Background
Inflammation and pain largely share a common course of progression; patients with inflammation may suffer hyperalgesia and/or allodynia to various mechanical, thermal and chemical stimuli [
1]. Inflammation results in an array of chemical mediators being released and triggering immune cell accumulation in the damaged area. Those activated immune cells further release pro-inflammatory cytokines and neurotrophins including nitric oxide (NO), interleukin-1β (IL-1β), interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α) and nerve growth factor (NGF) [
1‐
3] producing either central or peripheral sensitization [
3‐
5].
TNF-α is a potent pro-inflammatory cytokine that has been used frequently in laboratory studies to evoke inflammatory reactions. TNF-α activates the release of many cytokines, such as IL-1β, IL-6 and IL-8, and participates in the development of inflammatory hyperalgesia mainly through its receptor, TNFR1 and TNFR2 [
6‐
8]. TNF-α-dependent neuropathy or inflammatory pain appears to be largely mediated by TNFR1 [
9‐
11].
Neurotrophins like NGF, neurotrophin 3/4 (NT-3/4) and brain-derived neurotrophic factor (BDNF) can be released from DRG, acting to either support neuronal development [
12] or participate in the induction of hyperalgesia [
3]. NGF is recognized to play a potent role in the development of neurogenic pain by inducing hyperalgesia [
5,
13]. After release from immune cells, NGF up-regulates the expression of proteins involved in inflammatory pain transmission, TRPV1, BDNF, calcitonin gene-related peptide (CGRP) and substance P in the DRG via tyrosine protein kinase A (trkA) receptor [
2,
3,
14‐
17]. BDNF is expressed and synthesized in small- to medium-sized DRG neurons and co-expressed with trkA along with CGRP and substance P [
18,
19]. Hence, BDNF can be released in response to peripheral NGF via trkA stimulation and is known as a nociceptive modulator for both pain perception and sensitization at both spinal and supraspinal levels [
18]. In particular, nociceptor-drived BDNF has been demonstrated to regulate acute and inflammatory pain [
20]. Tyrosine protein kinase B (trkB) is a high affinity BDNF receptor [
18]. Recent ultrastructural evidence indicates that trkB receptor is not only expressed in post-synaptic neurons but also localizes to pre-synaptic terminals in spinal lamina II [
21]. BDNF in spinal cord lamina II is a neuromodulator of nociception at synapses. Recent studies indicate that BDNF acts on pre-synaptic trkB receptors to increase the frequency of glutamatergic EPSCs in spinal dorsal horn of complete Freund's adjuvant- (CFA-) treated rats [
22,
23]. Hence, BDNF-induced thermal hyperalgesia can be inhibited by intrathecally applied NMDA receptor antagonist D-APV [
24]. Further, it has been found that trkB phosphorylation is increased under noxious mechanical, thermal or chemical stimulation [
25]. This effect is tightly associated with extracellular signal-regulated kinase (ERK) and phospholipase C (PLC)/phosphoinositide-3 kinase (PKC) signals [
18].
The aim of this study was to investigate the role of BDNF-trkB within DRG during the progression of inflammatory pain, specifically focusing on whether and how the pro-inflammatory cytokine TNF-α influences trkB expression and signaling. To accomplish this aim, we used CFA-induced inflammation in rats as a chronic pain model to examine neurochemical changes in lumbar DRG. Levels of the pro-inflammatory cytokine TNF-α were first determined in the inflamed tissues. Consequently, a set of animals was injected with TNF-α directly into hindpaw to assess inflammatory pain responses. In addition, we used DRG primary cultures and treated with TNF-α for mechanistic and functional studies. We demonstrate that BDNF and trkB receptors are up-regulated in DRG of CFA- and TNF-α-induced inflammatory pain models. In addition, retrograde transport of TNF-α was observed in the CFA-induced inflammation model. Both trkB receptors and trkB-induced ERK signaling were increased after chronic TNF-α-treatment in DRG primary cultures. Furthermore, both CGRP and substance P release were enhanced after chronic TNF-α pre-treatment or acute BDNF administration in DRG cultures. We conclude that experimental arthritis and TNF-α initiate a BDNF-trkB feed-forward pathway in peripheral sensory neurons, and this up-regulation of BDNF-trkB system may participate in pain sensitization.
