grown in serum free conditions and conditioned media (CM) was harvested at 24 hours and filtered. Following baseline (BL) feeding measurements, athymic animals were injected with either vehicle or 3ug/20ul injection ANA12 in the tongue and 1 hour later, 15ul of CM was injected in the tongue of both groups. Feeding behavior was tested 15 mins after CM injection (Post CM Injection). n=6–10 per group D. HSC2 cells were grown in serum free conditions and conditioned media (CM) was harvested at 24 hours and filtered. Baseline (BL) von Frey thresholds were taken for two groups of athymic animals: those that were later assigned to vehicle group [BL(VEH)] and those that were assigned to ANA12 group [BL(ANA12)]. Following BL measurements, animals were injected with either vehicle (VEH) or 3ug/20ul injection of ANA12 in the vibrissal pad and 1 hour later, 15ul of CM was injected in the vibrissal pad in both groups. von Frey thresholds were measured 15 mins after CM injection (VEH, CM and ANA12, CM). n=5 per group. Data are represented as mean±SEM. Data analyzed with one-way ANOVA with Bonferroni post hoc test at p<0.05.

Further, we have previously shown that HSC2 tumor induces mechanical hypersensitivity in the face suggesting central hypersensitivity[6]. Therefore, we tested whether local TrkB signaling in the tongue was involved in mediating OSCC-induced facial hypersensitivity. ANA12 (3ug/20ul injection) was injected in the tongue at day 6, 7 and 8 post cell inoculation to prevent persistent peripheral barrage induced by the tumor. Mechanical hypersensitivity was tested at day 9 post cell inoculation. As seen in Fig 4A, local injection of ANA12 in NOK injected animals did not affect mechanical hypersensitivity in the face (EC50 0.56mN Veh vs 0.60mN ANA12). However, ANA12 significantly blocked tumor-induced facial mechanical sensitivity (Fig 4B) (EC50 0.37mN Veh vs 0.60mN ANA12) suggesting that peripheral BDNF/TrkB axis in the tongue may contribute to the OSCC-induced central hypersensitivity, in addition to regulating the peripheral pain transmission in the tongue upon tumor growth. Repeated injections of ANA12 had no effect on tumor volumes or body weights as shown in Fig S3A and B.

Fig 4. Effect of Lingual TrkB inhibition on tumor-induced mechanical sensitivity in the face.

Athymic animals were injected with A. NOK cells and B. HSC2 cells in the tongue and injected with vehicle or 3ug/20ul injection ANA12 in the tongue on day 6,7 and 8 post cell inoculation and von Frey responses were measured at day 9 post cell-inoculation. n=6–10 per group. Data are represented as mean±SEM. Data analyzed with one-way ANOVA with Bonferroni post hoc test at p<0.05.

To test whether the effect of ANA12 is sex-dependent, we performed similar experiments in female mice. Interestingly, Injection of ANA12 in the tongue did not reverse tumor-induced reduction in feeding behaviors (Fig S4A) (0.3g ± 0.067 VEH vs 0.267g ± 0.078 ANA12, 2-way ANOVA , p=0.909). Moreover, ANA12 worsened mechanical hypersensitivity in the face (Fig S4B) (produced by the oral tumor (EC50 0.22mN Veh vs 0.095mN ANA12) suggesting that peripheral activities of TrkB in oral cancer pain may be sex-dependent. Tumor size and body weights were unaltered with repeated injections of ANA12 in female mice (Fig S4C and D). Since we observed a varied sex-difference in the behavioral experiments, we continued the rest of the study in male mice that produced consistent reversal of tumor-induced nociception with inhibition of BDNF signaling.

ANA12 blocked tumor-induced mechanical hypersensitivity of A-slow HTMR fibers innervating the tongue.

