Prostaglandins, synthesized in the spinal cord in response to noxious stimuli, are known to facilitate nociceptive transmission, raising questions about their role in neuropathic pain. The current study tested the hypothesis that spinal nerve ligation-induced allodynia is composed of an early prostaglandin-dependent phase, the disruption of which prevents allodynia.


Male Sprague-Dawley rats, fitted with intrathecal drug delivery or microdialysis catheters, underwent left L5-L6 spinal nerve ligation or sham surgery. Paw withdrawal threshold, brush-evoked behavior, and the concentration of prostaglandin E2 (PGE2) in spinal cerebrospinal fluid ([PGE2]dialysate) were determined for up to 24 days. PGE2-evoked glutamate release from spinal slices was also determined.


Paw withdrawal threshold decreased from at least 15 g (control) to less than 4 g, beginning 1 day after ligation. Brushing the affected hind paw evoked nociceptive-like behavior and increased [PGE2]dialysate (up to 257 +/- 62% of baseline). There was no detectable change in basal [PGE2]dialysate from preligation values. The EC50 of PGE2-evoked glutamate release (2.4 x 10-11 M, control) was significantly decreased in affected spinal segments of allodynic rats (8.9 x 10-15 M). Treatment with intrathecal S(+)-ibuprofen or SC-560, beginning 2 h after ligation, prevented the decrease in paw withdrawal threshold, the brush-evoked increase in [PGE2]dialysate, and the change in EC50 of PGE2-evoked glutamate release. R(-)-ibuprofen or SC-236 had no effect.


The results of this study provide solid evidence that spinal prostaglandins, synthesized by cyclooxygenase-1 in the first 4-8 h after ligation, are critical in the pathogenesis of prostaglandin-dependent and prostaglandin-independent allodynia and that their early pharmacologic disruption affords protection against this neuropathic state in the rat.

PROSTAGLANDINS are synthesized in the spinal cord in response to repetitive C-fiber (noxious) input, 1–5an effect independent of their role in pain and inflammation in the periphery. This finding is consistent with the localization of cyclooxygenase and prostaglandin binding sites in spinal laminae known to receive nociceptive afferent input 6–9and the ability of intrathecal nonsteroidal antiinflammatory drugs to block nociceptive behaviors and the concurrent release of prostaglandins into spinal cerebrospinal fliud. 10 

The facilitatory effect of spinal prostaglandins on nociceptive transmission 11suggests that prostaglandin synthesis early after nerve injury could be important in the development of allodynia, a neuropathic condition in which pain is evoked by a stimulus that does not normally evoke pain (e.g. , cold breeze or light touch 12). Indeed, studies of the effects of spinal prostaglandins on normal nociceptive neurotransmission provide clues as to how this could occur. Prostaglandins directly stimulate wide dynamic range neurons in the rat dorsal horn, 13sensitize these same neurons to noxious mechanical stimulation, 8enhance glutamate and substance P release from primary afferent terminals in the spinal cord, 14,15and reduce glycine-mediated inhibition in the spinal cord. 16 

Certainly, there is growing experimental evidence linking spinal prostaglandins to allodynia. Intrathecal prostaglandin D2(PGD2), prostaglandin E2(PGE2), and prostaglandin F(PGF) evoke robust allodynia in conscious mice and rats, an effect blocked by prostaglandin receptor antagonists. 14,17–19Allodynia induced by intrathecal strychnine 20features brush-evoked increases in the concentration of PGE2in spinal dialysate 21([PGE2]dialysate) and attenuation by intrathecal cyclooxygenase inhibitors. 22Bicuculline-induced allodynia is also blocked by cyclooxygenase inhibitors given locally to the affected spinal segments. 23In the spinal nerve ligation model, 24significant upregulation of the cyclooxygenase-2 protein has been reported in the spinal cord as early as 1 day after ligation, 25and allodynia can be temporarily reversed by an intrathecal cyclooxygenase inhibitor or prostaglandin subtype receptor antagonist. 26Experimental allodynia is also sensitive to spinal N-methyl D-aspartate (NMDA) receptor blockade. 23,27–32NMDA receptor activation, an essential trigger for central prostaglandin synthesis, 33is not normally involved in low-threshold neurotransmission. 34The emergence of NMDA and prostaglandin properties in allodynia suggests a convergence of cellular events by which low- and high-threshold input are processed. Thus, the enhancement of glutamate release, NMDA-mediated Ca2+influx, activation of cyclooxygenase, and prostaglandin synthesis may be synaptic events beginning immediately after nerve injury that initiate the subsequent and more complex changes leading to permanent (prostaglandin-independent) allodynia.

In the current study, we used the rat spinal nerve ligation model to test the hypothesis that mechanical allodynia is composed of an early spinal prostaglandin-dependent phase, the disruption of which prevents the establishment of delayed (prostaglandin-independent) allodynia.

Materials and Methods

All studies were conducted in accordance with the guidelines of the Institutional Animal Care Committee of Memorial University of Newfoundland.

