p38 Mitogen–activated Protein Kinase Activation by Nerve Growth Factor in Primary Sensory Neurons Upregulates μ-Opioid Receptors to Enhance Opioid Responsiveness Toward Better Pain Control

• Inflammation sensitizes nociceptive peripheral nerves, in part, by increasing nerve growth factor release—but nerve growth factor also increases their expression of μ-opioid receptors.

• In rats, peripheral inflammation increased μ-opioid receptor expression in nociceptive peripheral nerves by nerve growth factor activation of p38-mitogen–activated protein kinase.

DAILY clinical practice shows that opioids are the most effective and widely used drugs in the treatment of severe pain (e.g ., acute postoperative pain, chronic cancer-related pain).1,2Most opioids are administered systemically, passing the blood-brain barrier to activate corresponding opioid receptors in pain-related brain areas.3However, opioids are also effective on the central terminals of peripheral sensory neurons via  intrathecal or epidural delivery.4,5Likewise, they are effective at the peripheral terminals of sensory neurons via  local or topical application.6 

It has been shown that the μ-opioid receptor (MOR) is expressed mainly in cell bodies of nociceptive C and AΔ neurons within dorsal root ganglia (DRG).7–9In addition, there is growing evidence that inflammatory pain increases MOR expression in these neurons,10–12resulting in enhanced axonal transport of receptors from the DRG toward nerve terminals of sensory neurons.13,14This up-regulation in MOR expression occurs exclusively in ipsilateral DRG, suggesting that local (but not systemic) inflammatory mediators regulate opioid receptor expression of peripheral sensory neurons.13 

It is well established that nerve growth factor (NGF), interleukin-1, and other inflammatory mediators can affect the phenotype of nociceptive neurons by increased expression of neuropeptides, such as substance P and calcitonin gene–related peptide (CGRP),15–18as well as receptors and ion channels, such as transient receptor potential channel, vanilloid subfamily member 1, acid-sensing ion channel receptors, and purinoreceptors.19–21This inflammatory-associated neuroplasticity is the consequence of a combination of activity-dependent changes and specific signal-transduction pathways, which phosphorylate cellular proteins, activate transcription factors, and finally alter gene expression.22Several studies have investigated the signal-transduction pathways that are mainly involved in this enhanced expression.19,21,23However, little is known about the signaling pathways responsible for the regulation of sensory neuron opioid receptor expression during painful conditions, such as inflammatory pain.

Mitogen-activated protein kinases (MAPK) transduce extracellular stimuli into intracellular responses and, thereby, induce changes in transcription and posttranslational modifications of target proteins.24p38 MAPK and extracellular signal–regulated kinase (ERK) 1/2 are activated in primary sensory neurons after peripheral inflammation.25,26A recent study demonstrated that the transient receptor potential ion channel, vanilloid subfamily member 1, is regulated by NGF-induced activation of the ERK pathway in DRG neurons in vitro .27Moreover, Ji et al .19showed that p38-MAPK activation in the DRG is responsible for the inflammation-induced transient receptor potential ion channel, vanilloid subfamily member 1, up-regulation, and hyperalgesia. In addition, p38 MAPK is involved in substance P and N -methyl-d-aspartate receptor–induced pain sensitivity as well as the regulation of cyclooxygenase-2 up-regulation and prostaglandin E2 release in the spinal cord.28 

Because NGF has been implicated in the up-regulation of the number and efficacy of sensory neuron MOR, resulting in enhanced analgesic effects of opioids,13we set out to identify the NGF-dependent signaling pathways. Therefore, we examined whether: (1) exogenous as well as endogenous NGF, raised by Freund's complete adjuvant (FCA), contributes to enhanced antinociception of a full (fentanyl) or partial (buprenorphine) opioid agonist; (2) NGF-dependent increases in MOR binding, protein, and immunoreactive cells in peripheral sensory neurons are abolished by p38 MAPK and/or ERK-1/2 inhibition; (3) the number of immunoreactive MOR neuronal cells colocalizing with activated p-p38 MAPK within DRG neurons is increased after intraplantar NGF but return to baseline levels after concomitant intrathecal inhibition of p38 and/or ERK-1/2 MAPK; (4) local NGF enhances the anterograde axonal transport of MOR along the sciatic nerve via  activation of p38 MAPK and/or ERK 1/2; (5) NGF-induced enhanced antinociceptive effects of local opioids are abolished by intrathecal inhibition of p38 and/or ERK-1/2 MAPK.

