THIS ARTICLE HAS BEEN RETRACTED: Toll-like receptor 3 contributes to spinal glial activation and tactile allodynia after nerve injury

Authors

Errata

This article is corrected by:

  1. Errata: RETRACTION Volume 121, Issue 4, 693, Article first published online: 16 March 2012

Address correspondence and reprint requests to Koichi Noguchi, MD PhD, Department of Anatomy and Neuroscience, Hyogo College of Medicine, 1-1 Mukogawa-cho, Nishinomiya, Hyogo 663-8501, Japan. E-mail: noguchi@hyo-med.ac.jp

Abstract

Toll-like receptors (TLRs) play an essential role in innate immune responses and in the initiation of adaptive immune responses. Microglia, the resident innate immune cells in the CNS, express TLRs. In this study, we show that TLR3 is crucial for spinal cord glial activation and tactile allodynia after peripheral nerve injury. Intrathecal administration of TLR3 antisense oligodeoxynucleotide suppressed nerve injury-induced tactile allodynia, and decreased the phosphorylation of p38 mitogen-activated protein kinase, but not extracellular signal-regulated protein kinases 1/2, in spinal glial cells. Antisense knockdown of TLR3 also attenuated the activation of spinal microglia, but not astrocytes, caused by nerve injury. Furthermore, down-regulation of TLR3 inhibited nerve injury-induced up-regulation of spinal pro-inflammatory cytokines, such as interleukin-1β, interleukin-6, and tumor necrosis factor-α. Conversely, intrathecal injection of the TLR3 agonist polyinosine–polycytidylic acid induced behavioral, morphological, and biochemical changes similar to those observed after nerve injury. Indeed, TLR3-deficient mice did not develop tactile allodynia after nerve injury or polyinosine–polycytidylic acid injection. Our results indicate that TLR3 has a substantial role in the activation of spinal glial cells and the development of tactile allodynia after nerve injury. Thus, blocking TLR3 in the spinal glial cells might provide a fruitful strategy for treating neuropathic pain.

Abbreviations used
AS

antisense

dsRNA

double-stranded RNA

ERK

extracellular signal-regulated protein kinase

GAPDH

glyceraldehyde 3-phosphate dehydrogenase

GFAP

glial fibrillary acid protein

Iba1

ionized calcium-binding adapter molecule 1

IL

interleukin

JNK

c-Jun N-terminal kinase

MAPK

mitogen-activated protein kinase

MM

mismatch

ODN

oligodeoxynucleotide

poly(I : C)

polyinosine–polycytidylic acid

p-p38

phosphorylated-p38

SNL

spinal nerve ligation

TLR

Toll-like receptor

TRPV1

transient receptor potential ion channel

Toll-like receptors (TLRs) play a key role in host defense during pathogen infection by regulating and linking innate and adaptive immune responses (Akira and Takeda 2004; Akira et al. 2006; Marshak-Rothstein 2006). TLRs belong to a family of receptors that recognize pathogen-associated molecular patterns. TLRs may also have pathogen-independent roles, by initiating responses to host-derived, endogenous ligands. In mammals, 12 members of the TLR family have been identified so far. Among them, TLR3 is known to be a major mediator of the cellular response to viral infection, because it responds to double-stranded RNA (dsRNA), a common byproduct of viral replication (Alexopoulou et al. 2001). Upon activation by dsRNA, TLR3 is tyrosine-phosphorylated in the cytoplasmic domain and TLR3 phosphorylation is essential for its ability to signal (Sarkar et al. 2004, 2007). Microglia, the resident myeloid cells that constitute the innate immune system in the CNS, express TLRs (Olson and Miller 2004; Jack et al. 2005), and the TLR family has been implicated in the microglial response to nerve damage (Block et al. 2007; Guo and Schluesener 2007). Indeed, TLR3 is constitutively expressed in the CNS, and microglia directly recognize dsRNA through the TLR3 pathway (Town et al. 2006). However, the role of TLR3 in neuropathic pain after nerve injury has not yet been defined.