Using single fiber tongue-nerve electrophysiology, we have previously shown that OSCC tongue tumor produces mechanical hypersensitivity of C-nociceptors as well as of A-slow high threshold mechanoreceptors (HTMR)[19]. To determine if TrkB activities contribute to tumor-induced mechanical hypersensitivity of these lingual fiber types, we acutely applied vehicle or 0.06ug/ul of ANA12 on the tongues of tumor-bearing preparations prior to recording von Frey thresholds as well as impulses to increasing mechanical force. As shown in Fig 5A, S5A and B, ANA12 did not affect C-fiber mechanical responses (2-way ANOVA, p=0.7161 for Interaction), conduction velocity (CV) (0.655m/s ± 0.058 ANA12 vs 0.59m/s ± 0.048 Veh, Unpaired T-test, p=0.4328) or von Frey thresholds (0.73g ± 0.168 ANA12 vs 0.68g ± 0.12 Veh, Unpaired T-test, p=0.0827), compared to vehicle treated group, post tumor growth respectively. Interestingly, local application of the drug in the tongue, significantly reversed tumor-induced mechanical hypersensitivity at forces 25–200mN (Fig 5D, S5C and S5D) of A-slow HTMR fibers (2way ANOVA, p=0.0060 for Interaction), while leaving CV (1.71 m/s ANA12 vs 1.53m/s Veh, Unpaired T-test, p=0.3477) and von Frey thresholds (1.05g ± 0.3942 ANA12 vs 1.047 ± 0.3942 Veh, Unpaired t-test, p=0.9928) unaffected (Fig 5E and F).

Fig 5. Effect of ANA12 on Tumor-Induced Mechanical Sensitivity of C and A-slow Lingual Nerve Fibers.

Athymic male animals were injected with HSC2 cells and at day 9 post-cell inoculation, tongue-lingual nerve was dissected following which single lingual nerve fibers were isolated and identified by conduction velocity (CV). Tongue tissue was applied with 1.5–2ml solution of vehicle (Veh) or 0.06ug/ul of ANA12 for 5 min following which fibers were tested for von Frey thresholds and nerve discharges at mechanical forces from 10–200mN applied for 10 seconds each at 60 sec intervals. A, B and C show mechanical impulses, CV and von Frey thresholds of C-HTMR fibers respectively and D, E and F show mechanical impulses, CV and von Frey thresholds of A-slow HTMR fibers respectively. n=10–15 fibers per group. Data are represented as mean±SEM. Data analyzed by unpaired Student’s T- test for CV and von Frey threshold comparisons and two-way ANOVA with Bonferroni post hoc test for mechanical responses. Significance marked at p<0.05.

TrkB is expressed in multiple subtypes of lingual sensory neurons.

We then characterized TrkB expression in sensory neurons innervating mouse tongue. Lingual sensory neurons were retrogradely labeled with WGA injection in the tongue and tested for TrkB expression in TG sections. As shown in Fig 6A, a number of neurons co-localized with WGA and TrkB indicating that lingual sensory neurons expressed the TrkB receptor. Further characterization of the receptor expression was performed with single cell RTPCR of WGA positive neurons for full-length TrkB (long TrkB), truncated TrkB (short TrkB) (i.e. TrkB without the intracellular tyrosine kinase domain), CGRP and NFH (Fig 6B). Neurons were divided into small, medium and large cells. Our data revealed that a majority of full-length TrkB and truncated TrkB were expressed in large and medium sized neurons (approx. 60% of all large, 60% of all medium and 20% of all small neurons) of which, the truncated isoform was expressed in a greater proportion of these neurons compared to the full-length isoform (47.7% truncated vs 13.6% full-length of all large neurons and 43.47% truncated vs 15.21 of all medium neurons) (Fig 6C). Interestingly, a small proportion of small neurons also expressed both isoforms of TrkB receptor (13.3% truncated vs 6.6% full-length of all small neurons) (Fig 6C). A majority of the full-length receptor was expressed in myelinating neurons (as marked by NFH positivity) (87.5% of all full-length expressing neurons) whereas 55% non-myelinating neurons expressed the truncated form and 45% expressed the full-length form of the receptor. (Fig 6D). Further classification based on CGRP expression revealed that 87.5% of full-length TrkB is expressed in CGRP negative myelinating neurons with a small proportion (12.5%) expressing the isoform in CGRP negative non-myelinating neurons (Fig 6E). On the contrary, the truncated form of the TrkB receptor was expressed equally at small proportion of 14.89% of all CGRP negative non-myelinating neurons as well as CGRP positive myelinating neurons with a majority of it expressed in CGRP positive non-myelinating neurons (40.42%) followed by in CGRP negative myelinating neurons (29.78%) (Fig 6E).