Experiments were conducted in male Wistar rats (200–250 g) housed in cages lined with ground corncob bedding. Rats were kept in climate- and light-controlled rooms (22 ± 0.5°C; relative humidity, 60–65%; 12-h light/dark cycle) with standard rodent food pellets and water ad libitum . Experiments were performed in accordance with The Development of Science-based Guidelines for Laboratory Animal Care ∥and were approved by the local animal care committee (Landesamt für Arbeitsschutz, Gesundheitsschutz und technische Sicherheit, LAGetSi, Berlin, Germany). All efforts were made to minimize the number of animals used and their suffering.

Intrathecal catheters were used according to previously published protocols.29In brief, animals were anesthetized with isoflurane in oxygen via  nose cone. A longitudinal skin incision was made in the lumbar region directly above the spinous processes of the L4–L6 vertebrae. The needle, through which the catheter was set up, was inserted at a 30° angle between the L5 and L6 vertebra. The catheter (PE10, 15 cm, 0.61-mm OD; Portex Ltd, Hythe, Kent, United Kingdom) was carefully advanced while rotating it between the thumb and forefinger. This rotation facilitated the penetration through the intervertebral space and dura. Dura penetration was observed by movement of the tail and hind limbs. The catheter was then carefully pushed upward 1–2 cm into the intrathecal space. The needle was then carefully removed before the catheter was fixed with glue to the tissue and secured with two additional sutures. Another skin incision was made at the dorsal neck, where the catheter was tunneled under the skin and pulled out at the neck before suture. Intrathecal catheter location was confirmed after surgery by administration of 10 μl lidocaine, 2%. Lidocaine caused 10–15 min of bilateral hind limb paresis whereas no deficits occurred in animals injected with sterile phosphate buffered saline (PBS). Animals were allowed to recover for at least 2 days before experiments. Only animals exhibiting no motor deficits were used for further testing.

FCA, a water-in-oil emulsion with killed mycobacteria (Calbiochem, San Diego, CA) was injected (150 μl) intraplantarly. Affinity-purified polyclonal rabbit anti-NGF antiserum (CEDARLANE, Burlington, Ontario, Canada) was dissolved in PBS at 8 μg/100 μl. The first and intraplantar injection was administered 30 min before FCA.13According to a previous protocol,30the β subunit of NGF (β-NGF; R&D Systems, Inc., Minneapolis, MN) was reconstituted in sterile PBS at 4 μg/100 μl and injected intraplantarly once daily for 4 consecutive days. According to manufacturer instructions, the p38-MAPK inhibitor 4-(4-fluorophenyl)-2-(4-methylsulfonylphenyl)-5-(4-pyridyl)-1H-imidazole (SB203580) and the MEK/ERK-1/2 inhibitor 2′-amino-3′-methoxyflavone (PD98059; Calbiochem) were dissolved in dimethyl sulfoxide diluted with sterile distilled water to a final concentration of 2%. Twice before removal of DRG, intrathecal boluses (1 μg/10 μl) were injected for each drug.17,31The first injection was delivered 30 min before intraplantar NGF. For 96-h experiments, mini-pumps (1 μg/1 μl; delivery rate 0.5 μl/h; ALZET Osmotic Pumps, Cupertino, CA) were filled with drugs or vehicle, implanted subcutaneously, and attached to anchored intrathecal catheters.29,32Fentanyl (N -(1-phenethyl-4-piperidyl) propionanilide citrate (1:1) or buprenorphine hydrochloride (21-(cyclopropyl-7α-[(S)-1-hydroxy-1,2,2-trimethylpropyl]-6,14-endo-ethano-6,7,8,14-tetrahydrooripavine hydrochloride) (Sigma-Aldrich, Taufkirchen, Germany) were dissolved in isotonic saline at different concentrations before application.

Rats were divided into nine groups with six to eight animals per group. The first group was treated with intraplantar FCA for 96 h. The second group of rats received intraplantar FCA plus intraplantar affinity-purified anti-NGF antiserum 30 min before FCA and then every 12 h. The third group, which served as a control, was injected intraplantar with isotonic saline. The fourth group received intraplantar NGF once daily for 1 or 4 days (24 or 96 h). The fifth group received intraplantar NGF and intrathecal SB203580. The sixth group received intraplantar NGF and intrathecal PD98059.

The dose of each MAPK inhibitor was selected based on previously published experimental protocols.19,31For visualization of anterograde axonally transported MOR, rats from the last three groups also received a tight ligation of the sciatic nerve at the midthigh level 24 h after the onset of treatment under isoflurane anesthesia and were sacrificed 24 h later.13,14Taken together, all experimental groups were tested at 24 h for changes at the level of the DRG, at 48 h for changes at the level of the sciatic nerve, and at 96 h (4 days) for behavioral changes at the level of the hind paw after local NGF treatment to examine consecutive changes along the peripheral sensory neuron.

