The FASEB Journal express article 10.1096/fj.01-0971fje. Published online March 12, 2002. A novel control mechanism based on GDNF modulation of somatostatin release from sensory neurones Marzia Malcangio*, Stephen Getting‡, John Grist*, Joanna R. Cunningham†, Elizabeth J. Bradbury*, Peter Charbel Issa*, Isobel J. Lever*, Sophie Pezet*, and Mauro Perretti‡ *Neuroscience Research Centre, Guy’s, King’s and St. Thomas’ School of Biomedical Sciences, King’s College London, London, United Kingdom; †Vision and Ophthalmology Research Group, Rayne Institute, St Thomas' Hospital, Lambeth Palace Road, London, United Kingdom; and ‡The William Harvey Research Institute, Charterhouse Square, London, United Kingdom Corresponding author: Marzia Malcangio, Sensory Function, Centre for Neuroscience, Hodgkin Building, KCL, Guy’s Campus, London Bridge, London SE1 1UL U.K. E-mail: Marzia.Malcangio@kcl.ac.uk ABSTRACT Small-diameter sensory neurones found in the rat dorsal root ganglia (DRG) include cells sensitive to glial cell line-derived neurotrophic factor (GDNF), which express the inhibitory peptide somatostatin (SOM). Here we addressed the functional relationship between GDNF and sensory neurone-derived SOM. Topical application of GDNF through the rat isolated dorsal horn of the spinal cord promoted activity-induced release of SOM from central terminals of sensory neurones. Once released by sensory neurones, SOM is known to act, at least in part, by opposing the action of Substance P (SP) in neurogenic inflammation. Therefore, we evaluated GDNF ability to modulate two well-documented effects of peripherally and centrally administered SP. Local application of GDNF in the mouse air pouch reduced SP-induced leukocyte migration. This effect of GDNF was mimicked by the SOM analog octreotide (OCT) and required intact SOM neuronal pools. Intrathecal injection of GDNF activated rat lumbar dorsal horn neurones and inhibited intrathecal SP-induced thermal hypersensitivity. This effect of GDNF was reversed by the SOM antagonist c-SOM and mimicked by OCT. In conclusion we propose GDNF regulation of neuronal SOM release as a novel mechanism that, if explored, may lead to new therapeutic strategies based on local release of somatostatin. Key words: Substance P • evoked release • spinal cord • thermal hypersensitivity • inflammation G lial cell line-derived neurotrophic factor (GDNF) belongs to the transforming growth factor β (TGF-β) superfamily. Its biological actions are brought about by binding to GFRα-1 or α-2 (GDNF family of receptors α) and subsequent activation of the signaling receptor RET, a tyrosine kinase receptor (1–3). Beneficial effects of GDNF have been reported in models for Parkinson’s disease and neuropathic pain (4, 5). In the sensory system of adult rats, the GDNF receptor components α-1 and α-2 and RET are found in the dorsal root ganglia (DRG), in small- as well as large-diameter cell bodies of primary afferent neurones (6). Some small cells expressing the receptor components for GDNF also express the inhibitory tetradecapeptide somatostatin (SOM) (6–8). Primary afferent neurones send their axons centrally in the dorsal horn of the spinal cord and peripherally to the skin and most internal tissues where they serve afferent and efferent functions, respectively (9). Neuronal SOM released in the periphery has been suggested to act as an endogenous anti-inflammatory peptide that opposes the actions of pro-inflammatory peptides, such as substance P (SP) in models of neurogenic inflammation (10–14). The effect of pro-inflammatory peptides and other mediators predominates during active phases of inflammation, but promotion of SOM release from activated sensory neurones may play a role during the phase of resolution. Consequently, a potential role for SOM in treating major features of inflammation, such as pain and swelling, has been proposed (13–15). Accordingly, SOM and its stable analog octreotide (OCT) are delivered intrathecally in humans to treat pain that does not respond well to conventional therapies (16, 17), and SOM synthetic analogs exert anti-inflammatory and analgesic effects in rats (18, 19). We have shown recently that prolonged intrathecal administration of GDNF increased SOM expression in DRG and promoted release of the peptide in the rat isolated dorsal horn (20). In this study, we assessed first whether GDNF would act locally and acutely to influence SOM release in the dorsal horn. Then, we tested the effect of single GDNF treatment in animal models in which SOM had been shown to play anti-inflammatory roles. MATERIALS AND METHODS In vivo methods Intrathecal catheter implantation in rats Adult male Wistar rats (Charles River, Margate, Kent, UK, 250 g body weight) were used. All procedures were in accordance with U.K. Home Office regulations. Rats were anaesthetized with sodium pentobarbitone (45 mg/kg i.p.), and a midline incision was made over the mid-thoracic spinal cord. Muscles were separated from bone by blunt-dissection, and a small laminaectomy was made at the sixth or seventh thoracic vertebra. A cannula was inserted under the dura mater, such that the tip rested over the lumbar enlargement. The opposite end was externalized at the top of the head. The muscles and the skin were closed in anatomical layers. All animals recovered without incident. Each animal was used at least 3 days after surgery for a maximum of four times with at least 4 days between intrathecal treatments. Thermal threshold in rats At a minimum of 3 days after surgery, rats were allocated randomly to different experimental groups and hind-paw thermal thresholds were measured for 2 consecutive days, before the day of the experiment. Each rat was placed in a clear plastic compartment of a three-compartment box with a glass floor and allowed for acclimate for 10–20 min. We elicited a paw-flick response with a high-intensity infrared beam focused on the plantar surface of left and right hind paws (21). We monitored the time that elapsed to reflex removal of the hind paw from the beam (paw withdrawal latency, PWL). PWL was defined as the mean of three measurements. On the day of experiment, baseline PWLs were assessed and then rats received intrathecal injection of 10 µl of either vehicle (buffer acetate, 0.07 M, pH 5 (Sigma, Gillingham, Dorset, UK), containing in mM: NaCl, 128.5; KCl 3; MgCl2 0.8; CaCl2, 1.15, and 1 mg/ml rat serum