In Vivo  Experiments


Male Sprague-Dawley rats were obtained from the Vivarium of Memorial University of Newfoundland and were housed in standard cages with wood chip bedding. Animals had free access to food and water and were housed singly after surgery. A 12- to 12-h light cycle (lights on at 0700 h) was used throughout.

Intrathecal Catheterization.

Intrathecal microdialysis catheters were constructed using a method modified from Marsala et al.  35and implanted in rats weighing 100–120 g, as previously described. 36A 6.5-cm catheter was used for termination near the lumbar enlargement in these smaller animals. The catheter was sterilized with 70% alcohol and filled with sterile saline. Under halothane anesthesia, the catheter was inserted through an incision in the atlantooccipital membrane of the cisterna magna. The catheter was externalized behind the head and sealed with stainless steel wire. Drugs were delivered intrathecally through a separate channel of the microdialysis catheter. Rats with normal motor behavior were housed separately and allowed to recover for 3 days before spinal nerve ligation. A laminectomy was performed at the end of the experiments to visually confirm the position of the intrathecal catheter except for rats used in the in vitro  experiments where this was not feasible. No animals were excluded from the results because of an improperly positioned catheter.


Neuropathy was induced using the method of Kim and Chung, 24as modified by Chaplan et al.  37Rats were anesthetized with halothane, and a dorsal midline incision was made from L3–S2. Using blunt dissection, the left posterior interarticular process was located and resected to aid visualization of the L6 transverse process. This was partially removed to expose the L4 and L5 spinal nerves. Once separated, the L5 spinal root was tightly ligated with 6-0 silk thread. The L6 spinal root was then located medial and caudal to the sacroiliac junction and ligated in the same manner. The wound was closed with 4-0 silk sutures in two layers and cleaned with 70% alcohol, and a 5-ml bolus of lactated Ringer's solution was injected intraperitoneally. The animal was placed under a warming light and returned to the animal care facility after recovery from anesthesia.


Intrathecal microdialysis was conducted in conscious, unrestrained animals 1 day before and on selected days after spinal nerve ligation. Briefly, the inflow channel of the microdialysis catheter was connected to a microsyringe pump. Artificial cerebrospinal fluid, sterilized by filtration through a 0.22-μm pore filter (Micron Separations, Inc., Edmonton, Canada), was perfused at a rate of 10 μl/min. This flow rate, selected on the basis of previous work, 10,26produced an in vitro  recovery rate of 31.7 ± 2.7% of the external PGE2concentration. 26After a 30-min stabilization period, a control sample (100 μl, no paw stimulation) was collected. The plantar surface of the affected hind paw was then lightly brushed for 2 min with a cotton-tipped applicator followed by a 1-min rest period (no brushing). This stimulation protocol was repeated at regular intervals so that rats received a total of 6 min of brushing in a 10-min interval. Dialysate samples were collected on ice over 10 min and immediately frozen at −80°C until assayed. The concentration of PGE2was determined using a commercially available enzyme-linked immunosorbent assay kit (Cayman Chemical, Ann Arbor, MI).


Mechanical allodynia was quantified by determining the 50% withdrawal threshold using von Frey filaments. 37Briefly, rats were placed in a plastic cage with a wire mesh bottom to allow access to the plantar surface of the left hind paw. Thresholds were measured after a 20-min acclimatization period. Testing was performed during the daylight portion of the circadian cycle (0800–1800 h), and allodynia was defined as a withdrawal threshold of no more than 4 g. The investigators were blinded to the identity of the treatments throughout the study.


All drugs were dissolved in 100% dimethyl sulfoxide and diluted with normal saline at the time of injection to yield a final concentration of 50%. Intrathecal drugs were injected using a Hamilton syringe and a hand-operated microsyringe pump. All equipment was sterilized with 70% alcohol before injection and thoroughly rinsed with 0.9% sterile saline. Intrathecal drugs were injected to conscious, unrestrained rats in a volume of 5 μl followed by 8 μl sterile saline. The intrathecal catheter was immediately resealed with the stainless steel plug. S(+)-ibuprofen, the active isomer of the nonselective cyclooxygenase inhibitor, and R(−)-ibuprofen, the inactive cyclooxygenase inhibitor, were purchased from Research Biochemicals International (Natick, MA); SC-236, a selective cyclooxygenase-2 inhibitor, and SC-560, a selective cyclooxygenase-1 inhibitor, were generous gifts from Searle (Skokie, IL).

In Vitro  Experiments

Tissue Preparation.

Rats were anesthetized with urethane (1.5 g/kg intraperitoneally) and killed by decapitation. The spinal cord was removed immediately using hydraulic expulsion. After careful removal of the dural and arachnoid membranes, the lumbar region was visually identified, excised, and mounted on cutting blocks. The tissue was submersed in sucrose-modified artificial cerebrospinal fluid aerated with 95% O2and 5% CO2and sectioned using a Vibratome. Spinal cord slices (600–800 μm) were placed in artificial cerebrospinal fluid aerated with 95% O2and 5% CO2and stored at room temperature.