Fig 6. TrkB expression in mouse lingual sensory neurons.

Naïve male animals were injected bilaterally with 1% WGA-488 in the tongue and 2 days later, A. animals were perfused, and trigeminal ganglia were harvested for immunohistochemistry. TG sections were stained with specific antibody for TrkB and images visualized for co-localization of WGA and TrkB positivity. n=2 animals used. Images taken at 20x magnification. Arrows indicate co-localization of WGA and TrkB. B. WGA positive small, medium and large cells were manually picked and subjected to single-cell RTPCR for expression of full-length TrkB, truncated TrkB, CGRP and NFH. Data represented as heatmap of relative gene expression normalized to the internal control. n=44–46 cells per size. C. Percentage of small medium and large neurons expressing full-length TrkB (TrkB long) and truncated TrkB (TrkB Short). D. Percentage of neurons expressing long and short TrkB that are NFH positive and NFH negative. E. Percentage of neurons expressing CGRP, NFH and long and short TrkB.

Neuronal TrkB contributes to BDNF mediated OSCC-induced nociceptive behaviors.

To test whether neuronal TrkB plays a role in OSCC-induced pain, we employed molecular inhibition of TrkB using intraganglionic siRNA for TrkB. Accell siRNA from Horizon Discovery has been previously used for in vivo knockdown of genes [46] and has been shown to specifically transfect into neurons over other cell types within the brain[38]. Using intraTG injections of FAM-labeled Accell non-targeting siRNA, we were able to visualize significant green fluorescence transfection within sensory neurons of the V3 division of trigeminal ganglion (Fig. S6) at 5 days post-injection. Therefore, using this approach, we used Accell siRNA against TrkB to knockdown the receptor within TG tissue. Injection of 5ug/5ul injection of TrkB siRNA produced up to 70% knockdown of long TrkB and 35% of short TrkB in V3 division of TG tissue (Fig S7A, B and C). Testing for feeding behavior and von Frey sensitivity in the face, our data showed that TrkB knockdown had no effect on any of the behavior assays in NOK-injected animals (Fig 7A and 7B), whereas TrkB knockdown prevented HSC2 tumor induced reduction in food intake (0.318g ± 0012 Scr siRNA vs 0.573g ±0.049 TrkB siRNA, 2-way ANOVA, p= 0.0007) (Fig 7C) as well as tumor-induced central hypersensitivity in the face (Fig 7D) (EC50 0.078mN Scrambled siRNA vs 0.196mN TrkB siRNA). These data demonstrate that OSCC-released BDNF may directly act on TrkB expressing lingual sensory neurons, contributing to oral cancer pain.

Fig 7. Effect of TrkB knockdown in trigeminal ganglia on oral tumor-induced pain behaviors.

Following baseline measurements, athymic male animals were injected with A and C. NOK cells or B and D. HSC2 cells and at day 4 post cell inoculation, animals will either subjected to bilateral intraganglionic injections of 5ug scrambled siRNA or TrkB siRNA. At day 9 post-cell inoculation, A and B, animals were tested for feeding behavior and C and D animals were tested for von Frey responses. n=6–10 per group. Data are represented as mean±SEM. Data analyzed with one-way ANOVA with Bonferroni post hoc test at p<0.05

DISCUSSION

BDNF/TrkB signaling has been well-known for its significant contribution in the central pathway of nociception, in various pain conditions [1–3; 39; 48]. Only recently, it was reported that BDNF and TrkB, expressed in the synovium of osteoarthritis patients contributes to OA-related pain [18]. However, whether this pathway is involved in mediating peripheral transmission of orofacial pain at the site of injury/disease is entirely unknown.