Methods
Animals
Male Sprague-Dawley (SD) rats (BioLASCo Taiwan Co., Ltd) weighing 250-280 g (for inflammatory animal model) or 3-4 weeks of age (for primary culture) were used in the current study. After arrival, animals were acclimatized to the room with controlled temperature (22 ± 2°C), air humidity (50 ± 10%) and 12 hours day-night cycle (light on at 7:00 AM) for at least 7 days before experimentation. The animals were housed 2 or 3 per cage with food (Western Lab 7001, Orange, CA, USA) and water ad libitum. The animal handling and drug treatments were performed in strict accordance with the NIH Guide for the Care and Use of Laboratory Animals, and all procedures were approved by the Animal Care Committee of Chang-Gung University.
Chemicals
CFA (Sigma, St. Louis, MO, USA) was freshly prepared (1:1 with saline containing 0.05 mg heat-killed and dried Mycobacterium tuberculosis and paraffin oil mixed in a water-in-oil emulsion). Recombinant rat TNF-α protein (R & D systems, Minneapolis, MN, USA) was reconstituted with phosphate buffered saline (PBS)/0.1% bovine serum albumin (BSA). Recombinant human BDNF protein (Abcam, Cambridge, MA, USA) was reconstituted with ddH2O. NGF 2.5S (Chemican, Temecula, CA, USA) was reconstituted with PBS.
Animal model of inflammatory pain
To produce an in vivo model of inflammatory pain, animals were anesthetized with isoflurane and locally injected with CFA, TNF-α or vehicle. In the CFA-treated group, animals were injected with 100 μl CFA into the right hindpaw with the contralateral side serving as an untreated control. Other animals received saline injection in the hindpaw to serve as a vehicle control. In the TNF-α-treated group, animals received 500 ng TNF-α (30 μl) in the hindpaw twice daily for three consecutive days, followed by daily mechanical von Frey tests. Lumbar L3~L5 DRGs or hindpaw tissues were collected 72 hours after CFA or TNF-α injection and samples of DRG were divided into following different groups: CFA (ipsilateral), TNF-α (ipsilateral), control (contralateral) and saline groups.
Pain threshold test
Mechanical hyperalgesia was measured by the
von Frey method [
26]. The test was performed daily in CFA- or TNF-α-treated rats for a period of up to three days. To perform the von Frey test, rats were placed in a clear plexiglass box (20 × 19 × 20 cm) on an elevated mesh screen. A calibrated von Frey rigid tip (Electronic von Frey Anesthesiometer, IITC Life Science, USA) was applied to the plantar surface of each hindpaw in a series of logarithmically ascending forces. The responses were recorded in grams of paw withdrawal averaged over three to five applications referred to as mechanical pain threshold. The interval of each application was 5 min.
Primary DRG culture
Bilateral lumbar L1-L6 DRGs were dissected from 3-4-week-old SD rats. DRG tissues were digested with 1 mg/ml collagenase (Sigma) at 37°C for 30 min and then incubated with 0.25% (v/v) trypsin-EDTA (Biological Industries, Israel) at 37°C for 30 min. Tissues were centrifuged at 300 × g for 5 min and then re-suspended in DMEM/F12 medium (repeated three times). Next, tissues were manually triturated approximately sixty times using a flame-polished Pasteur pipette. The dissociated cells were suspended in DMEM/F12 medium containing 10% fetal calf serum, 100 μg/ml penicillin/streptomycin and 1 mM sodium pyruvate and plated onto poly-L-lysine-coated plates. Medium was replaced the following day with the addition of 10 μM Ara-C and 100 ng/ml NGF and changed every two days thereafter. After cell plating for two days, all drugs (TNF-α, BDNF or NGF) were added for the designated time periods and samples (medium or cultured cells) were all collected by the end of each experiment (including basal control).
