Nerve Conduction Blockade in the Sciatic Nerve Prevents but Does Not Reverse the Activation of p38 Mitogen-activated Protein Kinase in Spinal Microglia in the Rat Spared Nerve Injury Model

Male adult Sprague-Dawley rats (240–300 g) were housed (two per cage) with free access to water and rat chow. The animal room was artificially illuminated on a 12-h day–night cycle and maintained at ambient temperature (22°–23°C). All the animals were used under animal care institutional guidelines for Harvard Medical School, Boston, Massachusetts, and Lausanne University, Lausanne, Switzerland.

As described in our previous study,26bupivacaine microspheres were used to produce a slow release of the local anesthetic and a long-term conduction blockade of the sciatic nerve. The sciatic nerve was exposed from the emergence of the musculocutaneous branch to the trifurcation of the sural, tibial, and peroneal branches. A silicone tube (11 mm long) was incised on its long axis and placed carefully around the sciatic nerve. Two 5-0 silk ligatures were used to close the longitudinal slit. The distal end of the tube was sealed with fibrin glue (Tissucol; Baxter, Volketswil, Switzerland). Bupivacaine microspheres containing 75% bupivacaine (wt/wt) (provided by Dr. Charles Berde) were prepared in a fibrin solution (300 mg/ml). This solution also contained low concentration of dexamethasone (0.05%), which might increase the penetration of bupivacaine to nerve fibers, therefore prolonging the duration of nerve block.27The mixture solution (bupivacaine microsphere in fibrin solution, 50 μl) was then slowly poured inside the tube through the proximate end.26Special care was taken to avoid any stretch or damage to the sciatic nerve and its branches. A sham blockade procedure was performed by filling the tube with fibrin without bupivacaine microspheres, followed by SNI (vehicle control) or sham surgery (sham control). An additional control group was added by filling the perisciatic tube with a solution of fibrin containing dexamethasone (0.05%). All the animals received a subcutaneous injection of amoxicillin (150 mg/kg) before surgery. To block nerve injury–induced immediate discharge, SNI was performed 3 h after nerve blockade, because a complete nerve blockade can be achieved 3 h after treatment with bupivacaine microspheres.

To test the efficacy of the nerve blockade, rats were allowed to recover from anesthesia and then gently held with a cloth wrapped above their waist to restrain upper extremities, and sensitivity measurements were performed. Measurements were also performed daily for 3 days after nerve blockade.26The nociceptive sensory function was examined with noxious mechanical stimulation (pinch with forceps) and noxious heat stimulation (hot plate testing for the ipsilateral or contralateral hind paws, with 12 s as cutoff to avoid tissue injury). The motor function was examined by the presence or absence of hopping response and tactile placing.26 

Although we initially found p38 activation in the spinal nerve ligation model,6,28we chose the SNI model29in this study because (1) nerve conduction blockade is easier to perform on the sciatic nerve than on the spinal nerve and (2) the role of p38 has not been examined in the SNI model. To produce SNI, two of the three sciatic nerves' terminal branches, the common peroneal and tibial nerves, were ligated by a 5-0 silk suture and cut, and a 3-mm portion of the nerves was removed. The third branch, the sural nerve, was untouched, and great care was taken not to stretch this nerve during the surgical procedure. All the animals were killed 3 days after SNI, because at this time point (1) mechanical allodynia is fully developed and (2) p38 is induced in spinal microglia.6 

The p38 inhibitor FR167653 was provided by Fujisawa Pharmaceutical Company (Osaka, Japan). For the study of pretreatment (treatment given before and during SNI), an osmotic pump (0.5 μl/h, 7-day capacity; Alzet, Cupertino, CA) was filled with FR167653 (30 μg/μl, n = 5) or vehicle (30% dimethyl sulfoxide [DMSO], n = 8) to produce stable and sustained infusion of the inhibitor, starting 2 days before SNI. Laminectomy was performed under a microscope at the spinal level 2 cm below the lumbar enlargement. A polyethylene (PE5) catheter was implanted into the intrathecal space of the spinal cord (L4–L5). The other end of the catheter was connected to the pump. After filling, the pump was soaked in saline for 3 h before implantation and placed subcutaneously on the back.30,31The concentrations of FR167653 (30 μg/μl) and DMSO (30%) for pump infusion were chosen on the basis of previous studies.15,30 