At 96 h after treatment initiation, paw pressure threshold (PPT) was assessed using an algesiometer (modified Randall-Selitto test; Ugo Basile, Comerio, Italy) before (baseline) and after intraplantar injection of two different MOR opioid agonists: the full opioid agonist fentanyl (0–1.2 μg/100 μl) and the partial opioid agonist buprenorphine (1–5 μg/100 μl). Animals were gently restrained under paper wadding. Using a wedge-shaped, blunt piston, incremental pressure was applied on the dorsal surface of the hind paw by means of an automated gauge. The pressure required to elicit paw withdrawal, the PPT (250-g cutoff), was determined by averaging three consecutive trials separated by 10 s. The sequence of left and right paws was alternated among animals to preclude order effects. The experimenter was blind to group assignment.

At 24 h after treatment initiation, animals from the different groups were sacrificed. Ipsilateral L3–L5 DRGs were immediately removed, frozen in liquid nitrogen, and stored at −80°C. Western blot analysis was performed as previously described by Ji and Rupp.33In brief, samples were homogenized in boiling sodium dodecyl sulfate polyacrylamide sample buffer (100 mM Tris-hydrochloric acid, 2% sodium dodecyl sulfate, 20% glycerol). Protein concentration was measured using bicinchoninic acid assay (Pierce BCA Protein Assay; Pierce Biotechnology, Inc., Rockford, IL). 2-Mercaptoethanol and bromphenol blue were added before loading. Extracts were separated using 10% (MOR) or 12% (p38, p-p38, ERK 1/2) sodium dodecyl sulfate polyacrylamide gel electrophoresis 60-μg protein per lane and transferred onto nitrocellulose filters. Filters were blocked in 5% milk for 1 h and incubated with the following antibodies at 4°C overnight: rabbit polyclonal MOR antiserum (1:1,000 in 5% bicinchoninic acid; Gramsch Laboratories, Schwabhausen, Germany), rabbit polyclonal antiphospho-p44/p42 MAPK (p-ERK 1/2, 1:500), mouse monoclonal anti-p38 MAPK (1:1,000), and antiphospho-p38 MAPK (p-p38 MAPK, 1:500; Cell Signaling Technology, Inc., Beverly, MA). After incubation at room temperature for 2 h with secondary antibodies, specifically peroxidase-conjugated goat anti-rabbit (1:5,000; Amersham Pharmacia Biotech Europe, Freiburg, Germany) and rabbit anti-mouse (1:5,000; Abcam, Cambridge, MA), reactive bands were visualized in enhanced chemiluminescence solutions (Amersham Pharmacia Biotech Europe) for 1 min and immediately exposed to autoradiograph film for 5–20 min. Finally, blots were incubated in stripping buffer (62.6 mM Tris-hydrochloric acid [pH 6.7], 2% sodium dodecyl sulfate, 100 mM mercaptoethanol) at 56°C for 30 min and reprobed with monoclonal mouse anti–β-actin antibody (1:10,000; Sigma-Aldrich) as a loading control. These experiments were performed in duplicate and carried out three times.

Western blot bands of MOR, p38, and p-p38 MAPK were then quantified using an open-source image software program (Java Image; ImageJ#).34The area and density of pixels within the threshold values representing immunoreactivity were measured. The integrated density (i.e ., the product of the area and mean of gray value) was calculated. Integrated immunodensities of controls and treated groups were compared and analyzed.

At 24 h after treatment initiation, animals from the different groups were killed by an overdose of isoflurane. The lumbar (L3–L5) DRG ipsilateral to the NGF-injected hind paw were removed. Membranes were obtained from DRGs as previously described.35In brief, the tissue was placed immediately on ice in cold assay buffer (50 mM Tris-hydrochloric acid, 1 mM EGTA, 5 mM MgCl²[pH 7.4]). Tissue was homogenized (POLYTRON; Kinematica, Littau, Switzerland) and centrifuged at 48,000g  at 4°C for 20 min. The pellet was resuspended in assay buffer after 10-min incubation at 37°C to remove endogenous ligands. The homogenate was centrifuged at 42,000g  and resuspended in assay buffer. Membranes were aliquoted and stored at −80°C. DRG membranes were diluted in assay buffer. Specific binding of [³H]DAMGO (DAMGO =[D-Ala2, N-MePhe4, Gly-ol]-enkephalin) was performed by incubating 100 μg membrane protein with 2 nM [³H]DAMGO in the presence or absence of 10 μM unlabeled naloxone to determine nonspecific binding. Membranes were incubated at 22°C for 1 h in assay buffer. Reactions were terminated by rapid filtration under vacuum through Whatman grade GF/B glass fiber filters followed by four washes with cold buffer (50 mM Tris-hydrochloric acid [pH 7.4]). Bound radioactivity was determined by liquid scintillation spectrophotometry (PerkinElmer, Inc., Rodgau, Germany) at 60% counter efficiency after overnight extraction of the filters in 3 ml scintillation fluid.35All experiments were performed in duplicate and carried out three times.