Western blot
All samples were sonicated in 1% (w/v) sodium dodecyl sulfate (SDS) and heated to 100°C for 5 min. Protein concentration was determined by the Coomassie blue method with bovine serum albumin as standard. Equal amounts of proteins (10-40 μg) were separated by 10% SDS-polyacrylamide gel electrophoresis and transferred onto a polyvinylidene difluoride (PVDF) membrane (Millipore Corporation, Bedford, MA, USA) using 200 mA for 70 min at 4°C. Immunoblots were first blocked with 5% (w/v) non-fat milk for 1 hour, followed by incubation overnight with specific primary antibody. The antibodies used were anti-β-actin antibody (1:1000; Sigma), anti-phospho-p44/42 MAP kinase (Thr202/Tyr204) antibody (anti-phospho-ERK1/2, 1:1000; Cell Signaling, Beverly, MA, USA). On the second day, blots were incubated with secondary antibody conjugated with HRP (1:2000-1:10,000, GE Healthcare, UK) for 1 hour. After another 3 × 10 min washes with TBS-T buffer, the substrate solution (Amersham ECL kit) was applied for 1 min. The enhanced chemiluminescence signal was detected with X-ray film (Kodak, X-OMAT) and quantified by laser scanning densitometry (Molecular Dynamics, Sunnyvale, CA). Signals were normalized to β-actin.
Quantitative real-time PCR
Total RNA was isolated from cultured DRG neurons or tissues using TRIzol® reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. The mRNA was transcribed to cDNA via reverse transcriptase (HT BioTechnology, England UK). The cDNA for corresponding targets was measured by quantitative real-time PCR using a PE Applied Biosystems prism model 7000 sequence detection instrument (Applied Biosystems, Foster City, CA, USA). The PCR program was: 95°C for 10 min then 95°C for 15 sec and 60°C 30 sec for 40 cycles. SYBR (Invitrogen) was used as fluorescent dye and ROX as reference dye. The primer sequences are as follows: 18S-forward, 5' GCT GTG GTC CAA GGC CAT TTT 3'; 18S-reverse, 5' CCG AGT TAC TTT TCC CCA GAT GAC 3'[GenBank: NM_000996]; TRPV1-forward, 5' GCG AGT TCA AAG ACC CAG AG 3'; TRPV1-reverse, 5' GGC ATT GAC AAA CTG CTT CA 3'[GenBank: NM_031982]; trkB.FL-forward, 5' AAG ATC CTG GTG GCC GTG AAG A 3'; trkB.FL-reverse, 5' CGG CTT CGC GAT GAA AGT CCT T 3'[GenBank: NM_012731]; BDNF-forward-1, 5' CAA GGC AAC TTG GCC TAC CC 3'; BDNF-reverse-1, 5' GAG CAT CAC CCG GGA AGT GT 3'; BDNF-forward-2, 5' TTA CCT GGA TGC CGC AAA CA 3'; BDNF-reverse-2, 5' TGG CCT TTT GAT ACC GGG AC 3' [GenBank: NM_012513]; CGRP-forward, 5' AAC CTT AGA AAG CAG CCC AGG CAT G 3'; CGRP-reverse, 5' GTG GGC ACA AAG TTG TCC TTC ACC A 3'[GenBank: M11597]. Threshold cycle, Ct, which correlates inversely with the target mRNA levels, was measured. ABI Prism Primer Express was used to design the specific gene primers. After normalization to 18S values, data were expressed as percentage of corresponding controls.
Immunohistochemistry (IHC) for tissue sections
SD rats were sacrificed by decapitation; L3-L5 DRGs were removed and post-fixed in 4% (w/v) paraformaldehyde solution for 48 hours at 4°C, then cryo-protected with 20% (w/v) sucrose in potassium phosphate buffer saline (KPBS, KH2PO4 3.3 mM, K2HPO4 21.9 mM, NaCl 154 mM) overnight at 4°C. IHC images were collected from 14 μm sections prepared using a freezing microtome (LEICA CM3050S, Bannockburn, IL, USA).