Damage to peripheral nerves often results in spontaneous pain, hyperalgesia (increased responsiveness to noxious stimuli), and allodynia (painful responses to normally innocuous stimuli). In contrast to inflammatory pain, effective therapy for this neuropathic pain is lacking, and the underlying mechanisms are poorly understood. Following nerve injury, plastic changes occur in the expression of ion channels, receptors, neuropeptides, and signal transduction-related molecules in the PNS and CNS. In previous studies, much attention has been focused on the directly damaged primary afferents and their influence on the activity of spinal dorsal horn neurons (Woolf and Salter 2000; Scholz and Woolf 2002). However, there is compelling evidence indicating that glial cells in the spinal cord may also play a role in the pathogenesis of neuropathic pain (Watkins and Maier 2003; Marchand et al. 2005; Tsuda et al. 2005). Mitogen-activated protein kinases (MAPK) play a critical role in intracellular signal transduction and consist of extracellular signal-regulated protein kinases 1/2 (ERK1/2), p38 MAPK, and c-Jun N-terminal kinases 1/2 (JNK1/2) (Widmann et al. 1999; Chang and Karin 2001). Several lines of evidence indicate that nerve injury results in MAPK activation in spinal glial cells, and MAPK inhibitors diminish injury-induced pain hypersensitivity (Ji and Strichartz 2004; Katsura et al. 2006; Svensson et al. 2007).

In this study, we set out to investigate whether TLR3 in the spinal cord participates in pain hypersensitivity using the L5 spinal nerve ligation (SNL) model (Kim and Chung 1992). We now show that TLR3 is crucial for the activation of spinal glial cells and the development of tactile allodynia after peripheral nerve injury. Our findings point to the potential blockade of TLR3 in spinal cord glial cells as a new therapeutic strategy for pain caused by nerve injury.

Materials and methods

Animals

Male Sprague–Dawley rats weighing 200–250 g were used. Specific pathogen-free C57BL/6 mice, purchased from Clea Japan (Osaka, Japan), were also used. Tlr3−/− mice were backcrossed with C57BL/6 mice; F8 mice were used (Yamamoto et al. 2003). All mice weighed 25–30 g at the time of surgery. All procedures were approved by the Hyogo College of Medicine Committee on Animal Research and were performed in accordance with the National Institutes of Health guidelines on animal care.

Surgical procedures

All procedures were performed with the rats and mice under pentobarbital anesthesia (50 mg/kg, i.p.). Additional doses of the anesthetics were given as needed. In all animals, no surgery was performed on the right side. Special care was taken to prevent infection and to minimize the influence of inflammation. The hair of the animal’s lower back was shaved and the skin was sterilized with 0.5% chlorhexidine and covered with clean paper. Sterile operating instruments were used.

To produce an L5 SNL, a skin incision (3–4 cm) was made in the midline lumbar region (L4-S1). The L6 transverse process was identified, freed of muscular attachments, and partially removed with the help of bone ronguers. This exposed the L5 spinal nerve. The L5 ventral ramus was isolated, freed from the adjacent nerves and then the L5 spinal nerve was tightly ligated with silk suture and transected distal to the ligature. After surgery, the wound was washed with saline and closed in layers (fascia and skin) with 3-0 silk thread.

Polyinosine–polycytidylic acid injection and drug treatments

The intrathecal delivery of 10 μL of polyinosine–polycytidylic acid [poly(I : C)] (0.1, 1, 10, or 100 μg; Sigma, St Louis, MO, USA) or the p38 inhibitor, SB203580 (1 or 10 μg; Calbiochem, San Diego, CA, USA), was performed as previously described (Fukuoka et al. 2001). Briefly, a laminectomy of the L5 vertebra was performed under adequate anesthesia with sodium pentobarbital. The dura was cut, and a soft tube (Silascon; outer diameter 0.64 mm; Kaneka Medix, Osaka, Japan) was inserted into the subarachnoid space of the spinal cord, and the tip of the catheter was implanted at the L5 spinal segmental level. Most experiments were conducted at day 1 after 100 μg poly(I : C) injection.

Antisense-oligodeoxynucleotide (AS-ODN; 5′-AACAATTGCTTCAAGTCC-3′), mismatch ODN (MM-ODN; 5′-ACTACTACACTAGACTAC-3′), and FITC-labeled ODN directed to TLR3 were designed and manufactured by Biognostik (Gottingen, Germany). To obtain a sustained drug infusion, an Alzet osmotic pump (7 days pump, 1 μL/h; Durect, Cupertino, CA, USA) was filled with AS-ODN (0.05 or 0.5 nmol/μL) or MM-ODN (0.5 nmol/μL) in normal saline, and the associated catheter was implanted intrathecally 12 h before SNL or poly(I : C) injection, or 7 days after SNL. Normal saline was used as the vehicle control.

Behavioral tests

All rats and mice were tested for mechanical and heat hypersensitivity of the plantar surface of the hindpaw 1 day before surgery and 3, 5, 7, 10, 12, or 14 days after surgery. Room temperature (22–24°C) and humidity remained stable for all experiments. On each testing day, the rats and mice were brought into the behavior room 1 h prior to the test session to allow them to habituate to the environment.

