Intrathecal adenosine produces a remarkably prolonged effect to relieve mechanical hypersensitivity after peripheral nerve injury in animals. The purpose of the current study was to investigate whether this reflected an alteration in kinetics of adenosine in cerebrospinal fluid or in the number of spinal A1 adenosine receptors after nerve injury.


Male rats were anesthetized, and the left L5 and L6 spinal nerves were ligated. Two weeks later, a lumbar intrathecal catheter and intrathecal space microdialysis catheter were inserted. Adenosine, 20 microg, was injected intrathecally in these and in normal rats, and microdialysates of the intrathecal space were obtained. Radioligand binding studies of adenosine A1 receptors were determined in spinal cord tissue from other normal and spinal nerve-ligated rats.


Adenosine disappeared from rat cerebrospinal fluid within 30 min after intrathecal injection, with no difference between normal and spinal nerve-ligated animals. A1 adenosine receptor binding sites in the spinal cord were increased after spinal nerve ligation. This increase disappeared when adenosine deaminase was added to the membrane homogenates, suggestive of decreased endogenous adenosine in the membranes of nerve-ligated animals.


These data show that prolonged alleviation of hypersensitivity observed with intrathecal adenosine in this animal model of neuropathic pain is not due to prolonged residence in cerebrospinal fluid, although pharmacokinetics in tissues are unknown. Similarly, there is no evidence for up-regulation in spinal A1 adenosine receptors after spinal nerve ligation, and the adenosine deaminase experiment is consistent with a depletion of adenosine in spinal cord tissue after spinal nerve ligation.

A1ADENOSINE receptor stimulation produces analgesia by both peripheral and central mechanisms, and a variety of molecules are being developed, including direct agonists, allosteric modulators, and uptake or metabolism inhibitors to provide analgesia through this nonopioid mechanism. 1Peripherally, A1receptor stimulation inhibits excitation of dorsal root ganglion cells in culture 2and reduces release of calcitonin gene–related peptide and substance P from afferent terminals in the spinal cord. 3,4Centrally, A1receptors are present on spinal cord neurons, 5and stimulation of these receptors reduces C fiber–driven responses in dorsal horn neurons 6and causes antinociception to noxious thermal stimuli. 7Additionally, intrathecal injection of adenosine agonists reduces mechanical allodynia after peripheral nerve injury in rats, 8a model of hypersensitivity observed in patients with chronic pain, indicating a role in inhibition of central sensitization.

Previous studies have focused on synthetic analogs of adenosine and only indirectly examine the regulation of endogenous adenosine to modulate pain transmission in normal or pathologic conditions. Adenosine is a ubiquitous, water-soluble nucleoside, proposed to function in the nervous system as an inhibitory neuromodulator. 9Adenosine's action in the extracellular space is terminated mostly by reuptake via  equilibrative and concentrative transporters and subsequent intracellular metabolism. This metabolism is dependent on two enzymes, adenosine kinase and adenosine deaminase. Adenosine kinase facilitates a phosphorylation of adenosine to ADP and dominates control of extracellular adenosine concentrations under normal circumstances. 10The half-life of adenosine in blood is a few seconds, primarily because of rapid cellular uptake.

In contrast to the extremely short-lasting effects of adenosine after intravenous administration, we recently observed a prolonged (24-h) antiallodynic effect of intrathecally injected adenosine in rats with peripheral nerve injury. 11One purpose of the current study was to determine whether this prolonged duration of action reflected a pharmacokinetic mechanism. To test this, we used a recently described method of microdialysis of the intrathecal space 12in conscious rats with and without peripheral nerve injury and mechanical hypersensitivity. In addition, we previously noted in rats that intrathecal adenosine, in contrast to synthetic analogs, was essentially inactive against acute thermal pain but was active to reduce the degree of allodynia after nerve injury. 11,13A secondary purpose of the current study was to determine whether this increased efficacy of intrathecal administration of the endogenous compound adenosine in hypersensitivity states reflected up-regulation of A1adenosine receptors in the spinal cord in animals.

After Animal Care and Use Committee approval (Wake Forest University School of Medicine, Winston-Salem, NC), male rats (Harlan Sprague-Dawley) weighing 200–250 g at the time of surgery were studied. Animals were housed at 22°C and under a 12 h light–12 h dark cycle, with free access to food and water. Spinal nerve ligation was performed as previously described. 14During halothane anesthesia, the left L5 and L6 spinal nerves were isolated adjacent to the vertebral column and tightly ligated with 6-0 silk sutures distal to the dorsal root ganglion. After a recovery period of 13 days, left paw tactile hypersensitivity was confirmed by measuring hind paw thresholds to von Frey filaments, using an up–down method previously described. 15Only rats with a threshold below 4 g were further included for study.