For single intrathecal injection, a spinal cord puncture was made with 27-gauge needle between L5 and L6 level to deliver the drug and vehicle to the cerebral spinal fluid space around the lumbosacral spinal cord.31,32Animals were briefly anesthetized with sevoflurane before the puncture, and 20 μl liquid was injected with a microsyringe. Immediately after the needle entry into subarachnoid space (change in resistance), a brisk tail flick was observed after the needle puncture. For single injection, R167653 was dissolved in 10% DMSO and intrathecally injected in 20 μl at the doses of 20, 50, and 100 μg (n = 8 for each dose). DMSO, 10% (20 μl), was also injected as vehicle control (n = 5). The doses of FR167653 and the concentration of DMSO for bolus injection were chosen on the basis of previous studies15,30and our pilot study.

Three days after SNI or sham surgery, animals (n = 4 for each group) were terminally anesthetized with isoflurane and perfused through the ascending aorta with saline followed by 4% paraformaldehyde with 1.5% picric acid in 0.16 m phosphate buffer (pH 7.2–7.4, 4°C). After the perfusion, the L4 spinal cord segments were removed, postfixed in the same fixative for 24 h, and then replaced with 15% sucrose overnight. Transverse spinal sections (free-floating, 30 μm) were cut in a cryostat and processed for immunofluorescence.6,30All of the sections were blocked with 2% goat serum in 0.3% Triton for 1 h at room temperature and incubated overnight at 4°C with anti–phospho-p38 (p-p38) antibody (rabbit, 1:500; Cell Signaling, Beverly, MA). The sections were then incubated for 1 h at room temperature with Cy3-conjugated secondary antibody (1:300; Jackson Immunolab, West Grove, PA). For double immunofluorescence, spinal sections were incubated with a mixture of polyclonal p-p38 and monoclonal OX-42 (for CD11b, a microglia marker, 1:5,000; Chemicon, Temecula, CA) antibodies overnight at 4°C, followed by a mixture of fluorescein isothiocyanate–conjugated and Cy3-conjugated secondary antibodies for 1 h at root temperature. The stained sections were examined with a Nikon fluorescence microscope (Tokyo, Japan), and images were captured with a Spot charge-coupled device microscope digital camera (Diagnostic Instruments, Sterling Heights, MI). The specificity of the staining and antibodies has been tested in previous studies.6,18 

The free-floating spinal sections prepared for immunostaining were also used for the staining of isolectin B4 (IB4). The spinal sections were incubated with IB4 (1:200; Sigma, St. Louis, MO) for 2 h in phosphate-buffered saline solution at room temperature.33The sections were then washed in phosphate-buffered saline and examined under a Nikon fluorescence microscope.

Western blotting was performed as previously reported.6,31,32Animals were rapidly killed 3 days after SNI or sham surgery (n = 8), and the L4–L5 spinal segments (dorsal part) were quickly removed and homogenized in a lysis buffer containing a cocktail of proteinase and phosphatase inhibitors (Sigma). Protein samples (40 μg) were separated on sodium dodecyl sulfate polyacrylamide gel and transferred to polyvinylidene fluoride blots (Millipore, Billerica, MA). The blots were blocked with 5% milk and incubated overnight at 4°C with antibody against p-p38 (anti-rabbit, 1:1,000; Cell Signaling), or p38 (anti-rabbit, 1:1,000; Cell Signaling), or glyceraldehyde-3-phosphate dehydrogenase (GAPDH; anti-rabbit, 1:10,000; Abcam, Cambridge, England). These blots were further incubated with horseradish peroxidase–conjugated secondary antibody (1:5,000; Amersham, Piscataway, NJ), developed in ECL solution (PerkinElmer, Boston, MA), and exposed onto hyperfilms (Amersham). The pixels of the specific bands were quantified with Image Quant TL software (Zurich, Switzerland), and the ratios of p-p38/p38, p38/GAPDH, and p-p38/GAPDH were determined.