Tissue sections were rinsed twice with KPBS. Endogenous peroxidase was quenched by pre-treating the sections with 0.03% (v/v) H2O2. The sections were then rinsed with KPBS followed by pretreatment with 1% (w/v) sodium tetrahydridoborate for 5 min. Afterward, the sections were incubated in primary antibody in 2% (w/v) normal goat serum/0.3% (v/v) TX-100/KPBS at 4°C for 48 hours. The antibodies used were anti-trkB antibody (1:2000; Santa Cruz, Santa Cruz, CA, USA), anti-BDNF antibody (1:1000; Abcam), anti-CGRP antibody (1:8000; Chemicon). Afterward, sections were incubated with 1:200 biotinylated goat anti-rabbit IgG secondary antibody for 1 hour. Immunoreactivity was visualized by the avidin-biotin complex method (Vector Lab., Burlingame, CA, USA). After incubating with the avidin-biotin complex for 1 hour, sections were developed using 3, 3'-diaminobenzideine (DAB; Sigma) and nickel ammonium sulfate in KPBS. Sections were mounted onto gelatin-coated slides, dehydrated through a serial alcohol gradient, degreased with xylene and then covered slipped with Entellan (Merck, Germany). Co-localization of BDNF and TNFR1 was evaluated by immunofluorescent double stain. Primary antibodies (anti-BDNF, 1:200; Abcam and anti-TNFR1, 1:20; Santa Cruz) were mixed and incubated at 4°C for 72 hours. After washed with KPBS, sections were incubated with secondary antibodies (Rodamine-conjugated goat-anti-mouse and Dylight 488-conjugated goat-anti-rabbit, 1:200, Jackson ImmunoResearch, West Grove, PA, USA) for 1 hour at room temperature. Sections were then incubated with 1:1000 DAPI (4',6-diamidino-2-phenylindole, Roche, Switzerland) for 5 min before mounting with glycerol. Signals of immunofluorescent stain were observed by fluorescence microscope.
Immunocytochemistry for cultured cells
Cultured DRG neurons were fixed with 2% (w/v) paraformaldehyde solution (2% paraformaldehyde containing 0.3% TX-100, pH 7.2) for 30 min. Cells were rinsed with PBS/0.3% TX-100 following blocking with 2% normal goat serum in PBS/0.3% TX-100 for 30 min. Cells were then incubated in primary antibody in PBS/0.3% TX-100/2% goat serum at 4°C for 48 hours. The antibodies used were anti-trkB antibody (1:2000), anti-BDNF antibody (1:1000), anti-CGRP antibody (1:8000). TrkB staining was visualized via immunofluorescence. After rinsing three times, cells were incubated with FITC-conjugated goat anti-rabbit IgG secondary antibody (1:200, Jackson ImmunoResearch) for 1 hour. Afterward, cells were then incubated with 1:1000 DAPI for 5 min before mounting with glycerol. Co-localization of TNFR1 and phospho-ERK were evaluated by immunofluorescent double stain. Primary antibodies (anti-TNFR1, 1:20 and anti-phospho-ERK1/2, 1:250) were mixed and incubated at 4°C for 72 hours. After wash with PBS, secondary antibodies, Rodamine-conjugated goat anti-mouse and Dylight 488-conjugated goat anti-rabbit (1:200, Jackson ImmunoResearch), were also mixed together and added for 1 hour. Slides were then incubated with 1:1000 DAPI for 5 min before mounting with glycerol. Immunofluorescent stain signal was observed by fluorescence microscope. On the other hand, BDNF and CGRP staining were visualized via avidin-biotin complex method. Similar to immunohistochemistry, endogenous peroxidase was quenched by pre-treating the sections with 0.03% H2O2 followed by pretreatment with 1% sodium tetrahydridoborate. After incubating with primary antibody and rinsing with PBS/0.3% TX-100, the cells were incubated in 1:200 biotinylated goat anti-rabbit IgG secondary antibody for 1 hour and rinsed with PBS. After incubation with avidin-biotin complex for 1 hour, ABC reaction was performed with a mixture of DAB and nickel ammonium sulfate. Cells were mounted on slides, dehydrated, and cover-slipped with mounting solution. The color of precipitate formed by DAB was monitored microscopically.