For visualization of anterograde axonally transported MOR with CGRP along the sciatic nerve, immunohistochemistry was done 48 h after treatment initiation and 24 h after nerve ligation. However, for visualization of MOR with p-p38 MAPK in DRG, immunohistochemistry was done at 24 h after treatment initiation. Rats were deeply anesthetized with isoflurane and transcardially perfused with 100 ml PBS (pH 7.4) and 500 ml paraformaldehyde, 4% w/v, in phosphate buffer (pH 7.4). After perfusion, the DRG and the ligated sciatic nerve (0.5 cm) were removed from treated and control animals, postfixed in the same fixatives for 90 min, and then cryoprotected at 4°C overnight in PBS with 10% sucrose. The DRG or sciatic nerve was embedded in Tissue-Tek OCT compound (Bayer Corporation, Pittsburgh, PA), frozen, and cut into 8-μm sections.

DRG or sciatic nerve–mounted tissue sections were incubated with the following rabbit primary antibodies: rabbit polyclonal MOR antibody (1:1,000; Gramsch Laboratories) alone or in combination with mouse monoclonal p38-MAPK antibody (1:1,000), mouse monoclonal p-p38-MAPK antibody (Cell Signaling Technology, Inc.), or guinea pig polyclonal antibody against CGRP (1:1,000; Peninsula Laboratories, Belmont, CA). Tissue sections were washed with PBS and incubated with the appropriate secondary antibodies: Texas red conjugated goat anti-rabbit antibody alone or in combination with fluorescein isothiocyanate conjugated donkey anti-mouse or anti–guinea pig antibody. Finally, tissues were washed in PBS, mounted on VECTASHIELD (Vector Laboratories, Inc., Burlingame, CA), and viewed under a laser-scanning microscope (LSM 510; Carl Zeiss MicroImaging, Göttingen, Germany).

To demonstrate staining specificity, the following controls were included as detailed elsewhere: (1) preabsorption of antibody against MOR with a synthetic peptide for MOR (Gramsch Laboratories) at 4°C for 24 h; (2) omission of either the primary antisera, the secondary antibodies, or the avidin-biotin complex; (3) omission of either the first or second primary antibody and either the first or second secondary antibody.13 

The method of quantification for DRG staining has been detailed elsewhere.9,13,36In brief, every fourth section of DRG that was serially cut at 10 μm for each animal (n = 5) was stained. For neuron counting, only those immunostained neurons containing a distinct nucleus were counted for a total of a minimum of 400 neurons using the microscope (×40). After MOR immunostaining, the total number of immunoreactive MOR neurons was counted by an observer blinded to the experimental protocol. This number was divided by the total number of neurons in each DRG section, and the percentage was calculated. The immunoreactive MOR neurons, which were also immunoreactive for p-p38 MAPK, were counted. The proportion was calculated as a percentage of the total number of immunoreactive MOR neurons in each DRG section and represented as percentages for vehicle (NGF v  s . NGF/p38-MAPK inhibitor–treated animals). Data were obtained from four sections of each DRG and five rats per group.

For quantification of MOR immunoreactivity at the ligated area of the sciatic nerve, images of red (Texas red) immunofluorescence were obtained using a laser scanning microscope laser-scanning microscope (Carl Zeiss MicroImaging). To quantify changes in immunodensities, Modular Image-Analysis software was applied (version 2.5 [service pack 2]; Carl Zeiss MicroImaging).13,37The settings of the confocal microscope were established using a control section and kept unchanged for all subsequent image acquisitions. Six to eight images were sampled per animal. Images were thresholded to exclude background fluorescence. They were gated to include intensity measurements only from positively stained cells. For images analysis, a standardized box was positioned over the proximal part of the ligated sciatic nerve of all groups to determine the mean product of the area (μm²) and the mean intensity of pixels within the threshold value as well as to calculate the integrated optical intensity (product of area and mean intensity). Five rats per group were used for analysis.