Animals were habituated to the testing environment daily for 3 days before baseline testing. The room temperature and humidity remained stable for all experiments. For testing mechanical sensitivity, animals were put in plastic boxes (11 × 13 × 24 cm) on an elevated metal mesh floor and allowed 30 min for habituation, and mechanical allodynia was tested using von Frey filaments. Mechanical paw withdrawal thresholds were determined using the up–down method described by Chaplan et al.  34Hind paws were pressed with one of a series von Frey hairs with logarithmically incremental stiffness (0.6–26 g; Stoelting, Wood Dale, IL), presented perpendicular to the plantar surface for 3–5 s for each hair.31The 50% withdrawal threshold was determined using the up–down method of Dixon.35The experimenter was not aware of drug treatment.

Five nonadjacent sections (30 μm) from the L4 spinal cord segments were randomly selected, and the numbers of p-p38–immunoreactive cell profiles with morphology of microglia were counted. The region of the dorsal horn sampled in all experiments is the medial two thirds of the dorsal horn (laminae I–III), where most p-p38–immunoreactive cells were found after SNI. This area was captured under 20× objective in a square box (450 × 338 μm), with the top line of the square in parallel with the edge of the gray matter of the dorsal horn.18,36The size of the box was kept the same in all conditions. The observer was unaware of the treatments during counting.

To minimize the number of animals used in this study, we included four rats per group for quantification of immunohistochemistry and Western blotting18,30,33,36and five to eight rats per group for behavioral studies.18,31Because paw withdrawal thresholds determined by up–down method passed normality test for all the groups, we compared differences using parametric tests as previously shown.37Differences were compared (GraphPad InStat; GraphPad Software, Inc., San Diego, CA, and SPSS 10; SPSS Inc., Chicago, IL) using the Student t  test (two groups) or one-way analysis of variance (multiple groups) followed by the Fisher protected least significant difference post hoc  test.18,31Data are presented as mean ± SEM. The criterion for statistical significance was P < 0.05. The effect sizes were also estimated using Cohen's d  values that are larger than 1.8 when P < 0.05.

Isolectin B4 has been widely used to label nonpeptidergic primary sensory neurons in the dorsal root ganglion. IB4 binding was also found in the central terminals of dorsal root ganglion neurons in lamina II of the spinal cord. After peripheral nerve injury, IB4 binding is rapidly lost in the dorsal root ganglion neurons and their central terminals.38Therefore, we used IB4 as a marker to characterize the projection of injured and intact nerve axons after SNI.

Three days after SNI, there was a significant reduction of IB4 binding in the medial two thirds of the dorsal horn (figs. 1A and B), which is terminated by the injured tibial and peroneal nerve. However, IB4 binding in the lateral one third of the dorsal horn, which is terminated by the intact sural nerve,39remained unchanged (figs. 1A and B). This IB4 staining pattern validated the SNI model we produced.

A specific anti–p-p38 antibody was used to study changes in p38 phosphorylation. The basal expression of p-p38 was low in the noninjured (contralateral) side of the spinal cord after nerve injury, as previously shown.6SNI induced a marked increase in p-p38 expression in the spinal cord 3 days after nerve injury (figs. 1C and D). Numerous p-p38–immunoreactive cells were found in the ipsilateral spinal cord, especially in the medial two thirds of superficial dorsal horn (laminae I–III; figs. 1C and D). This result indicates that p38 is mainly activated in the spinal region innervated by the injured axons after SNI.