Determination of Glutamate Concentration.

The basal and PGE2-evoked release of glutamate from spinal cord slices was determined using a modified enzymatic assay. 38,39Briefly, the slices were transferred from the storage buffer to a 1 × 1-cm cuvette containing (mM): NaCl (120), KCl (3.1), NaH2PO4(1.25), HEPES (25), glucose (4), MgCl2(1), CaCl2(2), glutamate dehydrogenase (40 U/ml; BioVectra, Charlottetown, Canada), and nicotinamide adenine dinucleotide phosphate (1) (pH 7.4). The cuvette was placed in a temperature-controlled Shimadzu RF-1501 spectrofluorometer. The contents of the cuvette were continuously mixed using a magnetic stirrer, and the temperature was held at 37°C. HEPES buffer containing the tissue slice was oxygenated with 100% O2throughout the experiment. Glutamate released from the tissue was immediately oxidized to α-ketoglutarate by glutamate dehydrogenase, thereby preventing neuronal reuptake of glutamate. 39,40The reduced form of nicotinamide adenine dinucleotide phosphate generated from this reaction was quantitated using spectrofluorometry (excitation, 335 nm; emission, 430 nm; delay, <1 s). PGE2(Biomol) was initially dissolved in ethanol and evaporated under 100% nitrogen gas. It was then dissolved in normal saline, diluted with the same to yield the desired concentrations, and added directly to the cuvette using a microsyringe. Each concentration of PGE2was tested using a separate slice so that a full PGE2concentration-response curve was determined in each animal. Standard curves were prepared on each day of analysis.

Statistical Analysis

All data are reported as the mean ± SEM. Paw withdrawal thresholds are presented in grams. The concentration of PGE2in microdialysis samples ([PGE2]dialysate) is presented in pmol/ml or as the percent of baseline. Glutamate release is reported as pmol · min−1· mg−1protein. Statistical testing was performed using Sigmastat® 2.03 for Windows® (SPSS, Inc., Chicago, IL). Pretreatment and posttreatment values were compared within each treatment group using one-way repeated-measures ANOVA followed by the Newman-Keuls test. Comparisons were also made across all drug- and vehicle-treated groups at each time point using one-way completely randomized ANOVA followed by the Newman-Keuls test. Concentration-response analysis was performed using methods from Tallarida and Murray. 41 P < 0.05 was considered to be statistically significant.


Rats undergoing L5–L6 spinal nerve ligation showed a significant decrease in paw withdrawal threshold, from ≥ 15 g to ≤ 4 g, beginning 1 day later (fig. 1A). This enhancement in mechanical sensitivity remained stable for at least 20 days and was confined to the plantar surface of the hind paw ipsilateral to nerve ligation. In addition, the affected hind paw was often kept in an elevated and cupped position, minimizing contact with the cage floor. These rats were otherwise healthy and showed normal feeding behavior and regular weight gain. The paw withdrawal thresholds of all sham-operated rats remained unchanged from presurgical values (fig. 1A). The basal concentration of PGE2in spinal dialysate samples, which ranged from 1.6 ± 0.4 pmol/ml to 4.1 ± 1.4 pmol/ml, was unchanged from preligation values up to 20 days after nerve ligation (fig. 1B).

Before nerve ligation, brushing the plantar surface of the left hind paw had no effect on [PGE2]dialysate(fig. 2A), nor did it evoke any protective or nocifensive-like behavioral responses. In contrast, the same stimulus evoked a significant increase in [PGE2]dialysate(257 ± 62% of baseline) (fig. 2B) beginning 1 day after ligation and nocifensivelike behaviors, including protection of the affected hind paw, abrupt paw withdrawal, vocalization, and avoidance of the stimulus applicator. The [PGE2]dialysatepeaked during brushing and declined gradually (up to 30 min) thereafter (fig. 2B). This effect persisted for at least 5 days (data not shown). By day 10, brushing the left hind paw was without effect on the [PGE2]dialysate(fig. 2C), although the animals were still allodynic (paw withdrawal threshold, ≤ 4 g) (fig. 1A). Brushing the contralateral (right) hind paw of nerve-ligated or sham-operated rats had no effect on the [PGE2]dialysateor behavior up to 5 days after ligation (fig. 2D).