The results of behavioral experiments are given as mean ± SD or SEM of the percentage of maximum possible effect (% MPE) according to the following formula: (PPT^{postinjection}− PPT^basal)/(250^cut-off− PPT^basal). All statistics (two-tailed testing) were performed using STAT software (version 2.03; SPSS Science Software, Chicago, IL). Data were compared by a one-way or two-way analysis of variance (ANOVA) if the normality test passed (Kolmogorov-Smirnov test). Otherwise, the Kruskal-Wallis test was used. Post hoc  multiple pair-wise comparisons were performed by the Newman-Keuls test. Data of two groups were compared using the unpaired Student t  test. For the analysis of behavioral dose-dependent effects, a linear regression ANOVA was applied. A P  value of less than 0.05 was considered statistically significant.

In PBS-treated Wistar rats, intraplantar injection of the full MOR agonist fentanyl (0.5–1.0 μg) elicited a significant and dose-dependent increase in PPT consistent with the development of dose-dependent antinociception (fig. 1A; P < 0.05, linear regression one-way ANOVA, n = 6–8). In rats with FCA hind paw inflammation, the dose-response curve elicited by 0.5–1.0 μg fentanyl significantly shifted to the left, toward enhanced potency. Immunoneutralization of endogenous NGF in FCA-inflamed hindpaws reversed this leftward shift in potency (P < 0.05, two-way ANOVA, Newman-Keuls, n = 6–8). In contrast, pretreatment of naive rats with exogenous NGF resulted in a similar leftward shift toward enhanced potency of intraplantar fentanyl (fig. 1B). In PBS- treated Wistar rats, intraplantar injection of the partial opioid agonist buprenorphine (1–5 μg) did not increase PPT consistent with a lack of peripheral efficacy (fig. 1C), although higher doses revealed systemic antinociceptive effects (data not shown). In rats with FCA hind paw inflammation, administration of buprenorphine (1–5 μg) significantly and dose-dependently increased PPT (P < 0.05, linear regression one-way ANOVA, n = 6–8). However, the maximum peripheral antinociceptive effects of buprenorphine were lower than those elicited by fentanyl. Immunoneutralization of NGF in FCA-treated rat hindpaws significantly abolished the antinociceptive efficacy elicited by buprenorphine (P < 0.05, two-way ANOVA, Newman-Keuls, n = 6–8). In contrast, pretreatment of naive rats with exogenous NGF resulted in a similar degree of enhanced efficacy of intraplantar fentanyl (fig. 1D).

MOR-specific [³H]DAMGO binding sites were evaluated in the DRGs of rats treated with intraplantar NGF with or without intrathecal p38-MAPK inhibitor SB203580 or MEK/ERK-1/2 inhibitor PD98059 compared with vehicle. Twenty-four hours after NGF treatment, the number of [³H]DAMGO binding sites in DRG was significantly increased compared with that of vehicle-treated rats (fig. 2A; P < 0.05; one-way ANOVA, Newman-Keuls, n = 6). This increase in MOR-specific [³H]DAMGO binding sites induced by NGF was significantly reversed after intrathecal p38-MAPK inhibitor SB203580 (P < 0.05; one-way ANOVA, Newman-Keuls, n = 6), but not after intrathecal MEK/ERK-1/2 inhibitor PD98059.

Analysis of MOR immunofluorescence staining showed a significant up-regulation in the percentage of immunoreactive MOR cells in DRG (40.2 ± 7.3%,) 24 h after local administration of NGF in comparison with vehicle-treated animals (32.1 ± 6.2%). Intrathecal application of the p38-MAPK inhibitor SB203580 (27.7 ± 4.6%) significantly reduced the percentage of immunoreactive MOR cells in comparison with NGF-treated animals (fig. 2B; P < 0.05; one-way ANOVA, Newman-Keuls, n = 20).

Western blot analysis of DRG tissue extracts of control animals revealed immunoreactive MOR protein bands at the expected molecular weight of 50 kDa (fig. 2C). Twenty-four hours after intraplantar NGF treatment, the integrated optical density of immunoreactive MOR protein bands was increased by 67% compared with that of vehicle-treated animals (P < 0.05; one-way ANOVA, Newman-Keuls, n = 6). This increase was again prevented after intrathecal administration of p38-MAPK inhibitor SB203580 (P < 0.05; one-way ANOVA, Newman-Keuls, n = 6).

In parallel, at 24 h after NGF treatment, the integrated optical density of immunoreactive phosphorylated p-p38–MAPK protein bands at the expected molecular weight of 43 kDa increased eightfold compared with that of vehicle-treated animals (fig. 2D; P < 0.05; one-way ANOVA, Newman-Keuls, n = 6). This increase was reduced 5.6-fold by intrathecal administration of p38-MAPK inhibitor SB203580 (P < 0.05; one-way ANOVA, Newman-Keuls, n = 6).