Mechanical hypersensitivity was assessed with a Dynamic Plantar Aesthesiometer (Ugo Basile, Comerio, Italy), which is an automated von Frey-type system (Kalmar et al. 2003; Lever et al. 2003). To measure animal hindpaw mechanical thresholds, animals were placed in plastic cages with a wire mesh floor and allowed to acclimate for 15 min before each test session. A paw-flick response was elicited by applying an increasing force (measured in grams) using a plastic filament (0.5 mm diameter) focused on the middle of the plantar surface of the ipsilateral hindpaw. The force applied was initially below detection threshold and then increased from 1 to 5 or 50 g in 0.1 or 1 g steps over 20 s, then held at 5 or 50 g for a further 10 s. The rate of force increase was 0.25 or 2.5 g/s. The threshold was taken as the force applied to elicit a reflex removal of the hindpaw. This was defined as the mean of three measurements at 1 min intervals. The variability between trials was approximately 0.2 or 2 g. Heat hypersensitivity was tested using the Hargreaves radiant heat apparatus (7370; Ugo Basile). A radiant heat source beneath a glass floor was aimed at the plantar surface of the hindpaw. Three measurements of latency were taken for each hindpaw in each test session. The hindpaws were tested alternately, with intervals between consecutive tests of > 5 min. The three measurements of latency per side were averaged.

Data are expressed as mean ± SEM. Differences in changes of values over time of each group were tested using one-way anova, followed by individual post hoc comparisons (Fisher exact test). One-way anova, followed by individual post hoc comparisons (Fisher exact test) was used to assess differences of values between the intrathecal groups. A difference was accepted as significant if < 0.05.

Immunohistochemistry

The rats and mice were deeply anesthetized with sodium pentobarbital and perfused transcardially with 1%p-formaldehyde in 0.1 M phosphate buffer, pH 7.4, followed by 4%p-formaldehyde in 0.1 M phosphate buffer, 7 days after surgery (= 4 at each time point). After the perfusion, the L5 spinal cord segments were dissected out and post-fixed in the same fixative for 12 h, then replaced with 20% sucrose overnight. Transverse spinal sections (free-floating, 20 μm) were cut and processed for glial fibrillary acid protein (GFAP), ionized calcium-binding adapter molecule 1 (Iba1), phosphorylated-p38 (p-p38), and p-ERK1/2 immunohistochemistry according to previously described methods (Noguchi et al. 1995). The rabbit polyclonal primary antibody for GFAP (1 : 400; Dako, Glostrup, Denmark), Iba1 (1 : 400; Wako, Osaka, Japan), p-p38 (1 : 400; Cell Signaling Technology, Beverly, MA, USA), and p-ERK1/2 (1 : 400; Cell Signaling Technology) was used for diaminobenzidine tetrahydrochloride staining.

To quantify positive cell profiles in the spinal cord, four to six sections from the L5 spinal cord segments were randomly selected. An image in a square (316 × 236 μm) centered on the medial two-thirds of the superficial dorsal horn (laminae I–III), as described previously (Molander et al. 1984), was captured under an 20× objective, and quantitative assessment of immunostaining was carried out by determining the intensity of immunolabeling within the fixed area of the dorsal horn. Because a stereological approach was not used in this study, quantification of the data may represent a biased estimate of the actual number of cells and neurons. An assistant, who was unaware of the treatment group of the tissue sections, performed all of the counting.

Data are expressed as mean ± SD. One-way anova, followed by individual post hoc comparisons (Fisher exact test) or pair wise comparisons (t-test) were used to assess differences of values between the intrathecal groups. A difference was accepted as significant if < 0.05.