Enzyme immunoassay (EIA)
The release of BDNF, CGRP and substance P were determined using BDNF Emax
® ImmunoAssay system (Promega, Madison, WI, USA), CGRP EIA kit (SPIbio, France) or substance P EIA kit (Cayman, Ann Arbor, Michigan, USA), respectively. For BDNF assay, after 5 nM TNF-α treatment for 48 hours, medium of DRG cultures were collected. The cells were lysised, sonicated and then centrifuged at 10,000 × g for 15 min. The supernatants were analyzed and represented as total BDNF content. For CGRP and substance P EIA assay, after 5 nM TNF-α treatment for 48 hours, culture medium were replaced by a fresh medium and allowed cells stable for 2 hours. Afterwards, 400 ng/ml BDNF was added for 30 or 60 min and the supernatant was collected and analyzed immediately according to the manufacturer's protocols. The microplate was read at 405 nm (yellow color) in an enzyme-linked immunosorbent assay (ELISA) plate reader (ASAY microplate reader, expert 96, Austria).
The amount of TNF-α in rat hindpaw or DRG was evaluated using a rat TNF-α EIA kit (R&D system). After CFA-injection for 72 hours, rat hindpaw or L3~L5 DRGs were isolated and analyzed immediately. Tissues were collected in PBS solution containing proteinase inhibitors leupeptin, aprotenin and PMSF (Sigma). After homogenization, tissues were centrifuged at 10,000 × g, 4°C for 20 min and supernatants were collected. Blood samples were withdrawn from the animals' trunks into a blood collection tube containing 5.4 mg K2 EDTA and centrifuged at 1,000 × g for 10 min in the room temperature to collect the serum. Samples were analyzed immediately according to the manufacturer's protocols. The concentration of TNF-α was determined by the absorbance at 450 nm with reference filter at 540 nm.
TNF-α retrograde transport tracing
Recombinant rat TNF-α protein (10 μg) was labeled with Alexa Fluor 488 (Molecular Probes, Invitrogen) according to the manufacturer's protocols. Animal received Alexa Fluor 488-TNF-α (25 μl) in CFA pre-treated hindpaw 24 hours before sacrifice (72 hours after CFA injection). Alexa Fluor 488 (25 μl) injection in CFA pre-treated hindpaw served as control. After injection, ipsilateral saphenous nerve was isolated immediately and monitored under fluorescence microscope. The fluorescent signals in the saphenous nerve were separately measured in three segments: distal (0.5 cm from the heel), middle (1.3 cm from the heel) and proximal (2 cm from the heel).
TUNEL reaction
Cell apoptosis was evaluated using an in situ cell death detection kit (TUNEL reaction, Roche, Germany), following the manufacturer's protocols, DRG primary cultures were fixed with 2% paraformaldehyde for 1 hour, then incubated with 0.1% Triton X-100/0.1% sodium citrate for 2 min on ice. Afterwards, the cultures were washed with PBS and then treated with 50 μl TUNEL reaction mixture (containing green Fluorescein) at 37°C for 60 min in the dark. After enzyme reaction, cultures were washed with PBS and double stained with DAPI (1:1,000) for 5 min. Numbers of apoptotic cells were evaluated by observation under a fluorescence microscope.