To test the relevance of spinal prostanoid synthesis to allodynia, separate groups of rats were treated with 100 μg of intrathecal S(+)-ibuprofen, R(−)-ibuprofen, or vehicle beginning 2 h after spinal nerve ligation (figs. 3 and 4). Rats treated with intrathecal S(+)-ibuprofen during an 8-h period (four intrathecal injections of 100 μg given every 2 h) exhibited normal posture (no cupped hind paw) and paw withdrawal thresholds (≥ 15 g) (fig. 3A) for at least 24 days after nerve injury. Brushing the ipsilateral hind paw had no effect on the [PGE2]dialysatein these same animals (figs. 4A and B). In contrast, ligated rats treated with intrathecal S(+)-ibuprofen for 4 h (100 μg every 2 h for a total of two injections) (fig. 3A), R(−)-ibuprofen for 4 h or 8 h (100 μg given every 2 h for a total of two or four injections, respectively) (fig. 3B), or vehicle using the same treatment schedules (fig. 3C) developed allodynia that was indistinguishable from untreated, ligated rats in terms of paw withdrawal threshold (fig. 1) and brush-evoked changes in behavior and [PGE2]dialysate(figs. 4A–C). Area-under-the-curve analysis showed no difference (P = 0.65) in the brush-evoked [PGE2]dialysateof the R(−)-ibuprofen treated group (fig. 4B) versus  ligated, untreated rats (fig. 2B). The ability of brushing to increase [PGE2]dialysatein the R(−)-ibuprofen (allodynic) group was lost by day 10 (fig. 4C), similar to that in ligated, untreated (allodynic) rats (fig. 2C).

To determine whether allodynic animals also show pharmacodynamic changes to spinal prostanoids, separate groups of rats undergoing spinal nerve ligation were treated with intrathecal R(−)- or S(+)-ibuprofen for 8 h (100 μg given every 2 h for a total of four injections). The concentration-response effect of PGE2on glutamate release was compared using spinal cord slices prepared from allodynic and nonallodynic animals. PGE2evoked the release from the slices of sham-operated (nonallodynic) rats in a concentration-dependent manner (fig. 5) (table 1). Whereas the PGE2concentration-response curves of ligated (untreated) and ligated (R[−]-ibuprofen treated) rats showed dramatic leftward shifts from control, there was no difference between sham-operated (nonallodynic) and ligated (S[+]-ibuprofen [8 h] treated, nonallodynic) rats (fig. 5) (table 1).

To determine whether this protective effect resulted from early cyclooxygenase-1 or cyclooxygenase-2 inhibition, rats were treated with intrathecal SC-560 or SC-236 (100 μg) beginning 2 h after spinal nerve ligation. Paw withdrawal thresholds in ligated rats treated with an 8-h regimen of intrathecal SC-560 were slightly, but not significantly, decreased as compared to preligation values for up to 20 days (fig. 6A). For the 4-h treatment, paw withdrawal thresholds on days 2 and 5 were statistically different from preligation values but never decreased below 9 g. In contrast, animals treated with intrathecal SC-236 (4- or 8-h regimen) showed a time course of paw withdrawal threshold (fig. 6B) unchanged from ligated, untreated rats (fig. 1A); all time points were significantly different from preligation values.


Ligation of the left L5–L6 spinal nerves produced a robust sensitization to innocuous mechanical stimulation on the plantar surface of the left hind paw beginning 1 day later. This was indicated by the marked decrease in the paw withdrawal threshold (≤ 4 g), brush-evoked nociceptive-like behavior, and concurrent stimulus-evoked increase in [PGE2]dialysate. The area of sensitization remained highly circumscribed over time and was absent in sham-operated controls. Brushing the hind paws of sham-operated animals or outside the affected dermatomes, including the contralateral hind paw, of ligated rats was without effect. These results indicate that prostanoid synthesis is recruited in the spinal cord of animals that underwent L5–L6 nerve ligation, a response that is spatially and temporally linked to a stimulus that elicited allodynia in the same animals during the prostaglandin-dependent phase of allodynia. The brush-evoked release of PGE2, which is in decline by day 5, 26was absent by day 10, even though brush-evoked allodynia persisted for up to 20 days independent of any detectable change in [PGE2]dialysate.

The connection between allodynia and spinal prostaglandins appears to involve more than just the emergence of brush-evoked prostanoid synthesis. There was also a marked decrease in the EC50of PGE2-evoked glutamate release in slices prepared from the affected (L5–L6) segments of allodynic rats compared with thoracic slices from the same animals (data not shown) or slices of L5–L6 spinal cord from sham-operated controls. Spinal glutamate has a recognized role in neuropathic states, including the NMDA receptor-coupled activation of phospholipase A2, nitric oxide synthase, cyclooxygenase, and various kinases, 42–44many of whose products diffuse into the extracellular space to enhance neurotransmitter release. The mechanisms underlying the increased potency of PGE2are currently being investigated but could include the amplification of spinal prostaglandin synthesis in nerve-injured segments in which increased expression of cyclooxygenase-2 has been shown to occur. 25To our knowledge, a marked pharmacodynamic change in the affected spinal segments of demonstrably allodynic animals has not been previously reported.