RT-PCR assay

For the RT-PCR, the rats were killed by decapitation under deep anesthesia 7 days after surgery and the L5 spinal cord segments were removed and rapidly frozen with powdered dry ice and stored at −80°C until ready for use. The procedure of extraction of total RNA using a RNA extraction regent Isogen (Nippon Gene, Tokyo, Japan) was described in our previous study (Fukuoka et al. 2001). PCR primers for TLR3, TLR4, Iba1, interleukin-1β (IL-1β), IL-6, tumor necrosis factor-α (TNF-α), and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) cDNA were designed corresponding to the coding region of the genes as follows: TLR3, 5′-CTGGAGCCAGAACTGTGCCA-3′ and 5′-CGCGGAGGCTGTTGTAGGAA-3′; TLR4, 5′-CTGCCTGAGACCAGGAAGCT-3′ and 5′-CCCTGAAAGGCTTGGGCTTG-3′; Iba1, 5′-CCATGAAGCCTGAGGAAATT-3′ and 5′-CCACCTCCAATTAGGGCA-3′; IL-1β, 5′-CCGTGGAGCTTCCAGGATGA-3′ and 5′-GCTCTGCTTGAGAGGTGCTG-3′; IL-6, 5′-CCACTGCCTTCCCTACTTCA-3′ and 5′-GCTCTGAATGACTCTGGCTT-3′; TNF-α, 5′-GCCCACGTCGTAGCAAACCA-3′ and 5′-GGGCTCATACCAGGGCTTG-3′; and Gapd (GAPDH), 5′-TGCTGGTGCTGAGTATGTCG-3′ and 5′-GCATGTCAGATCCACAACGG-3′. Subsequent PCR reaction was performed in a 50 μL solution of 1x PCR buffer (Perkin Elmer, Boston, MA, USA), 0.2 mM dNTP, and 1.25 U AmpliTaq (Perkin Elmer) with a pair of 20 pmol primers on Perkin DNA Thermal Cycler (Perkin Elmer) and the PCR program was 15 s at 94°C, 15 s at 57°C, and 45 s at 72°C. The intensity of stained bands was measured with a computer-assisted imaging analysis system (atto Densitograph version 4.02, Atto, Tokyo, Japan). The density of PCR product bands of TLR3, TLR4, Iba1, IL-1β, IL-6, TNF-α, and GAPDH mRNAs was increased between 25 and 35 PCR cycles, depending on the number of cycles, therefore, the number of PCR cycles of 30 was used. The ratio of TLR3, TLR4, Iba1, IL-1β, IL-6, or TNF-α to GAPDH mRNAs was considered to indicate the level of each transcript. The mRNA level was expressed as a percentage of the mRNA level in the normal control ganglia. Samples without the addition of reverse transcriptase or without the addition of RNA (negative controls) revealed no detectable product.

Data are expressed as mean ± SD. Differences in changes of values between the intrathecal groups were tested using one-way anova, followed by individual post hoc comparisons (Fisher exact test) or pair wise comparisons (t-test). A difference was accepted as significant if < 0.05.

Results

Inhibition of TLR3 expression attenuates nerve injury-induced tactile allodynia

To elucidate whether suppressing TLR3 in the spinal cord might prevent nerve injury-induced allodynia and hyperalgesia, rats with nerve injury were intrathecally treated with either an AS-ODN targeting TLR3 or a MM-ODN beginning 12 h before L5 SNL. We found that TLR3 AS-ODN significantly inhibited nerve injury-induced tactile allodynia, but not heat hyperalgesia, at days 3, 5, and 7 after surgery (Fig. 1a). Indeed, the level of TLR3 in the ipsilateral spinal cord of the AS-ODN-treated rats was significantly lower than that in the MM-ODN-treated rats, whereas there was no difference in TLR4 expression at day 7 (Fig. 1b). Intrathecal administration of AS-ODN in naive rats produced no significant changes in basal pain sensitivity (data not shown).

Figure 1.

 Attenuated tactile allodynia by a selective blockade of TLR3 expression. (a) Pre-treatment with TLR3 AS-ODN (0.5 nmol/μL/h) prevented mechanical, but not heat, hypersensitivity induced by nerve injury. Tactile allodynia was determined with a Dynamic Plantar Aesthesiometer at days 3, 5, and 7 after nerve injury. Heat hyperalgesia was tested using the plantar test. BL, baseline. Data represent mean ± SEM; = 8 per group. #< 0.05 compared with the MM-ODN (0.5 nmol/μL/h) group. (b) mRNA expression of TLR3 and TLR4 in the ipsilateral spinal cord at day 7 after L5 SNL, as detected by RT-PCR. Quantification of RT-PCR data is shown at right. Data represent mean ± SD; = 4 per group. #< 0.05 compared with the MM-ODN (0.5 nmol/μL/h) group.

To investigate whether inhibition of TLR3 expression would reverse established tactile allodynia, a treatment mode more relevant to clinical situation, we infused TLR3 AS-ODN intrathecally via an osmotic pump at day 7 after establishment of L5 SNL-induced tactile allodynia. This treatment did not reverse the nerve injury-induced allodynia (Fig. 2a). We further observed that TLR3-deficient mice displayed significantly attenuated tactile allodynia at days 3, 5, and 7 after surgery, compared with the wild-type controls, whereas TLR3-deficient mice developed robust heat hyperalgesia (Fig. 2b). These results suggest that TLR3 in the spinal cord might have a role in the early establishment of tactile allodynia, but not heat hyperalgesia, after peripheral nerve injury.

Figure 2.