To further examine the SNI-induced changes in spinal microglia, we also performed immunohistochemistry using OX-42 antibody, which labels the microglial surface marker CD11b (also called complement receptor 3 or Mac-1). SNI induced robust increase in OX-42 immunoreactivity also in the medial two thirds of the superficial dorsal horn (laminae I–III), 3 days after injury (figs. 1E and F). This increase is more likely to reflect an increase in OX-42 expression or activation of microglia rather than a proliferation of microglia. Double immunofluorescence showed that p-p38 was almost completely colocalized with OX-42 (figs. 2A–C), indicating that p38 is exclusively activated in spinal microglia on post-SNI day 3.

To confirm p-p38 increase after SNI, we also performed Western analysis using dorsal horn tissues from rats 3 days after SNI or sham surgery. When normalized to total p38 levels, SNI only induced a 26% increase in p-p38 levels (unpaired t  test, P = 0.0114, t = 3.031, df = 11; figs. 2D and E). When normalized to GAPDH, SNI also induced a 25% increase in p38 levels (unpaired t  test, P = 0.0196, t = 3.158, df = 6). Thus, both phosphorylated and total p38 levels were increased after SNI, and the actual increase of dorsal horn p-p38 levels is 63% when normalized to GAPDH (fig. 2E). However, SNI only induced p-p38 increase in a very restricted area of the dorsal horn, i.e. , in the medial superficial dorsal horn of the gray matter. Therefore, collecting the whole dorsal horn (a quarter of spinal cord including both white and gray matter) for Western analysis largely masked the p-p38 increase. In the remaining studies, we measured p-p38 immunoreactivity in microglia of the medial superficial dorsal horn to quantify p-p38 levels, which showed a fivefold increase after SNI (see next section).

As we previously reported,26all animals demonstrated complete sensory and motor blockade 3 h after application of bupivacaine microspheres to the sciatic nerve that was checked before SNI. The block maintained for the whole period of study (3 days). Therefore, none of the rats receiving bupivacaine microspheres responded to noxious heat stimulus in hot plate testing (P < 0.0001, t = 20.49, df = 13; fig. 3). Nor did they respond to skin pinch in the territory of the sciatic nerve. In addition, paw positioning and hopping responses were abolished. In contrast, all sensory and motor functions were preserved in rats receiving a sham procedure for nerve blockade (vehicle). A complete sensory blockade lasted more than a week, without signs of nerve damage, because all of the animals recovered to full responsiveness in sensory and motor tests when the block wore off.26 

To investigate whether injury-induced spontaneous activity generated in axons is essential for the induction of p38 activation, we applied bupivacaine microspheres to the sciatic nerve (proximal to the surgery site) before SNI and examined spinal p38 activation 3 days after SNI. A comparison among four groups (naive, sham, SNI, and SNI plus nerve block) was statistically significant in the number of p-p38–immunoreactive cells (analysis of variance, F^3,12= 22.318, P < 0.001). A post hoc  comparison showed no difference between naive and sham control rats, but SNI induced a fivefold increase in p-p38 immunoreactivity (P < 0.001, compared with sham control). The nerve blockade completely prevented SNI-induced increase in spinal p-p38 immunoreactivity (figs. 4A–C) compared with the SNI group (post hoc  comparison, P < 0.001).

To examine whether the low concentration of dexamethasone (20 μg, 0.05%) contained in the blocking solution has any effect on SNI-induced p38 activation, we included an additional control group receiving treatment, before SNI, of a fibrin solution that contains dexamethasone but not bupivacaine. Dexamethasone did not inhibit SNI-induced p38 activation: The numbers of p-p38–immunoreactive cells/spinal section in the dexamethasone-treated and vehicle (fibrin solution only)–treated groups were 76 ± 3 and 73 ± 9, respectively (t = 0.2338, P = 0.8244, df = 5). Therefore, suppression of p38 activation by blockade solution is not caused by low concentration of dexamethasone.

To investigate whether injury-induced spontaneous activity generated in axons is essential for the maintenance of p38 activation, we also applied bupivacaine microspheres to the sciatic nerve 2 days after SNI and examined spinal p38 activation 24 h later on post-SNI day 3. This delayed nerve block did not reverse SNI-induced p38 activation (t = 1.774, P = 0.1265, df = 6; fig. 5).