albumin) or drug under examination. This was followed by a 10 µl artificial cerebralspinal fluid (CSF) flush (Harvard, Boston, MA). We then assessed PWLs at various time intervals. Inflamed air pouch in mice Male Swiss Albino mice (20–22 g body weight) were purchased from Bantin & Kingman (Hull, Humberside, UK) and maintained on a standard chow pellet diet with tap water ad libitum, with a 12:00 h light/dark cycle. Animals were used 3–4 days after the arrival. Animal work was performed according to Home Office regulations, UK. Air (2.5 ml s.c.) was injected on day 0 and day 3 to create a dorsal air pouch. On day 6, mice were treated locally with either carboxymethyl cellulose (CMC, 0.5% in phosphate buffered saline [PBS], 0.5 ml injected) alone or containing Sar9-SP (10 µg/7.3 nmol) (22). In other groups, carrageenin (1%) or zymosan A (1 mg) were injected in a total volume of 0.5 ml PBS (23). The somatostatin analog OCT or GDNF (single dose) was injected into the air pouch with the inflammatory agents on day 6. In another set of experiments, GDNF was also injected once a day (days 6 and 7), directly into the pouch (injection around 10:00 am) in a volume of 100 µl. On day 8, Sar9-SP (10 µg in 0.5 ml CMC) was injected with or without GDNF. The animals were then killed by carbon dioxyde exposure 6 h after carrageenin injection and 4 h after Sar9-SP and zymosan injections. The air-pouches were lavaged with 2 ml PBS supplemented with 3 mM EDTA and 25 U/ml heparin. Lavage fluids were then centrifuged and supernatants frozen at –20ºC until required for biochemical determinations. The pellet was resuspended in 2 ml of lavage fluid, an aliquot was stained in Turks solution, and differential cell counts were performed by using a Neubauer haematocytometer. In some cases, we extracted spinal cords at the end of the experiment and determined SOM content as described below. In other cases, mice were treated with cysteamine or GDNF in the absence of inflammatory stimuli, 4 h before spinal cord collection. In vitro and ex vivo methods Adult rat dorsal horn slice isolation and mounting in vitro Horizontal dorsal horn slices (400-µm thick), with or without dorsal roots attached, were obtained from the lumbar spinal cord of adult rats as previously described (24). One slice was obtained from each rat. Slices were mounted in the central compartment of three-compartment chambers and superfused continuously at room temperature (1 ml/min) with oxygenated (95% O2 and 5% CO2) Krebs’ solution (in mM: NaCl, 118; KCl, 4; MgSO4x 7H2O, 1.2; KH2PO4 1.2; NaHCO3 25; CaCl2 2.5, and glucose 11). After a 1-h equilibration period, we substituted Krebs’ solution by modified Krebs’ solution containing bovine serum albumin (0.1 %, BSA), protease inhibitors (bacitracin 20 µg/ml, 10 µg/ml, phoshoramidon, 1 µM, captopril 100 µM) and antioxidant (dithiothreitol 6 µM) (Sigma). Fractions of the superfusates (8 ml) were collected in icecooled tubes in acetic acid (0.1 M). The dorsal roots were placed in the two lateral compartments on bipolar platinum electrodes and were covered in mineral oil to avoid dehydration (Aldrich, Milwaukee, WI). The lateral compartments were separated from the central one by a leak-proof partition of high-vacuum grease (Dow Corning Corp., Midland, MI). Neuropeptide and glutamate release We examined the release of SOM, SP, and glutamate in the isolated dorsal horn preparation after electrical stimulation of the dorsal roots (25). We collected 8-min samples of superfusates in the following order: three fractions to measure basal outflow of transmitters, one fraction during electrical stimulation, and three fractions to measure return to transmitter basal levels. The dorsal roots were stimulated electrically at high-threshold fiber strength (square pulses of 0.5 ms duration and 20 V) for 8 min at 10 Hz (4,800 pulses) (26). Somatostatin extraction from mouse spinal cord Spinal cords were removed by hydraulic extrusion, blotted, weighed, and then frozen in liquid nitrogen and stored at –80°C. To extract SOM, frozen cords were boiled for 20 min in 1 ml of glacial acetic acid. Samples were then homogenized for 20 s, centrifuged (10, 195 g for 20 min), and the supernatant was retained. The supernatants were partially purified by using Sep-Pak C18 reverse-phase silica gel cartridges (see below). Purified samples were assayed for SOM content by ELISA. Recovery of SOM standard spiked into a spinal cord sample after Sep-Pak processing was 68%. Processing of collected samples To quantify SOM, SP, and glutamate levels in the superfusates or cell-free supernatants, we desalted and partially purified the samples by using Sep-Pak C18 reverse-phase silica gel cartridges (Waters Associates, Watford, UK) (20, 24–26). The cartridges were conditioned with acetonitrile (100%; HPLC grade; BDH Chemicals, Poole, UK) and trifluoroacetic acid (0.1%, TFA, HPLC grade, BDH). We then loaded the samples into the columns and eluted the peptide by using acetonitrile/TFA (80/20) solution. We dried the eluates by evaporation under nitrogen (recovery not less than 85% for SP and 67% for SOM). ELISA for somatostatin Dried samples were reconstituted in 200 µl assay buffer, and 50 µl aliquots were assayed by ELISA (0.6-30 fmol/50 µl/well, Peninsula Laboratories Inc., San Carlos, CA) as previously described (20). Radioimmunoassay for substance P Dried samples were reconstituted in 300 µl of phosphate buffer, and 100 µl aliquots were assayed by radioimmunoassay (RIA, 1-100 fmol/tube) by using scintillation proximity assay (Amersham, Buckinghamshire, UK) as described previously (24-26). HPLC for glutamate The remaining 100 µl of reconstituted samples that were assayed for peptide content were used for glutamate content determination by HPLC (25). Intrathecal injection of GDNF and immunostaining for phospho-ERK in the spinal cord Six adult rats were anaesthetized with urethane (1.25 g/kg, i.p.). Five minutes after injection of either vehicle (see above) or GDNF, rats were perfused transcardially with 100 ml saline, followed by 500 ml paraformaldehyde/picric acid (15%) in phosphate buffer 0.1 M (PB). The cord was removed, postfixed in the same fixative for 24 h, and cryoprotected in 30% sucrose in PB. Coronal sections (30 µm) were cut and collected into PB containing 0.9% NaCl (PBS). Freefloating sections were incubated overnight at room temperature in PBS containing 