 TLR3 contributes to the initial development of tactile allodynia after nerve injury. (a) Post-treatment with TLR3 AS-ODN did not reverse injury-induced tactile allodynia. AS-ODN was delivered intrathecally via an osmotic pump starting on post-SNL day 7. Data represent mean ± SEM; = 8 per group. (b) Mechanical, but not heat, hypersensitivity was significantly attenuated at days 3, 5, and 7 after L5 SNL in TLR3 knockout (KO) mice relative to wild-type (WT) mice. BL, baseline. Data represent mean ± SEM; = 8 per group. #< 0.05 compared with the WT control.

Down-regulation of TLR3 reduces the phosphorylation of p38 MAPK, but not ERK1/2, in spinal glial cells caused by nerve injury

Increasing evidence shows that ERK1/2 and p38 MAPK activation in spinal glial cells contributes to the generation of neuropathic pain (Ji and Strichartz 2004; Katsura et al. 2006; Svensson et al. 2007). For example, ERK1/2 is sequentially activated in neurons, microglia, and astrocytes by peripheral nerve injury (Zhuang et al. 2005), whereas nerve injury induces the activation of p38 MAPK, predominantly in spinal microglia (Jin et al. 2003; Katsura et al. 2006). To ascertain whether MAPK phosphorylation in the dorsal horn is regulated by TLR3, the levels of p-ERK1/2 and p-p38 were compared in the MM-ODN and AS-ODN groups (Fig. 3a and b). We found that TLR3 AS-ODN inhibited the injury-induced increase in p38, but not ERK1/2, phosphorylation at day 7 after nerve injury. These data suggest that TLR3 in spinal cord glial cells participates in tactile allodynia through the p38 MAPK pathway.

Figure 3.

 Effects of TLR3 AS-ODN on the phosphorylation of p38 MAPK and ERK1/2 in the spinal cord dorsal horn after nerve injury. (a) p-p38 and p-ERK1/2 immunostaining in the superficial dorsal horn (laminae I–III) in the MM-ODN and AS-ODN groups at day 7 after L5 SNL. (b) The intensity of p-p38 and p-ERK1/2 determined as the average pixel density in the ipsilateral dorsal horn. Results are presented as percentages of the values normalized to the mean values of the MM-ODN group. Data represent mean ± SD; = 4 per group. #< 0.05 compared with the MM-ODN group. Scale bar: 100 μm.

Down-regulation of TLR3 reduces the induction of spinal microglial markers and pro-inflammatory cytokines caused by nerve injury

After peripheral nerve injury, spinal glial cells change their morphology and number in the dorsal horn (Watkins and Maier 2003; Marchand et al. 2005; Tsuda et al. 2005). To explore the mechanisms by which TLR3 may mediate tactile allodynia, we assessed the effect of TLR3 AS-ODN on nerve injury-induced activation of spinal microglia and astrocytes. We found that the number and intensity of Iba1-immunoreactive cells in the ipsilateral dorsal horn increased strikingly at day 7 after surgery, but pre-treatment with AS-ODN blocked this increase (Fig. 4a and b). However, AS-ODN did not attenuate the injury-induced elevation of GFAP. We further observed that in injured TLR3-deficient mice, p-p38 and Iba1 immunoreactivity was reduced almost to baseline levels in the ipsilateral dorsal horn (Fig. 5a and b).

Figure 4.

 Effects of TLR3 AS-ODN on the activation of spinal microglia and astrocytes after nerve injury. (a) Iba1 and GFAP immunostaining in the superficial dorsal horn (laminae I–III) in the MM-ODN and AS-ODN groups at day 7 after L5 SNL. Antisense knockdown of TLR3 attenuated nerve injury-induced activation of spinal microglia, but not astrocytes. (b) The intensity of Iba1 and GFAP determined as the average pixel density in the ipsilateral dorsal horn. Data represent mean ± SD; = 4 per group. #< 0.05 compared with the MM-ODN group. Scale bar: 100 μm.

Figure 5.

 Knockout of TLR3 leads to attenuation in p38 MAPK phosphorylation and spinal microglial activation induced by nerve injury. (a) p-p38 immunostaining in the superficial dorsal horn (laminae I–III) in TLR3 KO mice and WT mice at day 7 after nerve injury. (b) Iba1 and GFAP immunostaining in the superficial dorsal horn in TLR3 KO mice and WT mice at day 7. Scale bar: 100 μm.

Spinal glial activation is likely involved in the production and release of pro-inflammatory cytokines, such as IL-1β, IL-6, and TNF-α, and thus increasing pain hypersensitivity (Watkins and Maier 2003; Marchand et al. 2005; Tsuda et al. 2005). We therefore examined the effect of TLR3 AS-ODN on nerve injury-induced up-regulation of these cytokines in the spinal cord using RT-PCR. Significantly lower spinal expression of mRNA for Iba1, IL-1β, IL-6, and TNF-α were observed at day 7 after nerve injury in the AS-ODN group, compared with the MM-ODN group (Fig. 6). Together, these findings indicate that after peripheral nerve injury, spinal microglial activation occurs through the TLR3/p38 signaling pathway and leads to the enhanced expression of pro-inflammatory cytokines in the spinal cord.