To examine whether this treatment of bupivacaine microspheres would produce systemic effects, we applied bupivacaine microspheres to the contralateral (noninjured) sciatic nerve before the ipsilateral nerve injury and examined p-p38 expression on post-SNI day 3. This contralateral blockade did not affect the activation of p38 in the ipsilateral spinal cord (t = 0.7470, P = 0.4833, df = 6; fig. 6A).

To determine whether this treatment would affect basal activation of p38 on the noninjured side, we also examined p-p38 expression on the contralateral spinal cord after block. This contralateral nerve blockade did not affect p-p38 levels on the contralateral (noninjured) side (t = 0.4368, P = 0.6775, df = 6), suggesting that basal activation of p38 does not require nerve activity (fig. 6B).

To determine whether p38 activation is required for the development of neuropathic pain in the SNI model that may have different sensitivity to drug treatment compared with other neuropathic pain models,40we examined whether intrathecal infusion of the p38 inhibitor FR167653 would prevent SNI-induced neuropathic pain. As previously shown,29SNI induced rapid mechanical allodynia that was robust on day 2 (fig. 7A). Intrathecal infusion of FR167653 (30 μg/μl) via  an osmotic pump, starting 2 days before SNI and continuing to post-SNI day 3, did not affect the basal mechanical pain threshold, but completely prevented SNI-induced mechanical allodynia on SNI day 2 (t = 5.514, df = 11, P = 0.0002) and day 3 (t = 6.868, df = 11, P < 0.0001; fig. 7A).

To further determine whether p38 activation is also essential for the maintenance of neuropathic pain, we gave a bolus injection of FR167653 (20, 50, or 100 μg) into the spinal cerebrospinal fluid space by lumbar puncture on post-SNI day 3. FR167653 produced a dose-dependent reversal of mechanical allodynia at 3 h after the injection (analysis of variance, F^3,25= 6.728, P = 0.0018; fig. 7B). The antiallodynic effect of FR167653 was maintained at 6 h after the injection (analysis of variance, F^3,25= 3.050, P = 0.0471; fig. 7B) but completely recovered 24 h later. Vehicle (10% DMSO) had no effect on SNI-induced mechanical allodynia (fig. 7B).

We have made several interesting observations in this study. First, SNI activated p38 mitogen-activated protein kinase in microglia in the medial two thirds of dorsal horn, which is terminated by the injured tibial and peroneal nerve. Second, intrathecal administration of the p38 inhibitor FR167653 both prevented and reversed SNI-induced mechanical allodynia on day 3. Third, nerve conduction block in the sciatic nerve with bupivacaine microspheres prevented but did not reverse spinal p38 activation in spinal microglia. Fourth, application of bupivacaine microspheres to the contralateral nerve did not prevent SNI-induced p38 activation.

Accumulating evidence suggests a glial activation of p38 in the spinal cord. We initially found that p-p38 is present in spinal glial cells and that chronic inflammation by complete Freund adjuvant induces a modest increase in spinal p-p38 levels.33Inflammation by formalin, capsaicin, or complete Freund adjuvant was also shown to activate p38 in spinal microglia.41–44In these inflammatory pain models, activation of peripheral nociceptors by inflammatory mediators may produce spontaneous activity in the axons and cell bodies of the primary sensory neurons.42Therefore, p38 activation in spinal microglia under inflammatory pain conditions could be activity dependent. However, in a recent study, application of bupivacaine (repeated 6 h later) to the sciatic nerve did not block spinal p38 activation examined 15 h after hind paw carrageenan injection.43This failure may be caused by incomplete conduction blockade due to afferent input via  saphenous nerves.43 

Compared with inflammation, peripheral nerve injury produces a more robust activation of p38 in spinal microglia.6,13p38 activation was also found in spinal microglia after spinal cord injury, a central neuropathic pain condition.44,45Growing evidence from different laboratories demonstrates that p38 inhibitors can attenuate neuropathic pain in different animal models.6,12–17In the current study, we further confirmed microglial activation of p38 in another neuropathic pain condition after SNI. We have also shown that a different p38 inhibitor, FR167653, which can inhibit p38 activation,46prevents and reverses mechanical allodynia by SNI.