0.3% Triton X-100 (PBST) with polyclonal anti-phospho ERK 1/2 (1:200, New England Biolabs, Hitchin, Hertfordshire, UK). After being washed, sections were incubated with the secondary antibody (anti-rabbit Cy3-labeled, Jackson Labs, West Grove, PA, 1:500) for 2 h at room temperature. After several washes, sections were mounted on gelatin-coated slides and all were coverslipped in Vectashield (Vector Laboratories, Burlingame, CA). We counted the number of phospho-ERK positive neurones in 1 out of 18 sections at spinal cord level L2-L5 (total of 15–20 sections per animal). Results are expressed as the mean number of phospho-ERK positive neurones per section. Statistical analysis The majority of data were analyzed by ANOVA, followed by Tukey. One-tail Student’s t-test was used for comparisons between two experimental groups. A threshold probability value <0.05 was taken as significant. Chemicals The following chemicals were used: Sar9-substance P, cysteamine, λ-carrageenin, and zymosan A (Sigma-Aldrich, Poole, UK); cyclo[7-aminoheptanoyl-Phe-D-Trp-Lys-Thr(Bzl)] (c-SOM) (Bachem, St. Helens Merseyside, UK); and octreotide (Sandostatin, Sandoz, 0.5 mg/ml); and human recombinant GDNF was a generous gift from Amgen (Thousand Oak, CA). RESULTS Effect of topical application of GDNF on release of sensory neuron transmitters in the dorsal horn We have shown recently that prolonged in vivo treatment with GDNF increased SOM expression in the DRG and SOM release in the isolated dorsal horn—with dorsal root attached in preparation as measured ex vivo (20). In this study, we confirmed initially that electrical stimulation of the dorsal roots did not increase SOM content in the superfusate fraction (32–40 min interval, Fig. 1A) compared with basal outflow fractions (8–32 min interval, Fig. 1A). Then we observed that addition of GDNF in the medium superfusing the dorsal horn preparation significantly increased SOM content in the fraction collected during electrical stimulation (32–40 min interval, Fig. 1A) compared with basal outflow contents (8–24 time interval, Fig. 1A). The effect of GDNF was fast, reversible, and exclusively on evoked release because SOM basal outflow was not modified by this trophic factor (24–32 min interval, Fig. 1A). GDNF-induced release of SOM was dose-dependent, with a calculated EC50 of 15.8 ng/ml (500 pM) (Fig. 1B). To rule out the possibility that the lack of SOM release in controls was due to failure in recruiting high-threshold fibers, we measured the release of SP that is induced by a wide range of high-threshold fiber firing patterns (25). Under the same experimental conditions used for SOM, SP content in control slices was increased significantly in superfusate fractions collected during electrical stimulation of the dorsal roots (32–40 min interval, Fig. 1C) compared with basal outflow fractions (8–32 min interval, Fig. 1C). However, GDNF did not modify the pattern of SP release and left the basal outflow and electrical-evoked release of the peptide unchanged (Fig. 1C). In addition, GDNF did not modify the pattern of glutamate release, which failed to show an increase over basal outflow following electrical stimulation in either control or GDNF-treated cord slices (Fig. 1D). Glutamate is contained in both low- and high-threshold fibers, and, as previously observed, the lack of detection of glutamate release under our experimental conditions may be explained by fast uptake of the amino acid by both neurones and glia (25). To characterize the mechanism(s) by which GDNF promoted SOM release from activated sensory neurones up to measurable levels, we first evaluated whether the effect of GDNF was calcium-dependent. Figure 2A shows that GDNF-induced activity-evoked SOM release was absent in calcium-free medium. Then, because the functional receptor for GDNF, RET, is a tyrosine kinase receptor, we tested whether a tyrosine kinase inhibitor (K-252a) would modulate the effect of GDNF in the dorsal horn slice preparation. Superfusion of K-252a (100 nM), although not modifying SOM basal outflow (24–32 min interval, Fig. 2B), abrogated GDNFinduced release of SOM during electrical stimulation of the dorsal roots (32–40 min interval, Fig. 2B). K-252a alone did not alter the pattern of SOM release (Fig. 2B). Next, we tested for potential consequences of GDNF-induced sensory neurone activation. Neuronal SOM has been suggested to act as a functional antagonist of the pro-inflammatory and pro-nociceptive peptide SP released by a different population of sensory neurones (10, 11, 27). Therefore, we determined GDNF ability to modulate SP-induced PMN migration in the mouse air-pouch model of inflammation and SP-induced hypersensitivity to a noxious thermal stimulus in the rat. Effect of GDNF and OCT on Sar9-SP-induced PMN migration in mouse inflamed air-pouch model In models of neurogenic inflammation in the rat, SP-induced neurotrophil chemotaxis/elicitation have been shown to be susceptible to SOM inhibition (11, 12). In this study, we evaluated whether the SOM stable analog OCT and GDNF exerted any effect on neutrophil migration induced by several inflammogens in the mouse air-pouch model. Injections of carrageenin, zymosan, or Sar9-SP into the mouse air pouch induced intense neutrophil migration (Fig. 3A). OCT injected into the air pouch at the same time as the inflammogens significantly reduced cell migration induced by zymosan and Sar9-SP but not by carrageenin (Fig. 3A). GDNF modified neither carrageenin- nor zymosan-induced leukocyte influx into the air pouches (Fig. 3B). However, GDNF, either given as single (3 µg) or repeated (3× 1 µg) doses, markedly inhibited Sar9-SP-induced PMN migration (Fig. 3C). The effect of GDNF was prevented in mice previously treated with systemic cysteamine (Fig. 3C) to deplete SOM pools selectively (28). At the selected dose of 100 mg/kg, cysteamine produced neither lethargy and hypotension nor did it alter neutrophil migration (Fig. 3C). To verify whether cysteamine had indeed depleted neuronal SOM pools, we extruded spinal cords from cysteamine-treated mice and control mice and extracted and quantified the peptide. Systemic cysteamine treatment (100 mg/kg s.c.) dramatically reduced spinal cord SOM contents measured 4 h after injection. SOM-LI values in controls that received water (5 ml/kg s.c., n=4) were 33.7±3.7 pg/mg wet tissue. This level dropped to 4.7±0.4 pg/mg wet