In contrast, the integrated optical density of total immunoreactive p38-MAPK protein bands at the expected molecular weight of 38 kDa increased slightly by 0.4-fold in NGF-treated animals compared with vehicle-treated animals and did not change significantly after intrathecal treatment with p38-MAPK inhibitor SB203580 (fig. 2E; P > 0.05, one-way ANOVA, Newman-Keuls, n = 6).

Double confocal immunofluorescence microscopy of MOR and phosphorylated p-p38 MAPK in DRG of control animals showed immunoreactive MOR neurons (51.1 ± 2.2%) colocalizing with phosphorylated p-p38 MAPK, although some neurons expressed MOR (48.9 ± 2.2%) alone (figs. 3A, B, and C). Twenty-four hours after intraplantar NGF treatment, the percentage of immunoreactive MOR neurons colocalizing with phosphorylated p-p38 MAPK increased significantly (83.2 ± 3.5%) compared with that of vehicle-treated animals (figs. 3A, B, C, D, E, and F; P < 0.05, one-way ANOVA, Newman-Keuls, n = 20). This NGF-induced elevation in MOR colocalization with p-p38 MAPK in DRG neurons (55 ± 2.9%) was significantly attenuated after intrathecal p38-MAPK inhibitor SB203580 (figs. 3G, H, and I; P < 0.05, one-way ANOVA, Newman-Keuls, n = 20).

Double confocal immunofluorescence microscopy examined the axonal MOR immunoreactivity proximal to a sciatic nerve ligature in rats treated with intraplantar vehicle (figs. 4A, B, and C), NGF (figs. 4D, E, and F), or NGF with intrathecal p38-MAPK inhibitor SB20358 (figs. 4G, H, and I). In vehicle-treated rats, MOR immunoreactivity expressed in CGRP nerve fibers accumulated mainly proximal to the ligature of the sciatic nerve. Compared with vehicle-treated animals, intraplantar NGF treatment induced a 44% increase in axonally transported MOR immunoreactivity on CGRP nerve fibers proximal to the ligature (P < 0.05, one-way ANOVA, Newman-Keuls, n = 30). Intrathecal pretreatment with the p38-MAPK inhibitor SB20358 prevented this increase in axonally transported MOR of immunoreactive CGRP nerve fibers (P < 0.05, one-way ANOVA, Newman-Keuls, n = 30).

Intraplantar NGF-dependent potentiation of intraplantar fentanyl antinociception was reversed after intrathecal administration of the p38-MAPK inhibitor SB203580 (P < 0.05; two-way ANOVA, n = 6–8), but not after the MEK/ERK-1/2 inhibitor PD98059 (figs. 5A and B). NGF-dependent enhanced efficacy of intraplantar buprenorphine antinociception was inhibited after intrathecal administration of the p38-MAPK inhibitor SB203580 (P < 0.05; two-way ANOVA, n = 6–8), but not after the MEK/ERK-1/2 inhibitor PD98059 (figs. 5C and D). It is noteworthy that intraplantar NGF immediately decreased PPT (i.e ., hyperalgesia) when compared with vehicle-treated animals (table 1). However, this hyperalgesia was reversed by intrathecal administration of p38-MAPK inhibitor SB203580, but not MEK/ERK-1/2 inhibitor PD98059.

The major finding of this study is that exogenous and endogenously up-regulated NGF, through activation of the p38-MAPK pathway, contributes to adaptive changes of the sensory neuron opioid system toward enhanced susceptibility to local opioids. This effect is established in five ways: (1) a potentiation of dose-dependent local antinociceptive effects of the full opioid agonist fentanyl as well as an enhanced antinociceptive efficacy of the partial opioid agonist buprenorphine 96 h after increased local NGF; (2) at 24 h, NGF treatment up-regulation of MOR in DRG ipsilateral to local NGF and its attenuation by intrathecal application of a p38-MAPK, but not a MEK/ERK-1/2, inhibitor; (3) increased phosphorylation and, thus, activation of p38 MAPK and its enhanced colocalization in immunoreactive MOR neurons of DRG 24 h after local NGF treatment; (4) increased anterograde axonal transport of MOR along the sciatic nerve 48 h after local NGF and its inhibition by intrathecal p38-MAPK inhibitor; (5) reversal of the NGF-induced enhanced antinociceptive effects of local opioids by intrathecal application of a p38-MAPK, but not a MEK/ERK-1/2, inhibitor.