Allodynia and spinal prostaglandins appear to be functionally linked in the early postinjury period. Brush-evoked spinal PGE2release and allodynia were significantly attenuated by intrathecal S(+)-ibuprofen, a cyclooxygenase-1 and cyclooxygenase-2 inhibitor, but not R(−)-ibuprofen, in the spinal nerve ligation model. 26The relevance of spinal prostaglandins to the development of allodynia is further strengthened by the results of the current study. When given early after ligation, all rats treated with 100 μg S(+)-ibuprofen developed none of the characteristic features of allodynia for up to 24 days. Paw withdrawal thresholds remained normal (≥ 15 g); brushing had no effect on the [PGE2]dialysate; and there was no change in PGE2-evoked glutamate release from spinal cord slices, compared with sham-operated controls. In contrast, R(−)-ibuprofen, an inactive cyclooxygenase inhibitor and control treatment in these experiments, had no protective effect whatsoever. R(−)-ibuprofen has been shown to inhibit the transcription factor, NFκB in vitro , 45,46a factor known to regulate the expression of cyclooxygenase-2 in macrophages and human gingival fibroblasts. 47,48To the extent that upregulation of cyclooxygenase-2 after injury is important in the establishment of allodynia, a factor yet to be confirmed, the inhibition of NFκB by R(−)-ibuprofen would be predicted to attenuate allodynia. No such effect was observed, but further studies are required to address this possibility. Overall, the results of these experiments support the hypothesis that spinal prostaglandins are critical factors in the initiation of changes (e.g. , central synaptic excitability and neuronal sensitization) leading to prostaglandin-dependent and prostaglandin-independent allodynia. The protective effect of S(+)-ibuprofen also depended on the duration of treatment (only the 8-h regimen was effective), thereby suggesting that prostaglandin synthesis must be inhibited for a minimum period of time after ligation to effectively prevent or interrupt the events leading to allodynia. The results of this study are in general agreement with and build on the preliminary work of Zhao et al. , 25who showed that the nonselective cyclooxygenase inhibitor, indomethacin, given spinally 2 h after spinal nerve ligation, partially attenuated the development of allodynia for up to 4 weeks in the rat.

To extend this work, cyclooxygenase-1- and cyclooxygenase-2-selective inhibitors were used to provide information about the isozymes catalyzing spinal prostaglandin synthesis in the nerve-injured cord. The doses of SC-236 (cyclooxygenase-2-selective) and SC-560 (cyclooxygenase-1-selective) were chosen from previous dose-response studies in our laboratory, 26in which we showed a dramatic difference in the inhibitory effect of SC-236 and SC-560 on established allodynia using near equimolar intrathecal doses suggesting that isozyme selectivity is maintained at the doses used. In the current study, the contrast in effects with SC-560 and SC-236 was also striking. Whereas SC-560 (4- and 8-h treatments) prevented the emergence of all features of allodynia, an effect comparable with that of S(+)-ibuprofen (8-h treatment), SC-236 (4- or 8-h treatment) was totally ineffective. These results indicate that cyclooxygenase-1, known to be constitutively expressed in the spinal cord, 49catalyzes the synthesis of prostaglandins in the period immediately after ligation. It is this early synthesis that appears necessary for the development of prostaglandin-dependent and prostaglandin-independent allodynia. In contrast, inducible cyclooxygenase-2, which is known to be upregulated 24 h after ligation, 25appears to be important in maintaining prostaglandin-dependent allodynia, a stage when cyclooxygenase-1 inhibitors are largely ineffective. 26Thus, cyclooxygenase-2 may need to achieve sufficient expression over time. The differential localization of constitutive cyclooxygenase-1 and constitutive cyclooxygenase-2 reported in primary afferent terminals of the rat 8may also explain the early sensitivity to cyclooxygenase-1 inhibitors.

An obvious question arising from these experiments is whether allodynia would have developed without drug treatment (e.g. , unsuccessful nerve ligation). This outcome is unlikely because the protective effect was stereospecific, selective for cyclooxygenase-1, and dependent on treatment duration. Moreover, more than 90% of naive animals undergoing L5–L6 spinal nerve ligation in our laboratory develop allodynia within 1 day of surgery, a result comparable with that reported by Chaplan et al.  37Thus, the uniform failure of nerve ligation within an experimental group is improbable and inconsistent with the results of this study. These data indicate that treatment with spinal cyclooxygenase inhibitors early after ligation prevents or, at the least, significantly delays (> 25 days), the emergence of prostaglandin-dependent and prostaglandin-independent allodynia.