We previously reported that BDNF/TrkB axis contributes to oral cancer pain[6]. Expanding on this study, the current study explored the role of peripheral BDNF/TrkB signaling in oral cancer pain at the site of tumor growth.

BDNF/TrkB has been shown to be expressed in human oral tumors and is implicated in tumor progression [10; 24; 26; 60]. Accordingly, we showed that BDNF is expressed in three different human OSCC cell lines and in tongue tumors in vivo; is released in the extracellular medium and its levels increased with tumor progression. We therefore asked whether the released BDNF contributes to tongue tumor- induced pain behaviors. Neutralization of BDNF within the tongue by the TrkB-Chimera fusion antibody reversed tumor-induced reduction in feeding without affecting tumor size, body weights or feeding of NOK-injected animals, suggesting of BDNF contribution to tumor-induced nociception at the site of its growth. We previously showed that BDNF is increased in sensory neurons post tongue tumor growth [6] and systemic administration of ANA12 reversed tumor-induced feeding behavior and facial mechanical hypersensitivity. Whether or not neuronal BDNF plays an additive role to BDNF signaling within tongue tumors is yet to be explored. However, recently, it was reported that BDNF expressed in sensory neurons may not contribute to allodynia in several pain models. While this study did not use orofacial pain models, it is likely that our previous study may have shown reversal of tumor-induced nociception either due to the peripheral BDNF signaling in the tongue or a central pathway involving medullary dorsal horn[11].

Notably, reversal of feeding behavior with local TrkB Chimera Ab injection in the tongue of tumor-bearing animals do not indicate the source of BDNF in tumor-tongue tissue as, in addition to OSCC cells, BDNF and proBDNF, can also be released from immune cells [22; 23], endothelial cells [37], muscle cells [33], Schwann cells [58], or perhaps even from neuronal axons via axonal transport of the peptide from the cell bodies (although this has only been reported at the central terminals till date)[17]. Therefore, to determine whether OSCC-derived BDNF contributes to painful behaviors, we injected OSCC conditioned media (CM) in naïve mice and observed significant reduction in feeding when injected into the tongue as well as increased mechanical hypersensitivity when injected into the vibrissal pad. To limit the media around the size of BDNF, CM was double- filtered prior to injection in vivo. Moreover, pre-treatment with ANA12 significantly inhibited CM-induced feeding behavior and facial mechanical sensitivity, further demonstrating that BDNF within OSCC CM contributed to pain-like behaviors. Collectively, our data indicated that BDNF derived from OSCC cells within the tongue tumor at least in part, contributes to pain-like behaviors at the site of tumor growth.

To further test the role of TrkB receptor in the tongue in mediating oral cancer pain, we injected ANA12 in tumor-bearing tongue and observed significant reversal of tumor-induced feeding behaviors. These data confirmed that peripheral BDNF/TrkB signaling significantly contributes to oral cancer pain. Interestingly, we also observed that persistent blockade of lingual TrkB at the time of tumor development prevented tumor-induced central hypersensitivity as measured by von Frey thresholds in the face, suggesting that peripheral TrkB signaling within the tongue mediates tumor-induced central hypersensitivity, probably by repeated peripheral barrage produced by the tumor. The role of TrkB in regulating nociception at the level of the spinal cord and/or the brain has been reported previously [2; 3]. However, our study demonstrated that TrkB activities at the site of the disease can also participate in secondary sensitization at uninjured sites in the orofacial region. The mechanism by which this phenomenon occurs remains to be elucidated. An alternative interpretation to our finding is that tumor-induced referred pain may be due to reported cross-talk occurring within neuronal soma within the ganglia [35], although it is generally believed that this cross-talk occurs between nearby cell bodies and testing facial sensitivity assesses neurons innervated by the V2 division of the ganglion whereas the tongue-tumor is innervated by the neurons of the V3 division. Therefore, whether face-innervating V2 neurons are in close enough proximity to tongue innervating V3 soma, for this cross-talk to occur, is yet to be determined.