tissue after cysteamine administration (n=3 mice, P<0.001, Student’s t-test). A single injection of GDNF into the air pouch of naïve mice at a dose effective for inhibiting Sar9-SP-induced neutrophil migration (3 µg/mouse) significantly reduced SOM content measured in spinal cords 4 h after injection. SOM-LI values were as follows: 35.8±1.5 pg/mg wet tissue in controls that received PBS into the air pouch (n=4) and 15.8 ±5.0 pg/mg wet tissue in mice that received GDNF (n=4; P<0.05 Student’s t-test). Our attempt to measure changes in SOM content in air-pouch exudates failed as the peptide levels fell below detection limit of our assay. These data show that GDNF inhibited leukocyte recruitment induced by SP and required intact SOM neuronal pools. The peripheral terminal of sensory neurones is likely to be both the source for SOM and the site of action for GDNF. Effect of GDNF and OCT on intrathecal Sar9-SP-induced thermal hypersensitivity In models of neurogenic inflammation, the damage to peripheral tissue causes increased sensitivity to noxious stimuli (hyperalgesia) that is initiated by enhanced release of glutamate and SP from central terminals of nociceptive fibers in the dorsal horn (29). Accordingly, in this study intrathecal injection of Sar9-SP (10 µg/rat) reduced thermal threshold in normal rats (thermal hyperalgesia) (Fig. 4A). However, sensory neurons can also release the inhibitory peptide SOM following noxious stimulation (30). In our study, single intrathecal injections of SOM stable analog OCT (3.2 µg/rat) or the SOM antagonist, c-SOM (0.3 µg/rat) did not alter rat threshold to noxious thermal stimuli (Fig. 4B and C). Similarly, GDNF (12 µg/rat) did not modify rat thermal threshold (Fig. 4D). However, when OCT was administered 10 min before Sar9-SP (Fig. 5B), it completely prevented Sar9-SP-induced thermal hyperalgesia observed 15 min post-injection in vehicle-treated rats (Fig. 5A). OCT inhibition of thermal hyperalgesia was still observed at 30- and 45 min-intervals after Sar9-SP injection (not shown). Similarly, GDNF injected 10 min before Sar9-SP (Fig. 5C) or 10 min after Sar9-SP (Fig. 5D) significantly inhibited or abolished the hyperalgesia that developed 15 min after Sar9-SP injection, respectively. However, unlike OCT, GDNF did not prevent Sar9-SP-induced hyperalgesia at 30- and 45-min intervals after injection (data not shown). When the SOM antagonist c-SOM was co-administered with GDNF 10 min after Sar9-SP, normal thermal hyperalgesia developed at 15-min intervals after Sar9-SP (Fig. 5E), as in vehicletreated rats (Fig. 5A), and persisted up to 45 min (not shown). These data show that activation or blockage of SOM receptors in the dorsal horn did not change rat threshold to noxious thermal stimuli. However, centrally administered OCT and GDNF inhibited the hypersensitivity to thermal stimuli that followed intrathecal injection of SP. The effect of GDNF was likely to be mediated by SOM, as was prevented by a SOM receptor antagonist. To prove that GDNF had reached its proposed site of action in the dorsal horn, we evaluated whether a single intrathecal injection of GDNF induced activation of extracellular signalregulated kinase, ERK. This is because after binding to its receptor components, GDNF activates mitogen-activated protein kinase (MAPK) (31, 32). In the dorsal horn of the spinal cord, MAPK activation results in an increase in the phosphorylated form of ERK, which can be visualized immunohistochemically (33). We observed that phospho-ERK labeling was undetectable in the dorsal horn of control animals (Fig. 6A), whereas GDNF injection induced phosphorylation of ERK in neuron-shaped cells in laminae I and II (thin arrows in Fig. 6B and C). Some of them had the typical shape of flattened marginal cells located in lamina I. We observed a second type of diffuse labeling in round-shaped cells and processes in lamina II (thick arrows in Fig. 6B and C). These processes could be either lamina II neurones and /or terminals of primary afferent fibers. DISCUSSION The present study provides experimental evidence that supports the possibility that GDNF can exert fast modulatory effects towards the population of sensory neurons that express the receptor components for this trophic factor and contain the inhibitory non-opioid neuropeptide SOM. This mechanism underlies novel functional effects of GDNF in experimental inflammation and central hyperalgesia. We suggest that GDNF modulated sensory neuron synaptic activity at both central and peripheral terminal levels based on the following evidence: i) acute application of GDNF through the rat dorsal horn isolated in vitro, enhanced up to measurable levels the release of SOM evoked by electrical stimulation of the dorsal roots. This effect of GDNF was dependent on external calcium ions and on tyrosine kinase activation in the dorsal horn. ii) Single, as well as repeated, application of GDNF selectively reduced neutrophil migration induced by the chemoattractant SP in the mouse air-pouch model of inflammation. The effect of GDNF was mimicked by OCT and absent when neuronal SOM pools were depleted. iii) Single intrathecal injection of GDNF in the rat lumbar spinal cord had restored to normal values the reduced threshold to thermal stimuli that developed after spinal delivery of SP. The effect of GDNF was reversed by the SOM antagonist c-SOM and mimicked by OCT. At the time when GDNF was effective in inhibiting SP-induced thermal hypersensitivity (5 min post injection of the trophic factor), the extracellular signal-regulated kinase ERK was activated in the dorsal horn. We have shown previously that prolonged intrathecal treatment with GDNF increased the number of SOM-containing sensory neurones in the DRG and the activity-induced release of SOM in the dorsal horn isolated ex-vivo (20). These actionsof GDNF resembled the regulatory effect of NGF on the TrkA expressing population of sensory neurones that contain the peptides SP and calcitonin gene-related peptide (CGRP) (34). However, we report here that, in contrast to NGF, which does not have acute modulatory action on SP and CGRP release in the dorsal horn (34), topical application of GDNF promotes SOM release by a mechanism requiring calcium ion entry and activation of tyrosine kinase receptor (possibly RET). This is the first report to our knowledge that suggests GDNF may act as a fast neuromodulator, mimicking