Figure 7is a proposed model of the early (synaptically mediated) events effecting prostaglandin-dependent and prostaglandin-independent allodynia after L5–L6 spinal nerve ligation. Glutamate, released from the central terminals of primary afferent fibers beginning immediately after nerve injury, 50,51activates postsynaptic NMDA receptors, an essential step in central sensitization and allodynia. 52,53Subsequent intracellular events, including an increase in Ca2+concentration and the release of arachidonic acid via  PLA2, trigger the early synthesis of spinal prostaglandins by constitutive enzymes (cyclooxygenase-1 more so than cyclooxygenase-2). After diffusion to the extracellular space, spinal prostaglandins enhance the excitability of adjacent cells 8,13and feed back to reinforce glutamate release. 14,15,54Sustained NMDA receptor activation initiates the expression of inducible cyclooxygenase-2, 49thereby enhancing the capacity for stimulus-evoked prostaglandin synthesis in the spinal cord (cyclooxygenase-2 more so than cyclooxygenase-1) and the emergence of prostaglandin-dependent allodynia. 26During the next few weeks, prostaglandin-dependent allodynia recedes and leaves long-term, prostaglandin-independent allodynia. The latter may reflect phenotypic changes in primary afferents, sprouting of primary afferents into outer laminae, and altered gene regulation. 55–59The ability of cyclooxygenase inhibitors to prevent the emergence of prostaglandin-dependent and prostaglandin-independent allodynia strongly suggests that these two states are related.

In conclusion, the results of this study provide solid evidence that spinal prostaglandins, synthesized immediately after spinal nerve ligation, are critical in the development of prostaglandin-dependent and prostaglandin-independent allodynia, and that their early pharmacologic disruption affords protection against this neuropathic state. They build on and are consistent with previous work showing the prostaglandin-dependent nature of allodynia in this model, 25,26but are seemingly at odds with anecdotal reports that nonsteroidal antiinflammatory drugs provide little, if any, benefit to patients with allodynia. How can this apparent contradiction be resolved? Our data indicate that the timing of treatment, the type of cyclooxygenase inhibitor used, and perhaps the route of administration are critical factors influencing the effectiveness of nonsteroidal antiinflammatory drugs, not in relieving neuropathic pain per se , but in preventing its emergence. They suggest that nonsteroidal antiinflammatory drugs would need to be given early (i.e. , hours) after injury, after which their clinical value would decrease. The latter prediction is seemingly consistent with current clinical experience. Although further research is needed, the potential implications of this research for clinical neuropathic pain are exciting.

The authors thank Tony L. Yaksh, Ph.D., and Martin Marsala, M.D., for their guidance with the microdialysis and spinal nerve ligation model.