Figure 6.

 Effects of TLR3 AS-ODN on the induction of spinal pro-inflammatory cytokines after nerve injury. mRNA expression of Iba1, IL-1β, IL-6, and TNF-αin the ipsilateral spinal cord at day 7 after L5 SNL, as detected by RT-PCR. Quantification of RT-PCR data is shown at right. Data represent mean ± SD; = 4 per group. #< 0.05 compared with the MM-ODN group.

TLR3 agonist polyinosine–polycytidylic acid produces tactile allodynia, spinal microglial activation, and pro-inflammatory cytokines up-regulation

Microglia have been demonstrated to recognize poly(I : C) dsRNA via TLR3 (Olson and Miller 2004; Town et al. 2006). To further investigate the effects of TLR3 in the spinal cord on pain behaviors, we injected poly(I : C) intrathecally into naive rats. A single intrathecal injection of poly(I : C) produced tactile allodynia, but not heat hyperalgesia, at day 1 in a dose-dependent manner (Fig. 7a and b), and this allodynia persisted for 7 days (Fig. 7b). Selective p38 MAPK inhibitor, SB203580, dose-dependently blocked the poly(I : C)-induced allodynia, whereas it had no effects on the threshold of vehicle-treated rats at day 1 (Fig. 7c). When rats were pre-treated intrathecally with TLR3 AS-ODN before poly(I : C) treatment, the development of poly(I : C)-induced allodynia was completely prevented, whereas the MM-ODN had no effect (Fig. 7d). Similar blockade of poly(I : C)-induced tactile allodynia was also observed in TLR3-deficient mice at day 1 (Fig. 7e). Intrathecal injection of poly(I : C) caused an increase in the expression of p-p38, but not ERK1/2, in the spinal dorsal horn at day 1 after injection (Fig. 8a and b). Up-regulation of Iba1, but not GFAP, was also observed after poly(I : C) (Fig. 8c and d), and these increases in p-p38 and Iba1 were abolished by TLR3 AS-ODN (Fig. 8a–d). We further found that in TLR3-deficient mice, poly(I : C) injection did not induce an increase in p-p38 and Iba1 in the dorsal horn (Fig. 9a and b). TLR3 AS-ODN also led to a significant decrease in the spinal expression of Iba1, IL-1β, IL-6, and TNF-α mRNA at day 1 after poly(I : C) (Fig. 10).

Figure 7.

 Induction of behavioral allodynia by poly(I : C) dsRNA through the p38 MAPK pathway. (a) Intrathecal administration of poly(I : C) induced tactile allodynia at day 1 in a dose-dependent manner. Data represent mean ± SEM; = 8 per group. #< 0.05 compared with the vehicle control group. (b) Time course for poly(I : C)-induced tactile allodynia and heat hyperalgesia. #< 0.05 compared with the pre-operative value. (c) Prevention of poly(I : C) action by pre-treatment with a p38 MAPK inhibitor, SB203580. #< 0.05 compared with the vehicle control group. (d) Effects of TLR3 AS-ODN on tactile allodynia induced by poly(I : C) injection. #< 0.05 compared with the MM-ODN group. (e) Absence of poly(I : C)-induced allodynia in TLR3 KO mice. Data represent mean ± SEM; = 8 per group. #< 0.05 compared with the WT control.

Figure 8.

 Antisense knockdown of TLR3 inhibited poly(I : C)-induced p38 MAPK phosphorylation and spinal microglial activation. (a) p-p38 and p-ERK1/2 immunostaining in the dorsal horn at day 1 after poly(I : C). (b) The intensity of p-p38 and p-ERK1/2 determined as the average pixel density in the dorsal horn. (c) Iba1 and GFAP immunostaining in the dorsal horn at day 1 after poly(I : C). (d) The intensity of Iba1 and GFAP determined as the average pixel density in the dorsal horn. Data represent mean ± SD; = 4 per group. #< 0.05 compared with the MM-ODN group. Scale bar: 100 μm.

Figure 9.

 Knockout of TLR3 attenuates p38 MAPK phosphorylation and spinal microglial activation induced by poly(I : C) injection. (a) p-p38 immunostaining in the superficial dorsal horn (laminae I–III) in TLR3 KO mice and WT mice at day 1 after poly(I : C). (b) Iba1 and GFAP immunostaining in the superficial dorsal horn in TLR3 KO mice and WT mice at day 1. Scale bar: 100 μm.