Data analyses and statistics
ImageJ (NIH, Bethesda, MD, USA) software was used to quantify the immunostaining images. A threshold was set by ImageJ to define the positive neurons among different experiments of the same study. For DRG immunostaining, three different sections (spacing 250~300 μm) were selected to calculate the total number of positive neurons and expressed as the percentage of positive neurons relative to the total number of neurons visualized in DRG sections. For the data in Table
1, small, medium and large DRG neurons were identified and counted based on the diameter of the neuron: small neurons < 25 μm; medium neurons 25 to 40 μm; and large neurons > 40 μm. The results were calculated by two methods: 1) percentage of trkB-positive neurons in specific categories (small, medium, or large) relative to the total number of neurons in DRG sections; 2) percentage of trkB-positive neurons in specific categories relative to the total number of trkB-positive neurons in the DRG sections.
All data were analyzed with GraphPad InStat (GraphPad Software, San Diego, CA, USA). Results are expressed as mean ± SEM. The data were analyzed by one-way ANOVA followed by Tukey multiple comparisons test or un-paired Student's t-test. A value of p < 0.05 was considered significant.
Discussion
Numerous studies have attempted to understand the cellular mechanism underlying hyperalgesia and allodynia using a CFA-induced inflammation pain model. Our study focused on the neurochemical changes in lumbar DRG and clearly shows that protein levels of trkB, BDNF and CGRP are significantly increased after animals develop hyperalgesia. Up-regulation of BDNF and CGRP in DRG during inflammation is a well-known phenomenon along with the consequent effect of post-synaptic trkB over-expression in the dorsal horn of spinal cord [
18,
29]. On the other hand, up-regulation of trkB in DRG during progression of inflammation has rarely been documented. Lee
et al. (1999) reported that mRNA of truncated trkB receptor increases in DRG after CFA injection, but the full-length trkB receptor does not. Further, Chien
et al. (2007) reported that BDNF, CGRP, trkA and p75
NTR are increased in DRG after CFA treatment [
30], while levels of trkB receptor do not change. In addition to these reports, trkB expression and phospho-trkB signal have been found to be enhanced in rat DRG after the insults of colitis or cystitis as well as during spinal cord injury [
31‐
33]. These inconsistent findings imply that expression of DRG trkB receptor during inflammation is variable, possibly due to the complexity of chemical mediators that are recruited in designated disease progression.
In the present study, we found that BDNF protein but not mRNA was significantly enhanced in DRG during inflammation. Previously, it has been reported that levels of BDNF mRNA increase significantly in DRG 24 hours after CFA-injection [
29,
34], while CFA-induced BDNF protein is up-regulated in DRG (L5) for up to 4 days [
35]. Since we observed an increase in the amount of BDNF protein in DRG 72 hours after CFA injection, it is possible that transcriptional regulation occurs only during an early phase while translational regulation lasts longer. In further support of this concept, TNF-α was previously found to enhance BDNF protein expression at 24 hours, but to increase mRNA expression at 4 hours [
36]. Consistent with this notion, we did detect an up-regulation of BDNF mRNA 6 hours after TNF-α treatment in DRG cultures. The differential regulation of transcription and translation might also explain the results obtained in CGRP mRNA and protein, both
in vivo and
in vitro.
Our results reveal a statistically significant increase in protein levels of trkB, BDNF and CGRP as well as mRNA of trkB and TRPV1, mainly in L3, but CGRP protein in L4 and trkB and CGRP proteins in L5 DRG after CFA treatment. On the other hand, TNF-α treatment led to increased protein levels of trkB, BDNF and CGRP predominantly in L4, with relatively minor effects in L3 and L5 DRGs. This segregated effect was unexpected because rat hindpaw are innervated by primary afferents from L3, L4 and L5 DRGs. A previous report regarding the nociceptive dermatomes in Wistar rat hindpaw demonstrated that the L3 dermatome is contained in the inner part of the paw; the L4 dermatome is contained in the middle part of the paw and the L5 dermatome is contained in the outside part of the paw [
37]. Although swelling was visually observed in the entire hindpaw, a differential response was found among the individual segments of lumbar DRG. In support of the notion that segments of the lumbar DRG may respond differently to injury, it has been previously reported, for an animal model of neuropathic pain via L5 spinal nerve ligation, that expression of BDNF and TRPV1 are up-regulated in uninjured L4 DRG [
38]. Similarly, expression of TNFR1 and TNFR2 are increased in injured L5 and uninjured L4 DRG after L5 spinal nerve ligation [
39,
40]. In addition, our results showing differential effects in L3 and L4 in response to CFA and TNF-α might be due to drug formulation, i.e. chyle form of CFA vs. liquor form of TNF-α. A detailed explanation of the cause of uneven sensitivity between L3, L4 and L5 DRGs after CFA or TNF-α hindpaw treatment requires further investigation.