trophic factors such as BDNF and NT-3 (35). As previously argued (20), the lack of SOM release following electrical stimulation of the dorsal roots in control slices could not be attributed to a failure of electrical stimulation to recruit high-threshold fibers that contain the peptide. This is because, under the same conditions, SP that is also contained in high-threshold fibers (27) was released significantly. However, primary afferent fiber contribution to the total dorsal horn SOM content is much lower than SP content (36). It is possible that SOM release under normal conditions in the dorsal horn is too low to be detected and that GDNF increases activity-induced release up to measurable levels. In this study we have begun to determine the intracellular mechanism by which GDNF promoted the release of SOM. A potential pathway could involve activation of GDNF receptor-coupled tyrosine kinase activity that can lead to MAP-kinase activation. This possibility is based on two observations. First, the effect of GDNF on SOM release in the dorsal horn was prevented by K-252a, a tyrosine kinase inhibitor. Second, intrathecal injection of GDNF to anaesthetized rats induced ERK phosphorylation in the superficial laminae, where primary afferent fibers containing SOM terminate (7, 27). In addition, the calcium-dependency of GDNF effect on SOM release suggests that GDNF promoted a substantial buildup in cytosolic calcium ion concentration, which is known to be required for release of peptides from large, dense core vesicle (37). It remains to be established which SOM neuronal pool in the dorsal horn is targeted by GDNF treatment. These pools include sensory afferent fibers, intrinsic neurones, and some brain stem axons (8). However, it is likely that GDNF acted on primary afferents, because GFRα-1 immunoreactivity is found on axon terminals of sensory neurones in dorsal horn lamina II (38), some of which contain SOM (6) and also express RET (20). Moreover, because there is lack of evidence for RET mRNA expression in these neurones, it is unlikely that GDNF promoted release of SOM from interneurones (38). Therefore, our data so far indicate that GDNF promoted activity-induced SOM release in the dorsal horn following binding to its receptor components, likely GFR-α1 and RET (6, 20) on primary afferent fibers. GDNF ability to modulate sensory neurone activity prompts us to examine whether GDNF treatment would result in functional changes in the whole animal that could be attributed to facilitation of local SOM release from peripheral and/or central sensory neuron endings. Peripherally administered GDNF and OCT After tissue injury, the SOM released by peripheral terminals of sensory neurones acts as antiinflammatory peptide that directly opposes the actions of pro-inflammatory peptides such as SP in neurogenic inflammation (9, 10, 13, 14). SP activates NK1 receptors on the venular endothelium to increase microvascular permeability and promote plasma extravasation (9). SP also stimulates leukocyte adhesion to the vessel wall and their emigration into the inflamed tissue, an effect again brought about by NK1 receptors (39). SOM and OCT have been shown to reduce neutrophil elicitation in carrageenin-induced inflammation in the rat air-pouch model in vivo (12) and neutrophil chemotaxis promoted by SP in vitro (11). The latter cell type, as well as human mononuclear cells, expresses the receptors for both SP and SOM (40, 41). In this study, we confirmed that OCT locally applied into the mouse air pouch inhibited SP- and zymosaninduced neutrophil extravasation. However, we could not reproduce the OCT-inhibition of carrageenin-induced PMN migration reported in the rat air-pouch model (12). Differences in species or in protocols used could explain this discrepancy. Like OCT, GDNF that was injected locally into the mouse air pouch inhibited SP-induced PMN migration. In this experimental condition, however, GDNF did not alter the effect of either zymosan or carrageenin. The observation that depletion of neuronal SOM pools by cysteamine prevented the effect of GDNF strongly suggests that GDNF is counteracting the effect of SP by activating the endogenous SOM system. However, cysteamine treatment did not modify the extent of PMN migration promoted by the NK1 receptor activator, ruling out a role for endogenous SOM within the time frame of our observation. The peripheral terminal of sensory neurones is likely to be both the source for SOM and the site of action for GDNF as they express its receptor components (6, 20). In addition, the observation that SOM content was decreased in the spinal cord of mice that received GDNF directly into the air pouch supports the possibility that peripherally acting GDNF might have promoted release of SOM from central terminals of primary afferent fibers. For GDNF to act centrally it could either distribute systemically after local injection or is transported retrogradely. We favor neither of these two possibilities. First, in analogy to other proteins (e.g., IL-1ß, M. P. unpublished observations), GDNF is likely to remain localized into the air pouch; second, GDNF would have moved about 0.5 cm along peripheral nerve by fast axonal retrograde transport. Taken together, the data obtained in the mouse air-pouch model suggest that peripherally administered GDNF opposes the pro-inflammatory effect of SP likely by inducing release of SOM from primary afferent fibers. Centrally administered GDNF and OCT It is now well established that peripheral tissue inflammation causes increased sensitivity to noxious stimuli (hyperalgesia) consequent to a facilitated state of spinal processing (central sensitization). This is initiated by the release of excitatory transmitters, such as glutamate and SP from central terminals of afferent C-fibers and persists beyond activation of glutamate and NK1 receptors (29). Accordingly, intrathecal injections of SP or N-methyl-D-aspartic acid (NMDA) (a selective agonist to NMDA receptors for glutamate) induce thermal hyperalgesia in rats (29), and NMDA as well as NK1 receptor antagonists are antihyperalgesics in models of inflammation (42, 43). However, sensory neurones also release SOM following noxious stimulation of peripheral nerves (30) and in neurogenic inflammation (19). In the dorsal horn exogenous SOM mainly depresses the firing of dorsal horn neurones activated