In Vivo  Microdialysis

Sixteen male rats were randomly and equally divided for spinal nerve ligation or no surgery. Thirteen days later, all animals underwent implantation of intrathecal injection and loop dialysis catheters as previously described. 12,16During halothane anesthesia, polyethylene tubing and a loop microdialysis catheter were inserted through a small hole in the cisterna magna and directed caudad 8 cm such that their tips were in the lumbar intrathecal space. Rats showing neurologic deficits were immediately killed using an overdose of pentobarbital; the other animals were allowed to recover 4 or 5 days before the microdialysis experiment.

On the day of the study, microdialysis catheters were connected to 5-ml syringes containing artificial cerebrospinal fluid (CSF; 151.1 mm Na+, 2.6 mm K+, 0.9 mm Mg2+, 1.3 mm Ca2+, 122.7 mm Cl, 21 mm HCO3, 2.5 mm HPO4, and 3.5 mm dextrose) using a microinfusion pump at a flow rate of 5 μl/min. After a 1-h washout period and two 5-min baseline samples, 20 μg adenosine, in a volume of 10 μl, was injected through the intrathecal catheter, followed by 10 μl saline, and dialysis samples were collected on ice every 5 min for 25 min. Sample analysis was performed using high-pressure liquid chromatography, using a Rainin A1autosampler onto a 250 × 4.6 Luna C18 column (Phenomenex, Torrance, CA) at a flow rate of 1.3 ml/min with a mobile phase consisting of 10 mm ammonium phosphate, pH 6.0, with 14% methanol. Adenosine was determined using a Rainin Dynamax Model UV-D II absorbance detector at 254 nm, with limit of sensitivity of 10 pmol per injection and a coefficient of variation at 50 pmol of 7%.

Radioligand Binding

Preliminary experiments were performed to optimize binding conditions for pH, buffer components, adenosine deaminase concentration, time, temperature, and protein concentration. Spinal cords from normal and spinal nerve–ligated rats were quickly harvested and cut into lumbar and thoracic sections. These sections were dissected into left and right, dorsal and ventral quadrants and were stored at −70°C until used. Cords from 9 or 10 animals per group were prepared together. All tissue was stored for 1–4 weeks.

Tissue was thawed, quickly chopped, and then suspended in 170 mm Tris buffer of pH 7.4. Tissue was homogenized and then disrupted by sonication (5–7 s). The suspension was centrifuged at 3,000 rpm for 15 min (4°C). The supernatant was then centrifuged at 23,000 rpm for 30–40 min (4°C). The pellet was resuspended in the same buffer and disrupted by sonication for 5 s. Radioligand binding saturation experiments were performed with 250-μl aliquots of tissue (final volume 1 ml) and increasing concentrations of [3H]-DPCPX (90 pm–7 nm). Nonspecific binding was defined with the presence of 100 μm cold (unlabeled) adenosine. In some experiments, adenosine deaminase (1 IU/ml) was added. Competition experiments were performed with 200-μl aliquots of tissue. The final volume was 0.5 ml, the concentration of [3H]-DPCPX was 500 pm, and 1 IU/ml adenosine deaminase was added to the vials for total binding. The following adenosine A1receptor agonists were used in increasing concentrations: 2-chloroadenosine (2-CADO; 100 pm–10 μm), R-N6-phenylisopropyladenosine (R-PIA; 10 pm–10 μm), S-PIA (100 pm–100 μm), cyclohexyladenosine (CHA; 10 pm–1 μm), and cyclopentyladenosine (CPA; 10 pm–1 μm). The A2receptor agonist, 2-[4-(2-carboxyethyl)phenylethylamino]-5-N-ethylcarboxamidoadenosine (CGS21680; 100 pm–10 μm), was also used. Nonspecific binding was determined with 100-μm adenosine.

Tissues were incubated at room temperature for 1 h and then terminated by vacuum filtration over GF/B filters on a cell harvester (Brandel, Gaithersburg, MD) with three washes of 4 ml cold 170 mm Tris buffer. All experiments were performed in duplicate. Radioactivity bound was quantified by immersion of filters in 10 ml Biosafe II scintillation fluid counted in a scintillation counter. Saturation curves were constructed and analyzed with GraphPad Prism (GraphPad Software Inc., San Diego, CA).


Data are presented as mean ± SE or median ± quartiles, as indicated. Changes of adenosine concentration in CSF microdialysates after drug application compared with baseline were statistically analyzed using one-way repeated-measures analysis of variance followed by the Dunnett test. P < 0.05 was considered significant.