Figure 10.

 Antisense knockdown of TLR3 inhibited poly(I : C)-induced up-regulation of spinal pro-inflammatory cytokines. mRNA expression of Iba1, IL-1β, IL-6, TNF-α, TLR3, and TLR4 in the spinal cord at day 1 after poly(I : C) injection, as detected by RT-PCR. Quantification of RT-PCR data is shown at right. Data represent mean ± SD; = 4 per group. #< 0.05 compared with the MM-ODN group.

Discussion

The present study demonstrated the following new findings: (i) TLR3 AS-ODN alleviated L5 SNL-induced tactile allodynia, but not heat hyperalgesia, and prevented the activation of p38 MAPK, but not ERK1/2, in spinal glial cells. (ii) TLR3 AS-ODN attenuated L5 SNL-induced up-regulation of Iba1, IL-1β, IL-6, and TNF-α, but not GFAP, in the spinal cord. (iii) Intrathecal injection of poly(I : C) produced both tactile allodynia and up-regulation of p-p38, Iba1, IL-1β, IL-6, and TNF-α, which were prevented by TLR3 AS-ODN. (iv) TLR3-deficient mice did not develop tactile allodynia after L5 SNL or poly(I : C) injection.

There is accumulating evidence supporting a role for activated microglia in the pathogenesis of nerve injury-induced pain hypersensitivity. For example, the ATP receptor, P2X4, in the spinal dorsal horn is selectively expressed in activated microglia and contributes to tactile allodynia after nerve injury (Tsuda et al. 2003). Furthermore, the chemokine receptor 2 is also expressed in spinal microglia, and chemokine receptor 2-deficient mice do not develop tactile allodynia after nerve injury (Abbadie et al. 2003). In the present study, we found that TLR3 AS-ODN diminished nerve injury- or poly(I : C)-induced tactile allodynia and p38 MAPK activation in glial cells in the spinal dorsal horn. Furthermore, TLR3-deficient mice displayed significantly attenuated tactile allodynia and decreased levels of p-p38 in spinal glial cells, compared with the wild-type controls. There is compelling evidence indicating that MAPK activation in spinal glial cells has a critical role in the induction and maintenance of neuropathic pain (Ji and Strichartz 2004). Indeed, peripheral nerve injury induces the activation of p38 MAPK, predominantly in spinal microglia (Jin et al. 2003; Katsura et al. 2006). Taken together, these findings suggest that nerve injury induces p38 MAPK activation in spinal cord microglia via TLR3 and that this activation of the TLR3/p38 MAPK signaling cascade in spinal microglia contributes to tactile allodynia.

Upon binding of ligand, TLRs have been shown to activate a variety of signaling pathways, including phosphoinositide 3-kinase, ERK1/2, p38, JNK1/2, and nuclear factor-kappa B, each leading to the induction of numerous target genes involved in inflammation, cellular differentiation, and direct antimicrobial activity (Akira and Takeda 2004; Akira et al. 2006; Marshak-Rothstein 2006). MAPK activation appears to regulate the expression of pro-inflammatory cytokines, including IL-1β, IL-6, and TNF-α, as well as cyclooxygenase-2 and inducible nitric oxide synthase (Koistinaho and Koistinaho 2002; Ji and Strichartz 2004). In the present study, we found that AS knockdown of TLR3 expression reduced tactile allodynia, spinal microglial activation, and pro-inflammatory cytokines up-regulation induced by nerve injury or poly(I : C) injection. Because p38 MAPK activation can cause microglial activation in the brain (Koistinaho and Koistinaho 2002), our findings suggest that after nerve injury, microglial activation might occur through the TLR3/p38 MAPK signaling pathway and lead to the enhanced expression of spinal pro-inflammatory cytokines that sensitize dorsal horn neurons in the spinal cord. However, we cannot exclude the possibility that the effect of TLR3 phosphorylation on microglial activation might also function indirectly through the increase of pro-inflammatory cytokines.