by noxious stimulation (44, 45). Accordingly, SOM applied intrathecally has an antihyperalgesic effect in animal models of inflammation (46). SOM receptors have been identified in the dorsal horn matching the distribution of SOM (47–49), but SOM receptors and NK1 receptors are expressed by different population of neurones (49). In this study we tested the hypothesis that activation of SOM receptors in the dorsal horn would oppose the action of SP on NK1 receptors mimicking the relationship between these two systems delineated in the periphery. Therefore, we first showed that intrathecal OCT inhibited SP-induced thermal hyperalgesia, which suggests that activation of SOM receptors in the dorsal horn can functionally antagonize the excitatory effect of SP that results in thermal hypersensitivity. Then we proved the hypothesis that, similarly to OCT, GDNF would counteract SP-induced thermal hyperalgesia and that the effect of GDNF would be prevented by a SOM receptor antagonist. However, the effect of OCT after single intrathecal injection was long-lasting, whereas the effect of GDNF lasted only a few minutes. Several reasons could account for this short-lasting effect of the trophic factor. A plausible one is that if GDNF were antihyperalgesic by releasing SOM, the duration of the effect of the endogenous peptide would be limited by its enzymatic degradation (50, 51). However, we have not examined the time course of the effect of SOM in our model, but because OCT shows greater metabolic stability than SOM (51), its effect would last longer. Our observation that the extracellular signal-regulated kinase, ERK, was activated in superficial laminae as early as 5 min after the intrathecal GDNF injection indicated that this trophic factor had reached the dorsal horn at the time it was effective in inhibiting the effect of SP. This is because after binding to its receptor components, GDNF activates MAPK (31, 32). Finally, the short duration of the antihyperalgesic effect of GDNF could not be attributed to its degradation because a single injection of GDNF has been shown to have long-lasting effects in models of neuronal degeneration (4). The data obtained in the model of SP-induced thermal hypersensitivity indicate that in the dorsal horn OCT and GDNF functionally antagonized the effect of SP by mechanisms requiring direct and indirect activation of SOM receptors, respectively. A final comment is due to the modality of GDNF action. In all models used in this study, GDNF appeared to modulate SOM release exclusively from activated sensory neurones. In the dorsal horn preparation, sensory neurones were stimulated electrically. In the air-pouch model, sensory nerves were likely to be activated during the underlying tissue remodeling that occurs 6 days after the initial injection of air (52). In the plantar test for monitoring thermal sensitivity, sensory neurones were activated by a thermal noxious stimulus. All together, these arguments support a modulatory role for GDNF in the complex scenario characteristic of neurogenic inflammation. We propose that GDNF modulation of activity-induced release of SOM is a novel mechanism that deserves exploration as a potential new therapeutic strategy based on local release of SOM to control two major features of inflammation, pain and leukocyte recruitment. ACKNOWLEDGMENTS This work was supported by a Wellcome Trust Research Fellowship to M. M. and an Arthritis Research Campaign Fellowship to M. P. (grant P0567). S. P. was supported by the Guy’s & St. Thomas Charitable Foundation; E.J.B and J.G., by the Wellcome Trust. We thank Stephen B. McMahon for his cooperation over facilities and Amgen (Thousand Oak, CA) for its generous supply of GDNF. REFERENCES 1. Treanor, J., Goodman, L., Desauvage, F., Stone, D., Poulsen, K., Beck, C., Gray, C., Armanini, M., Pollock, R., Hefti, F., Phillips, H., Goddard, A., Moore, M., Bujbello, A., Davis, A., Asai, N., Takahashi, M., Vandlen, R., Henderson, C. and Rosenthal, A. (1996) Characterization of a multicomponent receptor for GDNF. Nature 382, 80-83 2. Trupp, M., Arenas, E., Fainzilber, M., Nilsson, A.S., Sielber, B.A., Grogoriou, M., Kilkenny, C., Salazar-Grueso, E., Pachnis, V. and Arumae, U. (1996) Functional receptor for GDNF encoded by the c-ret proto-oncogene. Nature 381, 785-789 3. Ibáñez, C.F. (1998) Emerging themes in structural biology of neurotrophic factors. Trends Neurosci 21, 438-444 4. Olson, L. (1997) The coming age of the GDNF family and its receptors: gene delivery in a rat Parkinson model may have clinical implications. Trends Neurosci 20, 277-280 5. Boucher, T.J., Okuse, K., Bennett, D.L.H., Munson, J.B., Wood, J.N. and McMahon, S.B. (2000) Potent analgesic effects of GDNF in neuropathic pain states. Science 290, 124-127 6. Bennett, D.L.H., Michael G.J., Ramachandran, N., Munson, J.B., Averill. S., Yan. Q., McMahon, S.B. and Priestley, J.V. (1998) A distinct subgroup of small DRG cells express GDNF receptor components and GDNF is protective for these neurons after nerve injury. J Neurosci 18, 3059-3072 7. Alvarez, F.J. and Priestley, J.V. (1990) Anatomy of somatostatin-immunoreactive fibres and cell bodies in the rat trigeminal subnucleus caudalis. Neuroscience 38, 343-357 8. Proudlock, F. , Spike, R.C. and Todd, A.J. (1993) Immunocytochemical study of somatostatin, neurotensin, GABA, and glycine in rat spinal dorsal horn. J Comp Neurol 372, 289-297 9. Holzer, P. (1998) Neurogenic vasodilation and plasma leakage in the skin. Gen Pharmacol 30, 5-11 10. Green, P.G., Basbaum, A.I. and Levine, J.D. (1992) Sensory neuropeptide interactions in the production of plasma extravasation in the rat. Neuroscience 50, 745-749 11. Kolasinski, S.L., Haines, K.A., Siegel, E.L., Cronstein, B.N. and Abramson, S.B. (1992) Neuropeptides and inflammation a somatostatin analog as selective antagonist of neurotrophil activation by substance P. Arthritis Rheum 35, 369-375 12. Karalis, K., Mastorakos, G., Chrousos, G.P. and Tolis, G. (1994) Somatostatin analogues suppress the inflammatory reaction in vivo. J Clin Invest 93, 2000-2006 13. Heppelmann, B. and Pawlak, M. (1997) Inhibitory effect of somatostatin on the mechanosensitivity of articular affrents in normal and inflamed knee joints of the rat. Pain 73, 377-382 14. Szolcsányi, J,, Helyes, Z., Oroszi, G., Németh, J. and Pintér, E. (1998) Release of somatostatin and its role in the mediation of