Similar to TrkB Chimera Ab, ANA12 had no effects on tumor size or body weights of the animals. These data were in contrast to prior reports demonstrating that BDNF/TrkB is overexpressed in oral tumors and can be potentially targeted for tumor regression[24; 26; 60]. However, these reports mostly imply the autocrine function of the pathway within OSCC cells. In our study, we did not observe any detectable levels of TrkB in HSC2 cells (Fig S8A), possibly accounting for no difference in tumor volumes by ANA12 or TrkB Chimera Ab treatment.

We also assessed the role of TrkB signaling in oral cancer pain in female mice. Interestingly, injection of ANA12 in the tongue did not reverse tumor-induced reduction in feeding behavior suggesting that the role of this signaling pathway is sex -dependent. These data are in accordance with prior studies reporting that inhibition of BDNF in microglia only reverses neuropathic pain in males, but not females [36; 45]. It is reported that pain intensity is higher in female oral cancer patients than in males [42]. Therefore, testing the effect of a higher dose of ANA12 in females in required to conclusively interpret the contribution of TrkB signaling in females. Even more intriguing were our results obtained with assessing the effect of local ANA12 injection on tumor-induced facial mechanical hypersensitivity. ANA12 not only had no effect on tumor-induced central hypersensitivity, but worsened mechanical sensitivity produced by the tumor. This finding was very surprising and is the first study reporting that local TrkB inhibition can increase nociceptive behaviors in female mice. Taken together, a thorough investigation of the sex-dependency of BDNF/TrkB signaling is critical to truly understand the mechanism of action of this pathway in oral cancer pain and will be pursued in our future studies.

We then investigated the sensory fiber types regulated by tumor-induced TrkB activities in male mice. Using single-fiber tongue-nerve electrophysiology, we assessed whether TrkB contributed to tumor-induced hypersensitivity of C-nociceptors and A-slow HTMR fibers. Acute application of ANA12 to tumor-bearing tongue preparations reversed tumor-induced mechanical hypersensitivity of A-slow HTMR fibers but not of C-fibers. Interestingly, while ANA12 reduced mechanical impulses of A-slow HTMR fibers, Von Frey thresholds remained unaffected. This indicates that acute inhibition of TrkB activities may at least reduce intensity or duration of pain produced by A-slow HTMR fibers. Also, despite not observing an effect of ANA12 on C-fibers, the contribution of TrkB activities in regulating peripheral C-afferents cannot be entirely ruled out yet, as ANA12 was only applied for 5 mins and perhaps a longer treatment period is required to observe a detectable difference in C-fiber responses or even a change in von Frey thresholds for both fiber types. TrkB is expressed in a variety of cell types during injury including endothelial cells [12], muscle cells [4] and keratinocytes[31], all of which can potentially release mediators to activate or sensitize C-nociceptors. In fact, our data showed that TrkB is expressed at detectable levels even in naïve mouse tongue tissue (Fig S8B). Therefore, additional studies are needed to elucidate the role of the TrkB receptor in regulating C-nociceptors and A-slow HTMR’s during oral cancer.