Malmberg AB, Yaksh TL: Hyperalgesia mediated by spinal glutamate or substance P receptor blocked by spinal cyclooxygenase inhibition. Science 1992; 257: 1276–9
Yaksh TL, Malmberg AB: Spinal actions of NSAIDs in blocking spinally mediated hyperalgesia: The role of cyclooxygenase products. Agents Actions Suppl 1993; 41: 89–100
McCormack K: Non-steroidal anti-inflammatory drugs and spinal nociceptive processing. Pain 1994; 59: 9–43
McCormack K: The spinal actions of nonsteroidal anti-inflammatory drugs and the dissociation between their anti-inflammatory and analgesic effects. Drugs 1994; 47: 28–47
Southall MD, Michael RL, Vasko MR: Intrathecal NSAIDS attenuate inflammation-induced neuropeptide release from rat spinal cord slices. Pain 1998; 78: 39–48
Matsumura K, Watanabe Y, Onoe H, Watanabe Y: Prostacyclin receptor in the brain and central terminals of the primary sensory neurons: An autoradiographic study using a stable prostacyclin analogue [3H]iloprost. Neuroscience 1995; 65: 493–503
Goppelt-Struebe M, Beiche F: Cyclooxygenase-2 in the spinal cord: localization and regulation after a peripheral inflammatory stimulus. Adv Exp Med Biol 1997; 433: 209–12
Willingale HL, Gardiner NJ, McLymont N, Giblett S, Grubb BD: Prostanoids synthesized by cyclo-oxygenase isoforms in the rat spinal cord and their contribution to the development of neuronal hyperexcitability. Br J Pharmacol 1997; 122: 1593–604
Vane JR, Bakhle YS, Botting RM: Cyclooxygenases 1 and 2. Annu Rev Pharmacol Toxicol 1998; 38: 97–120
Malmberg AB, Yaksh TL: Cyclooxygenase inhibition and the spinal release of prostaglandin E2and amino acids evoked by paw formalin injection: A microdialysis study in unanaesthetized rats. J Neurosci 1995; 15: 2768–76
Millan MJ: The induction of pain: An integrative review. Prog Neurobiol 1999; 57: 1–164
Merskey H: Classification of chronic pain, part II. Pain Suppl 1986; 3: S215–21
Baba H, Kohno T, Moore KA, Woolf CJ: Direct activation of rat spinal dorsal horn neurons by prostaglandin E2. J Neurosci 2001; 21: 1750–6
Ferreira SH, Lorenzetti BB: Intrathecal administration of prostaglandin E2 causes sensitization of the primary afferent neuron via spinal release of glutamate. Inflamm Res 1996; 45: 499–502
Hingtgen CM, Waite KJ, Vasko MR: Prostaglandins facilitate peptide release from rat sensory neurons by activating the adenosine 3′, 5′-cyclic monophosphate transduction cascade. J Neurosci 1995; 15: 5411–9
Ahmadi S, Lippross S, Neuhuber WL, Zeilhofer HU: PGE(2) selectively blocks inhibitory glycinergic neurotransmission onto rat superficial dorsal horn neurons. Nat Neuroscience 2002; 5: 34–40
Uda R, Horiguchi S, Ito S, Hyodo M, Hayaishi O: Nociceptive effects induced by intrathecal administration of prostaglandin D2, E2, or F2 alpha to conscious mice. Brain Res 1990; 510: 26–32
Minami T, Nishihara I, Sakamoto K, Ito S, Hyodo M, Hayaishi O: Blockade by ONO-NT-012, a unique prostanoid analogue, of prostaglandin E2-induced allodynia in conscious mice. Br J Pharmacol 1995; 115: 73–6
Minami T, Uda R, Horiguchi S, Ito S, Hyodo M, Hayaishi O: Allodynia evoked by intrathecal administration of prostaglandin E2 to conscious mice. Pain 1994; 57: 217–23
Yaksh T: Behavioral and autonomic correlates of the tactile evoked allodynia produced by spinal glycine inhibition: Effects of modulatory receptor systems and excitatory amino acid antagonists. Pain 1989; 37: 111–23
Milne B, Hall SR, Sullivan ME, Loomis C: The release of spinal prostaglandin E2 and the effect of nitric oxide synthetase inhibition during strychnine-induced allodynia. Anesth Analg 2001; 93: 728–33
Hall SR, Milne B, Loomis C: Spinal action of ketorolac, S(+)- and R(−)-ibuprofen on non-noxious activation of the catechol oxidation in the rat locus coeruleus: Evidence for a central role of prostaglandins in the strychnine model of allodynia. A nesthesiology 1999; 90: 165–73
Zhang Z, Hefferan MP, Loomis CW: Topical bicuculline to the rat spinal cord induces highly localized allodynia that is mediated by spinal prostaglandins. Pain 2001; 92: 351–61
Kim SH, Chung JM: An experimental model for peripheral neuropathy produced by segmental spinal nerve ligation in the rat. Pain 1992; 50: 355–63
Zhao Z, Chen SR, Eisenach JC, Busija DW, Pan HL: Spinal cyclooxygenase-2 is involved in development of allodynia after nerve injury in rats. Neuroscience 2000; 97: 743–8
Hefferan MP, Carter P, Haley M, Loomis CW: Spinal nerve injury activates prostaglandin synthesis in the spinal cord that contributes to early maintenance of tactile allodynia. Pain 2003; 101: 139–47
Lee YW, Yaksh TL: Analysis of drug interaction between intrathecal clonidine and MK-801 in peripheral neuropathic pain rat model. A nesthesiology 1995; 82: 741–8
Khandwala H, Hodge E, Loomis C: Comparable dose-dependent inhibition of AP-7 sensitive strychnine-induced allodynia and paw pinch-induced nociception by mexiletine in the rat. Pain 1997; 72: 299–308
Leung A, Wallace M, Ridgeway B, Yaksh T: Concentration-effect relationship of intravenous alfentanil and ketamine on peripheral neurosensory thresholds, allodynia and hyperalgesia of neuropathic pain. Pain 2001; 91: 177–87
Bennett AD, Everhart AW, Hulsebosch CE: Intrathecal administration of an NMDA or a non-NMDA receptor antagonist reduces mechanical but not thermal allodynia in a rodent model of chronic central pain after spinal cord injury. Brain Res 2000; 859: 72–82