The present study demonstrated that TLR3 AS-ODN did not attenuate nerve injury-induced activation of ERK1/2 in spinal glial cells. Furthermore, intrathecal injection of poly(I : C) did not produce the phosphorylation of ERK1/2 in the dorsal horn. Previous reports have shown that p38 MAPK, especially p38 β-isoform, is activated in microglia in the dorsal horn (Jin et al. 2003; Svensson et al. 2005; Katsura et al. 2006), whereas ERK1/2 is sequentially activated in neurons, microglia, and astrocytes by peripheral nerve injury (Zhuang et al. 2005). Our findings suggest that the TLR3/p38 MAPK signaling cascade participate in nerve injury-induced mechanical hypersensitivity, independent of ERK1/2 pathway. Indeed, either a p38 or ERK1/2 inhibitor only partially suppresses L5 SNL-induced mechanical hypersensitivity (Jin et al. 2003; Zhuang et al. 2005). Thus, we believe that blockade of both TLR3/p38 MAPK and ERK1/2 activation simultaneously may provide a more effective means to reduce neuropathic pain. Not only microglia but also astrocytes are activated in the spinal cord following nerve injury, and these activated astrocytes participate in the maintenance of the late phase of neuropathic pain (Watkins and Maier 2003; Marchand et al. 2005; Tsuda et al. 2005). For example, nerve injury induces JNK1/2 activation in spinal astrocytes 2–3 weeks after injury (Zhuang et al. 2006). Because microglial responses typically precede astrocyte activation, TLR3/p38 MAPK activation in microglia might have a role in the early establishment of neuropathic pain. Indeed, post-treatment with TLR3 AS-ODN did not reverse injury-induced tactile allodynia.

In the present study, we also found that TLR3 AS-ODN did not reduce nerve injury-induced GFAP up-regulation in the spinal dorsal horn. Furthermore, intrathecal injection of poly(I : C) did not induce an increase in GFAP expression in the spinal cord. Therefore, our data suggest that the activation of spinal astrocytes might not occur through the TLR3 signaling pathway. However, a recent report showed that neuropathic pain can be blocked by JNK inhibitors without accompanying reduction of GFAP expression, indicating that spinal GFAP up-regulation may be just associated with nerve injury but not a cause for neuropathic pain (Zhuang et al. 2006). Therefore, we cannot deny the possibility that TLR3 might be expressed in astrocytes in the spinal dorsal horn and contribute to the generation of neuropathic pain. Further studies are necessary to establish the direct relationship between the up-regulation of GFAP following nerve injury and the incidence of tactile allodynia, as well as the localization of TLR3.

Transient receptor potential ion channel (TRPV1), one of the transducer proteins, can generate depolarizing currents in response to noxious thermal stimuli, with an activation temperature of approximately 43°C (Moran et al. 2004; Dhaka et al. 2006). TRPV1 up-regulation in undamaged primary sensory neurons has been implicated in nerve injury-induced heat hyperalgesia (Ji et al. 2002; Obata et al. 2006). In the present study, AS knockdown of TLR3 expression did not attenuate nerve injury-induced heat hyperalgesia. Furthermore, a single intrathecal injection of poly(I : C) did not produce heat hyperalgesia. Therefore, these findings suggest that TLR3 might not be involved in the TRPV1 increase in sensory neurons, and heat hyperalgesia after peripheral nerve injury. Some reports have shown that activated glial cells participate in heat hyperalgesia, as well as tactile allodynia after nerve injury (Tanga et al. 2005). Nevertheless, our results indicate that TLR3 in spinal glial cells has a crucial role in the pathogenesis of tactile allodynia rather than heat hyperalgesia after peripheral nerve injury. In fact, TLR3-deficient mice developed robust heat hyperalgesia after nerve injury, and further, poly(I : C) injection did not produce heat hyperalgesia in rats.

Here, we show that TLR3 in spinal glial cells is required for tactile allodynia after nerve injury. There is growing appreciation that signaling through TLRs, which is key to generating innate responses to infection, may have pathogen-independent roles, by initiating responses to host-derived, endogenous ligands (Akira and Takeda 2004; Akira et al. 2006; Marshak-Rothstein 2006). Intriguingly, recent reports have shown that endogenous RNA, released from or associated with necrotic cells, could serve as an endogenous ligand for TLR3 (Kariko et al. 2004; Brentano et al. 2005). Indeed, TLR3 is localized intracellularly and recognizes nucleic acids in late endosomes-lysosomes (Lee et al. 2006). Although the discrimination between nucleic acids of mammalian versus microbial origin by TLR3 needs further investigation (Kanzler et al. 2007), host-derived nucleic acids may become available to TLR3 and serve as endogenous ligands under certain conditions, such as nerve injury. Considering that TLR4 and TLR2 up-regulation in spinal glial cells contributes to nerve injury-induced pain states (Tanga et al. 2005; Kim et al. 2007), the participation of TLRs in the nociceptive pathway may provide further insight into the function of these receptors.

Acknowledgements

This work was supported in part by Grants-in-Aid for Scientific Research, and an Open Research Center grant, Hyogo College of Medicine, both from the Japanese Ministry of Education, Science, and Culture. This work was also supported by a grant from the Japan Health Sciences Foundation. We thank Y. Wadazumi and N. Kusumoto for technical assistance. We thank D. A. Thomas for correcting the English usage.

Ancillary