the anti-inflammatory effect induced by antidromic stimulation of sensory fibres of rat sciatic nerve. Br J Pharmacol 123, 936-942 15. Carlton, S.M., Du, J., Davidson, E., Zhou, S. and Coggeshall, R.E. (2001) Somatostatin receptors on peripheral primary afferent terminals: inhibition of sensitized nociceptors. Pain 90, 233-244 16. Chrubasik, J., Meynadier, J., Blond, S., Scherpereel, P., Ackerman, E., Weistock, M., Bonath, K., Cramer, H. and Wunsch, E. (1984) Somatostatin, a potent analgesic. Lancet 2, 1208-1209 17. Paice, J.A., Penn, R.D. and Kroin, J.S. (1996) Intrathecal octreotide for releif of intractable nonmalignant pain: 5-year experience with two cases. Neurosurgery 38, 203-207 18. Helyes,Z., Pintér,E, Németh, J., Kéri, G., Thán, M., Oroski G., Horváth A and Szolcsányi J. (2000) Anti-inflammatory effect of synthetic somatostatin analogues in the rat. Br J Pharmacol 134, 1571-1579 19. Helyes,Z., Thán, M., Oroski G., Pintér,E, Németh, J., Kéri, G. and Szolcsányi J. (2001) Antinociceptive effect induced by somatostatin released from sensory nerve terminals and by synthetic somatostatin analogues in the rat. Neurosci Lett 278, 185-188 20. Charbel-Issa, P., Lever, I.J., Michael, G.J., Bradbury, E.J. and Malcangio, M. (2001) Intrathecally delivered glial cell line-derived neurotrophic factor produces electrically evoked release of somatostatin in the dorsal horn of the spinal cord. J Neurochem 78, 221-229 21. Hargreaves, K., Dubner, R., Brown, F., Flores, C. and Joris, J. (1988) A new and sensitive method for measuring thermal nociception in cutaneous hyperalgesia. Pain 32, 77-88 22. Perretti, M., Ahluwalia, A., Flower, R.J. and Manzini, S. (1993) Endogenous tachykinins play a role in interleukin-1-induced neutrophil accumulation: involvement of NK-1 receptors. Immunology 80, 73-77 23. Perretti, M., Ahluwalia, A., Harris, J.G., Harris, H.J., Wheller, S.K. and Flower, R.J. (1996) Acute inflammatory response in the mouse: exacerbation by immunoneutralization of lipocortin 1. Br J Pharmacol 117, 1145-1154 24. Malcangio, M. and Bowery, N.G. (1993) Gamma-aminobutyric acidB, but not γ-aminobutyric acidA receptor activation, inhibits electrically evoked substance P-like immunoreactivity release from the rat spinal cord in vitro. J Pharmacol Exp Ther 266, 1490-1496 25. Lever, I.J., Bradbury, E.J., Cunningham, J.R., Adelson, D.W., Jones, M.G., McMahon, S,B., Marvizón, J.C.G. and Malcangio, M. (2001) Brain-derived neurotrophic factor is released in the dorsal horn by distinctive patterns of afferent fiber stimulation. J Neurosci 21, 4469-4477 26. Malcangio, M., Ramer, M.S., Jones, M.G. and McMahon, S.B. (2000) Abnormal substance P release from the spinal cord following injury to primary sensory neurons. Eur J Neurosci 12, 397-399 27. Hökfelt, T., Elde, R., Johansson, O., Luft, R., Nilsson, G. and Arimura, A. (1976) Immunohistochemical evidence for separate populations of somatostatin-containing and substance P-containing primary afferent neurons in the rat. Neuroscience 1, 131-136 28. Ceccatelli, S., Hökfelt, T., Hallman, H., Nylander, I., Terenius, L., Elde, R. and Broenstein, M. (1987) Immunohistochemical analysis of the effect of cysteamine on somatostatin-like immunoreactivity in the rat central nervous system. Peptides 8, 371-384 29. Malmberg, A.B. and Yaksh, T.L. (1992) Hyperalgesia mediated by spinal glutamate and substance P receptor blocked by spinal cyclooxygenase inhibition. Science 257, 1276-1279 30. Morton, C.R., Hutchison, W.D. and Hendry, I.A. (1988) Release of immunoreactive somatostatin in the spinal dorsal horn of the cat. Neuropeptides 12, 189-197 31. Worby, C.A., Vega, Q.C., Zhao, Y., Chao, H.H.J., Seasholt, A.F. and Dixon, J.E. (1996) Glial cell line-derived neurotrophic factor signals through the RET receptor and activates mitogen-activated protein kinase. J Biol Chem 271, 23619-23622 32. Nicole, O., Ali, C., Docagne, F., Plawinski, L., MacKenzie, E.T., Vivien, D. and Buisson, A. (2001) Neuroprotection medaited by glial cell line-derived neurotrophic factor:involvement of a reduction of NMDA-induced calcium influx by the mitogen-activated protein kinase pathway. J Neurosci 21, 3024-3033 33. Ji, R-R., Baba, H., Brenner, G.J. and Woolf, C.J. (1999) Nociceptive-specific activation of ERK in spinal neurons contributes to pain hypersensitivity. Nat Neurosci 2, 1114-1119 34. Malcangio, M., Garrett, N.E. and Tomlinson, D.R. (1997) Nerve growth treatment increases stimulus-evoked release of sensory peptides in the rat spinal cord. Eur J Neurosci 9, 11011104 35. Schinder, A.F. and Poo, M. (2000) The neurotrophin hypothesis for synaptic plasticity. Trends Neurosci 23, 639-645 36. Tuchscherer, M.M. and Seybold, V.S. (1989) A quantitative study of the coexistence of peptides in varicosities within the superficial laminae of the dorsal horn of the rat spinal cord. J Neurosci 9, 195-205 37. Verhage, M., McMahon, H.T., Ghijsen, W.E.J.M., Boomsma, G.S., Wiegant, V.M. and Nicholls, D.G. (1991) Differential release of amino acids neuropeptides and catecholamines from isolated nerve terminals. Neuron 6, 517-524 38. Widenfalk, J., Lundsromer, K., Jubran, M., Brene, S. and Olson, L. (2001) Neurotrophic fcators and receptors in the immature and adult spinal cord after mechanical injury or kainic acid. J Neurosci 21, 3457-3475 39. Mancuso, F., Flower, R.J. and Perretti, M. (1995) Leukocyte transmigration, but not rolling or adhesion, is selectively inhibited by dexamethasone in the hamster post-capillary venule. Involvement of endogenous lipocortin 1. J Immunol 155, 377-386 40. Hartung, H., Wolters. J. and Toyka, K. (1986) Substance P: binding properties and studies on cellular responses in guinea-pig macrophages. J Immunol 136, 3856-3863 41. Payan, D.G., McGillis, J.P. and Organist, M.L. (1986) Binding characteristics and affinity labeling of protein constituent of the human IM-9 lymphoblast receptor for substance P. J Biol Chem 261, 14321-14329 42. Ren, K., Iadarola, M.J. and Dubner, R. (1996) An isobolographic analysis of the effects of Nmethyl-D-asparatte and NK1 tachykinin receptor antagonists on inflammatory hyperalgesia in the rat. Br J Pharmacol 117, 196-202 43. Sluka, K.A., Milton, M.A., Willis, W.D. and Westlund, K.N. (1997) Differential roles of neurokinin- and neurokinin 2 receptors in the development and maintenance of heat hyperlagesia induced by acute inflammation. Br J Pharmacol 120, 1263-1273 