The median baseline concentration of two consecutive samples of adenosine in the dialysate was 0.007 μm [0.005 μm 25th, 0.014 μm 75th percentiles] in normal rats and 0.008 μm [0.007 μm, 0.015 μm] in spinal nerve–ligated rats (P = not significant between normal and nerve-ligated animals;fig. 1). Intrathecal injection of 20 μg adenosine led to an increase of the adenosine concentration in both normal and spinal nerve–ligated animals with a maximum 5 min after injection in normal animals (median: 2.97 μm [0.97 μm, 22.1 μm]) and 10 min in ligated animals (median: 0.68 μm [0.28 μm, 8.8 μm]). In the following 10 min, the concentration decreased, and after 20 min, the adenosine had totally disappeared in both groups. The measured concentrations at 5 and 10 min were significantly different from baseline in both groups, but there was no change in the time course of adenosine's disappearance between normal and spinal nerve–ligated animals.

Animal Study: Radioligand Binding

[3H]-DPCPX bound to a single site, whether experiments were performed with or without adenosine deaminase (fig. 2). Maximum specific binding was 85% in the absence of adenosine deaminase and 95% in the presence of adenosine deaminase. In lumbar spinal cord, binding differed according to treatment (spinal nerve ligation vs.  normal), and incubation conditions (adenosine deaminase present vs.  absent). Thus, maximal binding (Bmax) was increased 75–100% in lumbar spinal cord in animals after spinal nerve ligation compared with controls, without a change in binding affinity (fig. 2and table 1;P < 0.01 between Bmaxin either group without and with adenosine deaminase). In contrast, lumbar spinal nerve ligation had no effect on total binding or binding affinity in dorsal quadrants of thoracic spinal cord, whether adenosine deaminase was added or not (table 2). Competition studies revealed a single site in spinal nerve–ligated animals, with inhibition constants Kifor A1ligands of 0.7 nm (CPA), 1.1 nm (CHA), 1.7 nm (R-PIA), 15 nm (2-CADO), and 39 nm (S-PIA), and Kifor the A2ligand CGS 21680 of 123 nm.

The current study describes three key aspects of mechanism of action of intrathecally administered adenosine for the treatment of pain: lack of pharmacokinetic explanation for the prolonged duration of action, lack of up-regulation of adenosine A1receptor number as an explanation for increased efficacy after spinal nerve ligation, and suggestion that there is a depletion of spinal cord adenosine after spinal nerve ligation. Our companion clinical article, although studying normal volunteers and acute sensitization by intradermal capsaicin injection rather than nerve injury, is in accordance with the current findings in showing a brief (< 4 h) residence time of adenosine in CSF but a prolonged (24 h) duration of action after intrathecal bolus. 17 

Intrathecal Space Pharmacokinetics

Sampling of CSF to determine pharmacokinetics of drugs in CSF after intrathecal administration, especially in small animals, such as rats, is problematic. Not only is the CSF volume small, but removal of fluid for analysis, even with concomitant replacement with an equivalent volume of artificial CSF, can alter drug kinetics and distribution in the spinal intrathecal space. This is suggested in humans by the dramatically different pharmacokinetics and dynamics observed between single intrathecal injection of neostigmine through a No. 27 Whitacre needle without CSF sampling compared with injection through a No. 22 catheter with sampling. 18For these reasons, we chose to use microdialysis of the intrathecal space, 12which does not alter drug distribution by sampling artifacts, although it does introduce problems of mechanical obstruction in the intrathecal compartment, variable efficiency of dialysis probe recovery, and limited time resolution. 19 

Despite the qualifications necessary in performing pharmacokinetics using microdialysis probes, the current study provides two clear observations regarding intrathecal adenosine disposition. First, there was an obvious discrepancy between the duration of action to reduce mechanical hypersensitivity in this model in recent studies from the same laboratory (12–24 h) 11,20and the rapid disappearance of increased concentrations of adenosine in CSF within 30 min in the current study. Of course, this does not exclude a depot of adenosine in other neuraxial compartments, such as epidural fat or spinal cord tissues, but clearly shows that adenosine does not produce long-lasting effects by residing in CSF. Second, there is no difference in the resting concentrations of adenosine in CSF between normal and spinal nerve–ligated animals, nor is there a difference in the disappearance from CSF between these groups after exogenous bolus of adenosine in CSF.

The causes for the long pharmacodynamic effect of intrathecal adenosine in rats with peripheral nerve injury are unknown. Possibilities include repletion of depleted adenosine stores in spinal cord tissue itself (see A1Adenosine Receptor Binding section, third paragraph); prolonged activation of A1receptors, leading to decreased activity of adenosine kinase, thereby decreasing adenosine's uptake 21,22; or transient blockade of the sensitization process, which could take several hours to recover.