In the present study, we examined whether endogenously increased NGF in inflamed paws, or exogenously applied intraplantar NGF, leads to enhanced antinociceptive effects of local opioids. Intraplantar injection of low, systemically inactive doses of the full MOR agonist fentanyl elicits dose-dependent antinociceptive effects in normal rats. In rats treated with intraplantar NGF or FCA, the dose-dependent antinociceptive effects of fentanyl shift to the left, indicating a potentiation. Moreover, intraplantar anti-NGF reversed this potentiation in FCA-induced inflammatory pain, suggesting a role of endogenous NGF that is commonly known to be up-regulated during FCA inflammation.38Consistent with a required higher receptor occupancy for partial opioid agonists39,40than for full agonists (e.g ., fentanyl) and a relatively low maximal number of MOR binding sites in DRG of normal rats,41peripheral administration of buprenorphine did not show antinociceptive efficacy in untreated rats. However, after local FCA or NGF treatment, intraplantar buprenorphine elicited significant and dose-dependent antinociceptive effects. However, these effects remained small compared with those observed for intraplantar fentanyl. The potentiation in the antinociception of the full opioid agonist fentanyl and the rise in the antinociceptive efficacy of the partial opioid agonist buprenorphine strongly suggest an increase in the number of MOR. Indeed, Western blot analysis of MOR receptor protein, MOR-specific [³H]DAMGO binding sites, and the number of immunoreactive MOR cells in DRG showed a significant up-regulation after local FCA or NGF treatment. These findings are consistent with those of Mousa et al .13and other reports,42,43showing an up-regulation of MOR in NGF-overexpressing transgenic mice. These findings are also consistent with a report by Cahill et al .,44which showed that intrathecal application of NGF can rescue the reduced opioid responsiveness seen in neuropathic pain.

We extended these investigations to identify the NGF-dependent signaling pathways responsible for the up-regulation of sensory neuron opioid receptor expression and efficacy. Increased concentrations of endogenous NGF during inflammatory pain and exogenously applied NGF are known to bind to its receptors TrkA and p75.45They are then retrogradely transported as early endosomes to the cell somata of sensory neurons within the DRG.45These NGF TrkA-carrying early endosomes also colocalize the signaling proteins p38 MAPK and ERK 1/2,45which are involved in different pain conditions.46In Western blot analysis, the NGF-induced up-regulation of MOR receptor proteins was concomitant with a 5.6-fold increase in phosphorylated (i.e ., activated) p-p38 MAPK in DRG. However, total p38 MAPK increased only slightly and not at a statistically significant level. Moreover, intrathecal administration of the p38-MAPK inhibitor SB203580—but not the MEK/ERK-1/2 inhibitor PD98059—reversed NGF-induced increases in MOR receptor proteins, MOR-specific [³H]DAMGO binding sites, and the number of immunoreactive MOR cells in DRG. Therefore, our findings suggest that the p38-MAPK–signaling pathway mainly contributes to the NGF-dependent up-regulation of sensory neuron MOR. These results are in line with previous studies19that showed increased activation of p38 MAPK after local NGF. However, the increased activation of p38 MAPK may also lead to enhanced nociception—most likely via  an increased number of ion channels, such as capsaicin receptor TRPV1—in painful conditions.19,46In line with this theory, intraplantar NGF produced hyperalgesia in our study, as reflected by diminished PPT when compared with vehicle-treated animals (table 1). Hyperalgesia was attenuated after intrathecal application of the p38-MAPK inhibitor SB203580. However, our results support the notion that, in parallel to enhanced hyperalgesia through p38-MAPK activation, adaptive changes occur within sensory neurons toward an up-regulation in MOR number and efficacy—resulting in enhanced opioid susceptibility and pain control.13This conclusion was confirmed in double confocal immunofluorescence microscopy, which demonstrated an increase in the number of immunoreactive MOR neurons colocalizing with phosphorylated (i.e ., activated) p38 MAPK, an effect that was suppressed by the intrathecal p38-MAPK inhibitor SB203580. In line with other studies,19this NGF-induced increase in p38-MAPK activation was restricted to small and medium size DRG neurons, most of which were nociceptors. This result indicates that p38-MAPK activation in DRG neurons is not a universal response of neurons to stress. Instead, it is restricted to nociceptive NGF-responsive and TrkA-expressing neurons.13Together, these results suggest that NGF, through the activation of the p38-MAPK pathway, contributes to enhanced expression of MOR within primary afferent sensory neurons. It is noteworthy that intrathecal administration of the MEK/ERK-1/2 inhibitor PD98059 had no effect on NGF-induced up-regulation of sensory neuron MOR, although the literature26,47has shown that, under these conditions, sensory neuron ERK-1/2 is activated.