44. Murase, K., Nedeljkov, V. and Randić, M. (1982) The actions of neuropeptides on dorsal horn neurons in the rat spinal cord slice preparation: an intracellular study. Brain Res 234, 170-176 45. Sandkühler, J., Fu, Q.G. and Helmchen, C. (1990) Spinal somatostatin superfusion in vivo affects activity of cat nociceptive dorsal horn neurones: comparison with spinal morphine. Neuroscience 34, 565-576 46. Chapman, V. and Dickenson, A.H. (1992) The effects of sandostatin and somatostatin on nociceptive transmission in the dorsal horn of the rat spinal cord. Neuropeptides 23, 147-152 47. Schindler, M., Holloway, S., Hathway, G., Woolf, C.J., Humphrey, P.P.A. and Emson, P.C. (1998) Identification of somatostain sst2a receptor expressing neurons in central regions involved in nociception. Brain Res 798, 25-35 48. Schulz, S., Schreff, M., Schmidt, H., Handel, M., Przewlocki, R. and Hollt, H. (1998) Immunocytochemical localization of somatostatin receptor sst2A in the rat spinal cord and dorsal root ganglia. Eur J Neurosci 10, 3700-3708 49. Todd, A.J., Spike, R.C. and Polgar, E. (1998) A quantitative study of neurons which express neurokinin-1 or somatostatin sst2a receptor in rat spinal dorsal horn. Neuroscience 85, 459473 50. Griffiths, E.C., Jeffcoate, S.L. and Holland, D.T. (1977) Inactivation of somatostatin by peptidases in different areas of the rat brain. Acta Endocrinol 85, 1-10 51. Patel, Y.C. (1999) Somatostatin and its receptor family. Frontiers Neuroendocrinol 20, 157198 52. Edwards, J.C., Sedgwick, A.D. and Willoughby, D.A. (1981) The formation of a structure with the features of synovial lining by subcutaneous injection of air: an in vivo tissue culture system. J Pathol 134, 147-156 Received January 7, 2002; accepted January 25, 2002. Fig. 1 Figure 1. Acute superfusion of GDNF promoted activity-induced release of SOM but did not modify SP or glutamate release. GDNF (100 ng/ml) was superfused for 16 min (horizontal open bar) during collection of the fraction before stimulation and the stimulated fraction (horizontal black bar, 20 V, 0.5 ms, 1–10 Hz). Data (means ± SE) on the release of SOM-like immunoreactivity (SOM-LI) (A), SP-LI (C) or glutamate (D) were obtained from at least 4 rat preparations in each group. Basal outflow values for glutamate were 97.7±1.5 pmol/ml. *P< 0.05 versus basal outflow values, # P<0.05 versus analog fraction in controls, ANOVA followed by Tukey test. B) shows GDNF dose-response curve on SOM-LI release. Values express peptide content in the stimulated fraction after subtraction of basal outflow values (32.2±1.1.fmol/8 ml, n=15 slices) by using GraphPad Prism™ software. Fig. 2 Figure 2. Modulation of GDNF effect on electrically evoked SOM release. A) GDNF induced the release of SOM-LI in the presence of Ca2+ions (■), but not when slices were superfused with Ca2+-free Krebs’ solution containing 10 mM EDTA (□). Electrical stimulation (black horizontal bar, 20 V, 0.5 ms, 10 Hz) was applied after collection of two basal outflow fractions. Points represent means ± SE of four preparations. *P<0.05 versus analog fractions collected in the presence of Ca2+ (ANOVA followed by Tukey test). B) GDNF effect on electrically evoked release of SOM was inhibited by a tyrosine antagonist. GDNF (30 ng/ml, ■) was present for 16 min (horizontal white bar) in the fraction before and during stimulation (horizontal black bar). The tyrosine antagonists K 252a (100 nM, horizontal white bar) was present in the fraction before and during stimulation either alone (
) or with GDNF (○). Points are means ± SE of at least four preparations for each group. *P<0.05 ANOVA followed by Tukey test. Fig. 3 Figure 3. Effect of OCT and GDNF on carrageenin-, zymosan-, and Sar9-SP-induced neutrophil migration in inflamed air-pouches in the mouse. A) Carrageenin (CARRA 0.1 %, 0.5 ml), zymosan A (ZYM , 1 mg/0.5 ml) and Sar9-Substance P (Sar9-SP, 10 µg/0.5 ml) were injected into the air pouches at time 0 and granulocytes (PMN) counted in lavage fluids obtained 6h (CARRA) or 4h (ZYM and Sar9-SP) after injections. OCT (25 µg/mouse) was injected with each inflammogens. *P<0.05 ANOVA followed by Tukey test (n=6-7). B) GDNF (3 µg/mouse) was co-injected into the air-pouches with CARRA or ZYM to groups of 12 or 8 mice, respectively. C) GDNF (single dose, 3 µg or repeated administration 1 µg for 3 times) was co-injected with Sar9-SP into the air pouches of 8 mice. PBS or GDNF (3 µg) and Sar9-SP were co-injected into the air-pouches of mice pretreated 4 h earlier with either water (5 ml/kg s.c.) or cysteamine (100 mg/kg s.c.). The number of mice was 6 in each group. *P<0.05 ANOVA followed by Tukey test. Fig. 4 Figure 4. Effect of Sar9-SP, OCT, c-SOM, and GDNF on rat hind paw sensitivity to thermal stimulation. A) Sar9SP (10 µg/rat, n=11) reduced paw withdrawal latency (PWL), whereas injection of vehicle (10 µl/rat, n=8) did not change thresholds from baseline values measured before treatment (time point 0). Neither OCT (3.2 µg/rat, n=6; B), the SOM antagonist c-SOM (0.3 µg/rat, n=6; C) or GDNF (12 µg/rat, n=6, D) altered rat thermal threshold from baseline values. The number of rats injected with vehicle was 5 in (B, C, and D). Fig. 5 Figure 5. Effect of OCT and GDNF on Sar9SP-induced thermal hypersensitivity. Basal thermal threshold latencies (time point 0) were reduced 15 min after intrathecal injection of Sar9SP (10 µg/rat, n=7, A). OCT (3.2 µg/rat) injected 10 min before Sar9SP prevented SP-induced reduction in PWL (n=5; B). GDNF (12 µg/rat) injected 10 min before Sar9SP inhibited the effect of Sar9SP (n=5; C). GDNF (12 µg/rat) injected 10 min after Sar9SP blocked Sar9SP-reduction in thermal threshold latencies (n=5; D). The SOM receptor antagonist, c-SOM (0.3 µg/rat), and GDNF (12 µg/rat) coinjected 10 min after Sar9SP did not inhibit Sar9SP-induced reduction in PWL (n=6; E) that developed as in controls (A). *P<0.05, one-tail Student’s t-test versus PWL values measured before drug injections (time point 0). Fig. 6 Figure 6. Intrathecal GDNF injection induced phosphorylation of ERK in superficial laminae of the dorsal horn. A–C) Immunofluorescent labeling showing phosphorylation of ERK in the dorsal horn of control rats (10 µl vehicle/rat) (A) or GDNF (12 µg/rat) (B and C). C is a higher magnification of (B). D) Quantification of the mean number of phospho-ERK positive neurons (n=3 preparations per group, *P<0.05, Student’s t-test).