The roles of the two known isoforms of TrkB in oral cancer pain is entirely unknown. In characterizing the expression of full-length TrkB and truncated TrkB in lingual sensory neuronal subtypes, our data revealed the following: 1) Truncated TrkB was expressed in higher proportion of neurons compared to full-length isoform. Gowler et al also demonstrated higher truncated TrkB expression compared to full-length TrkB in the synovium [18], indicating that the short form may have a predominant role than widely studied full-length isoform; 2) Full length isoform is likely expressed in A-fibers as determined by cell size, and NFH expression. Conversely, truncated TrkB is expressed in C- and a separate subpopulation of A-fibers than those expressing the long form. Collectively, these data imply that full-length TrkB expression may correlate with previously observed TrkB expression in subpopulation of Aδ DRG neurons [28; 51]. However, no prior study has reported the expression of truncated TrkB in neuronal subtypes. Our data suggests that truncated TrkB may have a role in regulating C- and Aδ fibers of TG neurons including peptidergic and non-peptidergic subtypes; 3) Traditionally, short TrkB was considered to be a dominant negative for long TrkB activities [2; 14; 20]. However, our data showed co-expression of the long and short isoforms in very few cells, further indicating that both isoforms may have independent roles from each other. While there is considerable literature on the regulation of full-length TrkB, its role in regulating other ion channels and its downstream signaling pathways[1; 3; 39; 48] [44; 53], such information for truncated TrkB isoform, especially pertaining to peripheral nociception remains to be a significant gap in knowledge.

Due to significant TrkB expression in lingual sensory neurons, we tested whether intraganglionic knockdown of TrkB reversed oral tumors-induced pain behaviors. TrkB knockdown using siRNA in TG tissue significantly reversed tumor-induced reduction in food intake as well as tumor-induced facial mechanical hypersensitivity, indicating that perhaps a direct mechanism of OSCC-derived BDNF on lingual neuronal TrkB activities may partly mediate the peripheral transmission of tumor-induced nociception within the tongue. Expectedly, we observed almost a complete reversal of pain behaviors with intraTG siRNA injection whereas a partial reversal with ANA12 injection in the tongue, possibly due to the differences in mechanism of TrkB inhibition between the two approaches (gene expression with siRNA versus receptor activity with ANA12), dosing as well as duration of inhibition. Notably, while, we observed a greater reduction of expression of full-length TrkB compared to truncated TrkB with siRNA; likely due to the higher number of neurons expressing the short form, it is unclear how many of the tongue-innervating neurons expressing each of the isoforms were transfected with the siRNA. Therefore, both receptors could be playing a significant role in mediating oral cancer pain and further studies are warranted in delineating whether one isoform plays a dominant role over the other.

Taken together, the current study provides compelling evidence that OSCC-released BDNF may regulate lingual neuronal hypersensitivity via peripheral TrkB activities in oral cancer pain. Several other questions remain to be answered to further understand the role of this pathway in oral cancer pain:

1. The mechanism by which BDNF is released from OSCC is not known yet. It is previously reported that NGF and/or ATP signaling can lead to increased BDNF release [7; 27; 29] [25; 30]. Both, NGF and ATP levels have been shown to be upregulated in oral cancer biopsies as well as implicated in oral tumor-induced pain [57; 59]. Thus, involvement of these pathways in BDNF release and signaling in oral cancer can be investigated.

2. The role of pro-BDNF and p75R signaling is not explored in our study.

3. While HSC2 cells don’t express detectable levels of TrkB (Fig. S8A), prior reports have demonstrated TrkB expression in human oral tumors as well as other OSCC cell lines [10]. Therefore, using TrkB expressing OSCC cells, an autocrine effect of BDNF/TrkB signaling in regulating peripheral lingual nerve terminals is warranted. Besides, as stated above, the naïve tongue itself expresses considerable amounts of TrkB. Thus, it is possible that BDNF could also modulate sensory neuronal activities via in indirect effect through other cell types, in addition to regulating neuronal TrkB activities.

4. A correlation of pain intensity with BDNF/TrkB expression in human samples will significantly aid in increasing the translational relevance of this signaling pathway in oral cancer pain.

5. Small-molecule inhibitors for TRK receptors have been widely explored for the treatment of a variety of cancers and many are being currently tested for clinical trials [13], indicating that use of such targeted therapy against the TRK receptors can have significant potential for anti-cancer as well as pain treatments allowing for considerable improvement in patient quality of life.