How might up-regulation in the number of sensory neuron MOR be mechanistically explained? One possible mechanism is that p38 MAPK increases transcription of MOR in DRG neurons. In previous work,11we observed a moderate increase in MOR messenger RNA (mRNA) 1–2 h after intraplantar FCA. However, a lack of changes in MOR mRNA does not rule out possible changes in MOR receptor proteins.10Although some studies48–50have investigated possible transcription factors that regulate MOR expression, many were performed using cell lines that endogenously express MOR. Up-regulation of MOR mRNA in human neuroblastoma SH-SY5Y cells after interleukin-6 stimulation was dependent on the transcription factors STAT1 and STAT 3.48Recent studies showed that p38 MAPK is involved in the activation of STATs49and CREB.50It is noteworthy that phosphorylation of STAT3 and CREB are increased in mouse DRGs after peripheral inflammation.51Alternatively, p38 MAPK has been shown to regulate protein expression under inflammatory conditions by mRNA stabilization. Unstable mRNAs often contain AU-rich elements in their 3′-untranslated region. (Au refers to nucleic acid bases adenin, uracil.) The characteristic motif is AUUUA. The AU-rich element–regulated mRNA stability is mediated by RNA-binding proteins, such as human antigen R. In response to proinflammatory stimuli, HuR has been shown to increase the mRNA stability of tumor necrosis factor-α, cyclooxygenase-2, and interleukin 8 in a p38-MAPK–dependent manner.52–55This p38-MAPK dependence was blocked by use of the p38-inhibitor SB203580. The 3′-untranslated region of human and mouse MOR contains several AU-rich elements56that regulate MOR expression.57,58 

Opioid receptors synthesized in DRG neurons undergo anterograde axonal transport toward peripheral nerve terminals. This process leads to enhanced density of opioid receptors on cutaneous nerve fibers.13,14,59,60Therefore, we examined whether the peripherally directed axonal transport of MOR is affected by NGF with or without intrathecal p38-MAPK inhibition. Double confocal immunofluorescence microscopy of MOR and immunoreactive CGRP neurons in the ligated sciatic nerve showed a significant increase of MOR after intraplantar NGF accumulating proximal to the ligature. This increase in axonally transported MOR was attenuated by intrathecal injection of the p38-MAPK inhibitor SB203580. These findings suggest that NGF enhanced the axonal transport of sensory neuron MOR toward the peripheral subcutaneous tissue through activation of the p38-MAPK pathway.

Finally, we examined whether NGF-induced MOR up-regulation in peripheral sensory neurons and enhanced antinociceptive effects of local opioids are mediated through the activation of the p38 or ERK-1/2 MAPK pathway. Axonal transport of MOR from the DRG toward peripheral nerve terminals requires a delay of approximately 2 days after FCA or NGF treatment.13,14Indeed, our previous work13showed a significant increase in immunoreactive MOR nerve fibers within the layers of the epidermis 96 h after intraplantar NGF, coinciding with the potentiation of opioid-mediated antinociception. Therefore, algesiometric experiments were performed 96 h after NGF treatment. Our current investigation found that, whereas the potentiation of dose-dependent antinociceptive effects of the full opioid agonist fentanyl was reversed by a rightward shift in the dose-response curve, the enhanced antinociceptive efficacy of the partial opioid agonist buprenorphine was abolished by intrathecal injection of the p38-MAPK inhibitor SB203580—but not the MEK/ERK-1/2 inhibitor PD98059. Thus, consistent with NGF-induced sensory neuron MOR up-regulation through p38-MAPK activation, p38-MAPK inhibition prevented the enhanced antinociceptive effects of local opioids.

In summary, we have shown that exogenous and endogenously up-regulated NGF, through activation of the p38-MAPK pathway, lead to adaptive changes in sensory neuron opioid receptors toward enhanced susceptibility to local opioids. After intraplantar NGF treatment, this effect occurs in three consecutive steps: increased MOR in DRG at 24 h, increased axonal MOR transport at 48 h, and increased MOR density at 96 h. Consequently, dose-dependent peripheral antinociceptive effects of locally applied full opioid agonists such as fentanyl are potentiated and those of partial opioid agonists such as buprenorphine are enhanced in efficacy which is reversed by intrathecal p38-MAPK inhibitor SB203580. This mechanism may act as a counter-regulatory response to p38-MAPK–induced painful conditions, such as inflammatory pain, to facilitate exogenously or endogenously mediated opioid antinociception.

The authors thank Claudia Spies, M.D. (Chairwoman and Professor, Department of Anaesthesiology and Intensive Care Medicine, Charité University Berlin, Berlin, Germany), for her continuous support. They also gratefully acknowledge Petra von Kwiatkowski, B.Sc. (Technician, Berlin, Germany), and Ute Oedekoven, B.Sc. (Graphic Designer, Berlin, Germany), for their technical assistance.