Hao JX, Xu XJ: Treatment of a chronic allodynia-like response in spinally injured rats: Effects of systemically administered excitatory amino acid receptor antagonists. Pain 1996; 66: 279–85
Minami T, Matsumura S, Okuda-Ashitaka E, Shimamoto K, Sakimura K, Mishina M, Mori H, Ito S: Characterization of the glutamatergic system for induction and maintenance of allodynia. Brain Res 2001; 895: 178–85
Dirig DM, Yaksh TL: In vitro prostanoid release from spinal cord following peripheral inflammation: Effects of substance P, NMDA and capsaicin. Br J Pharmacol 1999; 126: 1333–40
Urban L, Thompson SW, Dray A: Modulation of spinal excitability: Co-operation between neurokinin and excitatory amino acid neurotransmitters. Trends Neurosci 1994; 17: 432–8
Marsala M, Malmberg AB, Yaksh TL: The spinal loop dialysis catheter: Characterization of use in the unanesthetized rat. J Neurosci Methods 1995; 62: 43–53
Sherman SE, Loomis CW: Morphine insensitive allodynia is produced by intrathecal strychnine in the lightly anesthetized rat. Pain 1994; 56: 17–29
Chaplan SR, Bach FW, Pogrel JW, Chung JM, Yaksh TL: Quantitative assessment of tactile allodynia in the rat paw. J Neurosci Methods 1994; 53: 55–63
Nicholls DG, Sihra TS, Sanchez-Prieto J: Calcium-dependent and -independent release of glutamate from synaptosomes monitored by continuous fluorometry. J Neurochem 1987; 49: 50–7
Bezzi P, Carmignoto G, Pasti L, Vesce S, Rossi D, Rizzini BL, Pozzan T, Volterra A: Prostaglandins stimulate calcium-dependent glutamate release in astrocytes. Nature 1998; 391: 281–5
Vesce S, Bezzi P, Rossi D, Meldolesi J, Volterra A: HIV-1 gp120 glycoprotein affects the astrocyte control of extracellular glutamate by both inhibiting the uptake and stimulating the release of the amino acid. FEBS Lett 1997; 411: 107–9
Tallarida RJ, Murray RB: Manual of Pharmacologic Calculations with Computer Programs, 2nd edition. New York, Springer-Verlag, 1987, pp 10–30
Yaksh TL, Hua XY, Kalcheva I, Nozaki-Taguchi N, Marsala M: The spinal biology in humans and animals of pain states generated by persistent small afferent input. Proc Natl Acad Sci U S A 1999; 96: 7680–6
Yaksh, TL: Spinal systems and pain processing: Development of novel analgesic drugs with mechanistically defined models. Trends Pharmacol Sci 1999; 20: 329–37
Yang LC, Marsala M, Yaksh TL: Characterization of time course of spinal amino acids, citrulline and PGE2 release after carrageenan/kaolin-induced knee joint inflammation: A chronic microdialysis study. Pain 1996; 67: 345–54
Scheuren N, Bang H, Munster T, Brune K, Pahl A: Modulation of transcription factor NF-κB by enantiomers of the nonsteroidal drug ibuprofen. Br J Pharmacol 1998; 123: 645–52
Grilli M, Pizzi M, Memo M, Spano P: Neuroprotection by aspirin and sodium salicylate through blockage of NF-κB activation. Science 1996; 274: 1383–5
Mestre JR, Mackrell PJ, Rivadeneira DE, Stapleton PP, Tanabe T, Daly JM: Redundancy in the signaling pathways and promoter elements regulating cyclooxygenase-2 gene expression in endotoxin-treated macrophage/monocytic cells. J Biol Chem 2001; 276: 3977–82
Nakao S, Ogata Y, Shimizu-Sasaki E, Yamazaki M, Furuyama S, Sugiya H: Activation of NFκB is necessary for IL-1β-induced cyclooxygenase-2 (COX-2) expression in human gingival fibroblasts. Mol Cell Biochem 2000; 209: 113–8
Vanegas H, Schaible HG: Prostaglandins and cyclooxygenases in the spinal cord. Prog Neurobiol 2001; 64: 327–63
Kawamata M, Omote K: Involvement of increased excitatory amino acids and intracellular Ca2+ concentration in the spinal dorsal horn in an animal model of neuropathic pain. Pain 1996; 68: 85–96
Farooque M, Hillered L, Holtz A, Olsson Y: Changes of extracellular levels of amino acids after graded compression trauma to the spinal cord: An experimental study in the rat using microdialysis. J Neurotrauma 1996; 13: 537–48
Kajander KC, Wakisaka S, Bennett GJ: Spontaneous discharge originates in the dorsal root ganglion at the onset of a painful peripheral neuropathy in the rat. Neurosci Lett 1992; 138: 225–8
Yoon YW, Na HS, Chung JM: Contributions of injured and intact afferents to neuropathic pain in an experimental rat model. Pain 1996; 64: 27–36
Vasko M, Campbell W, Waite K: Prostaglandin E2 enhances bradykinin stimulated release of neuropeptides from rat sensory neurons in culture. J Neurosci 1994; 14: 4987–97
Wakisaka S, Kajander KC, Bennett GJ: Effects of peripheral nerve injuries and tissue inflammation on the levels of neuropeptide Y-like immunoreactivity in rat primary afferent neurons. Brain Res 1992; 598: 349–52
LaMotte RH, Shain CN, Simone DA, Tsai EF: Neurogenic hyperalgesia: Psychophysical studies of underlying mechanisms. J Neurophysiol 1991; 66: 190–211
Woolf CJ, Shortland P, Coggeshall RE: Peripheral nerve injury triggers central sprouting of myelinated afferents. Nature 1992; 355: 75–8
Herdegen T, Fiallos-Estrada CE, Schmid W, Bravo R, Zimmermann M: The transcription factors c-JUN, JUN D and CREB, but not FOS and KROX-24, are differentially regulated in axotomized neurons following transection of rat sciatic nerve. Mol Brain Res 1992; 14: 155–65
Cameron-Curry P, Aste N, Viglietti-Panzica C, Panzica GC: Immunocytochemical distribution of glial fibrillary acidic protein in the central nervous system of the Japanese quail (Coturnix coturnix japonica ). Anat Embryol Berl 1991; 184: 571–81