Pro-inflammatory cytokines play an important role in the process of inflammatory pain and their release results in pain sensitization [
2,
3]. TNF-α is a potent cytokine, and intraplantar injection of TNF-α in rats has been previously shown to produce mechanical hyperalgesia while TNF-neutralizing antibodies reverse TNF-α-induced nociception [
1,
8]. Furthermore, TNF-α increases expression of TRPV1 in DRG via TNFR1 receptor with a requirement for ERK signaling [
11]. Interestingly, anti-TNF-α treatment in patients with rheumatoid arthritis results in decreased plasma BDNF [
41]. TNF-α can also enhance expression of BDNF protein in trigeminal ganglion neuronal cultures [
36]. Hence, it is plausible that a casual relationship between TNF-α and the BDNF-trkB system in DRG mediates inflammation-induced hyperalgesia. In this study, we first demonstrated that levels of TNF-α were raised significantly in the inflamed tissues, as compared to uninjured hindpaw, thereby confirming a role for TNF-α during inflammation. Further, we proved that inflammatory hyperalgesia could be induced by direct TNF-α injection into hindpaw. The finding that TNF-α in DRG is undetectable after animals develop hyperalgesia is surprising, but may be explained by the following considerations. First, mono-arthritis developed under the current experimental conditions, which amounts to an early inflammatory stage confined mainly to the affected tissues. The resulting inability to detect TNF-α in serum is similar to the findings of a previous report [
42], and seems to support this argument. Second, TNF-α produced locally in injured tissues would act on TNF-α receptors found on free nerve endings and undergo retrograde transport to DRG. This phenomenon was recently demonstrated in a rat model of carrageenan-induced inflammation [
27] as well as a rat model of peripheral nerve injury [
28]. In the present study, we validated the ability of TNF-α to undergo retrograde transport along the saphenous nerve in a CFA-induced inflammatory rat model. The axonal transport pattern and speed are comparable with previous observations using a carrageenan-induced pain model [
27]. Hence, rather than massive TNF-α accumulation in DRG that would be expected to occur at a late stage of inflammation (progressive poly-arthritis), we speculate that in the current model, intracellular TNF-α carried by the TNF-α receptor is the main participant in the DRG, although we could not exclude the possibility that TNFR1 expression resulted from inflammation while independent from translocation.
Similar to the CFA-induced inflammatory condition, we found that TNF-α injection into hindpaw and application to DRG cultures produced a clear enhancement of trkB and BDNF expression in DRG. In addition, we observed a release of BDNF from DRG cultures after TNF-α treatment. Other than targeting the post-synaptic trkB receptor in the dorsal horn, we speculate that enhanced BDNF would also have chance to activate local trkB receptors in DRG, leading to trkB dimerization and auto-phosphorylation, which would ultimately trigger pre-synaptic signaling pathways [
43]. In support of this hypothesis, our results demonstrate that activation of trkB receptors by BDNF enhances phospho-ERK1/2 signal in TNF-α-pretreated animals, suggesting that a possible intra-DRG BDNF-trkB system could be functionally self-activated under chronic TNF-α exposure. However, we observed an inconsistent result for basal phospho-ERK1/2 signal after TNF-α pretreatment (i.e. an increase in IHC but a decrease in western blot). Other than a difference possibly due to analytical methods, the impact of TNF-α on overall basal MAPK signal requires further investigation. The dual observations that both BDNF and BDNF-triggered phospho-ERK signals largely co-localize with TNFR1 suggest that TNFR1 might associate with
in situ BDNF-trkB up-regulation. In addition to the effect of TNF-α on BDNF-trkB, it is well known that TNF-α can induce IL-6, IL-8 and granulocyte-macrophage colony-stimulating factor (GM-CSF) secretion through activation of MEK1/2, p38 MAPK and NF-κB pathways during inflammation [
44,
45]. Since a notable inflammation was observed in both CFA- and TNF-α-treated hindpaw, we could not exclude the possibility that a combined action of these chemical mediator(s) might work synergistically to up-regulate the BDNF-trkB system
in vivo.