A major finding of this study is that peripheral nerve blockade can prevent p38 activation in the spinal cord, suggesting that spontaneous activity, especially in the neuroma, is indispensable for initial p38 activation in spinal microglia. Spontaneous activity may release neurotransmitters (glutamate, adenosine triphosphate, substance P) from central terminals of the primary afferents to activate p38 in spinal microglia. Recently, we have shown that nerve injury may cause the release of the chemokine fractalkine, which is sufficient and required to produce spinal p38 activation by activating the CX3CR1 receptor expressed by microglia.18However, nerve conduction blockade in the sciatic nerve 2 days after SNI cannot reverse p38 activation in spinal microglia, indicating that maintenance of p38 activation may not require spontaneous activity. After activation, microglia produce multiple inflammatory mediators, such as the cytokines interleukin 1β, interleukin 6, and tumor necrosis factor α,47which could maintain p38 activation in spinal microglia, even in the absence of primary afferent input.

It is generally believed that spontaneous activity or ectopic discharge induced by peripheral nerve injury is responsible for the generation of neuropathic pain.19–25Local anesthetics are generally used to block nerve conduction and suppress spontaneous activity. The concentrations of local anesthetics required for blocking spontaneous activity are often lower than that required for blocking nerve conduction.48,49Local anesthetics may produce nonspecific effects, such as neurotoxicity and inhibition of fast axonal transport.2,50Because cytokines such as tumor necrosis factor α have been implicated in spinal p38 activation,12,16we cannot exclude the possibilities that bupivacaine may inhibit spinal p38 activation by suppressing (1) the local production of tumor necrosis factor α in the sciatic nerve (e.g. , in Schwann cells) and (2) axonal transport of tumor necrosis factor α to central axons in the spinal cord. However, severe disturbance of axonal transport would cause nerve damage, but we did not find obvious damage in the sciatic nerve treated with bupivacaine microspheres; all of the treated animals recovered to full responsiveness in sensory and motor tests when the block wore off.26Because low concentration of bupivacaine was shown to suppress inflammatory responses of immune cells and extracellular signal–regulated kinase activation in spinal neurons,51,52we tried to avoid the systemic effects of bupivacaine by using microspheres. The lack of systemic effects is supported by our data showing that application of bupivacaine microspheres to the contralateral side of the sciatic nerve does not change spinal p38 activation on the other side (fig. 6).

Abnormal peripheral activity is elicited by immediate injury discharge during axonal damage and sustained by ectopic activity in both injured and noninjured fibers.19–25Application of bupivacaine to the sciatic nerve in the SNI model should block spontaneous activity originated from both injured fibers in common peroneal and tibial nerves and intact fibers in sural nerves. However, p38 activation was mainly found in the medical dorsal horn, which is innervated by the injured tibial and peroneal nerve (fig. 1). Therefore, spontaneous activity in the injured axons should play a major role in spinal p38 activation. Although application of bupivacaine microspheres at the sciatic nerve should block spontaneous activity originated at the distal site (e.g. , neuroma) of the block, we do not know whether this block can also suppress spontaneous activity at the proximal site (e.g. , dorsal root ganglion soma and dorsal root).22Therefore, the role of proximal activity in p38 and microglia activation remains to be investigated.

In summary, this study has provided further evidence that p38 activation (p-p38 induction) in spinal microglia is a useful marker for microglia activation in the spinal cord. Importantly, this marker is directly correlated with chronic pain sensitization. In contrast, pain states may not be associated with the changes of the commonly used microglial surface marker CD11b (OX-42).5,9,53Our data also suggest that peripheral spontaneous activity is important for the induction but not the maintenance of p38 activation in spinal microglia after nerve injury. Therefore, afferent barrage after the nerve injury may trigger neuropathic pain by activating p38 in spinal microglia.