NGF is an important mediator of inflammatory pain [
13]. After release from immune cells, NGF binds with trkA to form an NGF/trkA complex which is then retrogradely transported to DRG to up-regulate expression of genes, such as those for TRPV1, BDNF, CGRP and substance P [
2,
3] via ERK, PI3K, p38 MAPK or CaMKII signaling [
5,
14,
15,
46]. TNF-α has also been shown to enhance NGF expression
in vivo [
42], although NGF does not appear to be involved in TNF-α-mediated BDNF up-regulation in trigeminal ganglia cultures [
36]. Our observation that TNF-α and NGF can both induce BDNF expression in DRG cultures suggests a functional synergism between these two chemical mediators. However, the increase of BDNF by TNF-α may be derived from astrocytes within the cultures [
47]. Although Ara-C was applied in our DRG culture system, we cannot totally exclude the possibility that enhanced TNF-α/NGF-induced BDNF could be partially due to contaminating satellite cells.
Our results show that BDNF treatment increases mRNA expression for trkB receptor in DRG cultures. BDNF is reported to co-express with trkA (20%) and trkB (10%) receptors in DRG [
18,
21]. In addition, BDNF is thought to act as an autocrine or paracrine signal regulating trkB in sensory neurons [
16,
48]. Low doses of BDNF up-regulate trkB receptors while high doses down-regulate trkB receptors [
49]. Our
in vitro results demonstrate that an increase in trkB receptors is triggered by either BDNF or TNF-α. This effect may eventually form a sensitization loop to increase pain transmission at the level of the DRG.
The finding that basal CGRP release is enhanced in DRG after chronic TNF-α treatment supports a previous study which showed that acute TNF-α treatment increases CGRP release in rat trigeminal ganglia [
50]. In addition, these results also correspond with our
in vivo and
in vitro observations that total CGRP protein increases after CFA-induced inflammation [
36,
50]. Similarly, we found that chronic TNF-α treatment results in enhanced basal substance P release in DRG cultures. Acute BDNF stimulation could induce an increase in substance P release 60 min after treatment, but had no effect after TNF-α pretreatment. The findings are consistent with previous reports [
23,
31] that BDNF can stimulate CGRP and substance P release under basal conditions via trkB receptor in spinal cord. TNF-α regulation of substance P release has never been studied in DRG, though Ansal
et al [
51] reported that substance P induces TNF-α mRNA expression and release in mast cells. In addition, substance P enhances mechanical allodynia and heat hyperalgesia in pre-protachykinin A
-/- mice through the actions of NGF and TNF-α [
52]. In this manuscript, we demonstrate that TNF-α pre-treatment enhances the production and basal release of CGRP and substance P, consequently facilitating pain transmission. The lack of a trkB effect on CGRP/substance P release during inflammation suggests that a maximal effect is already achieved by TNF-α pre-treatment.
Competing interests
The authors declare that they have no competing interests in conducting this study.
Authors' contributions
YTL carried out all the in vivo and in vitro experiments in this study and prepared the first draft of this manuscript. JCC conceived of the study, participated in its design and coordination and helped to draft the manuscript. LSR and HLW participated in experimental design and consulted on the study. All authors read and approved the final manuscript.