A1Adenosine Receptor Binding

The radioligand binding studies in the presence of adenosine deaminase to remove endogenous adenosine remaining in the membrane homogenates show no difference in either affinity or quantity of A1adenosine receptors in dorsal spinal cord after spinal nerve ligation. Others have speculated that the enhanced antinociceptive action of adenosine kinase inhibitors, like adenosine itself, in rats after spinal nerve ligation, might reflect increased A1adenosine receptors, 1which is clearly not supported by the current results. It is conceivable that receptor density in a small area of the spinal cord could change without being reflected in this homogenized tissue assay. We believe it is unlikely that we missed such a change because A1adenosine receptors are concentrated in superficial laminae of the spinal cord, and we examined dorsal spinal cord tissue, which was separated ipsilateral and contralateral to nerve injury, with no evidence of a difference between them.

However, there was a regional difference in adenosine receptor number, localized to the lumbar area—the site of nerve injury—when adenosine deaminase was not added to the membrane preparation. Endogenous adenosine, probably from both extracellular and intracellular sources, partitions into membranes during this preparation, and for this reason, adenosine deaminase is added to remove this endogenous competing ligand to determine receptor number in radioligand binding or G-protein activation experiments. 23Increased Bmaxin the lumbar region of nerve-injured compared with normal rats in the absence of adenosine deaminase in the current study is most consistent with a reduction in endogenous competing adenosine in the nerve-injured animals, allowing more binding sites available to the radioligand.

Adenosine depletion after spinal nerve ligation is supported by several observations: alleviation of allodynia by intrathecal injection of an adenosine reuptake inhibitor, 11reduced adenosine release by morphine in spinal cord from animals after spinal nerve ligation (preliminary observations), and reductions in CSF concentrations of adenosine in patients with chronic pain. 24Measuring resting intracellular adenosine concentrations or the half-life of increasing intracellular adenosine concentrations by bathing neurons with adenosine, such as would occur after intrathecal injection, would be difficult to study, given the rapid cycling of this purine nucleoside in transcription and energy processes. However, we propose that the selectivity of intrathecal adenosine to hypersensitivity states and its prolonged duration of action could share a common explanation: increased intracellular adenosine concentrations after intrathecal adenosine injection and restoration of depleted adenosine stores.

In summary, intrathecal adenosine disappears rapidly from the intrathecal CSF space in rats but has a prolonged duration of action, indicating that long residence time in CSF is not the cause of its prolonged effect. Radioligand binding studies are not consistent with up-regulation of spinal adenosine receptors as a cause for increased activity of intrathecal adenosine in hypersensitivity states. Rather, they support an association between depletion of spinal cord adenosine and allodynia in this model. We propose that intrathecal adenosine's effect is amplified in hypersensitivity conditions and is long lasting because it primarily acts to increase intracellular concentrations of adenosine in intrinsic spinal cord neurons.

Suzuki R, Stanfa LC, Kowaluk EA, Williams M, Jarvis MF, Dickenson AH: The effect of ABT-702, a novel adenosine kinase inhibitor, on the responses of spinal neurones following carrageenan inflammation and peripheral nerve injury. Br J Pharmacol 2001; 132: 1615–23
Sah DW: Neurotransmitter modulation of calcium current in rat spinal cord neurons. J Neurosci 1990; 10: 136–41
Cahill CM, White TD, Sawynok J: Substance P releases and augments the morphine-evoked release of adenosine from spinal cord. Brain Res 1997; 760: 294–7
Santicioli P, Del Bianco E, Maggi CA: Adenosine A1receptors mediate the presynaptic inhibition of calcitonin gene-related peptide release by adenosine in the rat spinal cord. Eur J Pharmacol 1993; 231: 139–42
Choca JI, Green RD, Proudfit HK: Adenosine A1and A2receptors of the substantia gelatinosa are located predominantly on intrinsic neurons: An autoradiography study. J Pharmacol Exp Ther 1988; 247: 757–64
Reeve AJ, Dickenson AH: The roles of spinal adenosine receptors in the control of acute and more persistent nociceptive responses of dorsal horn neurones in the anaesthetized rat. Br J Pharmacol 1995; 116: 2221–8
Sosnowski M, Stevens CW, Yaksh TL: Assessment of the role of A1/A2adenosine receptors mediating the purine antinociception, motor and autonomic function in the rat spinal cord. J Pharmacol Exp Ther 1989; 250: 915–22
Lee YW, Yaksh TL: Pharmacology of the spinal adenosine receptor which mediates the antiallodynic action of intrathecal adenosine agonists. J Pharmacol Exp Ther 1996; 277: 1642–8
Brundege JM, Dunwiddie TV: Role of adenosine as a modulator of synaptic activity in the central nervous system. Adv Pharmacol 1997; 39: 353–91
Arch JR, Newsholme EA: The control of the metabolism and the hormonal role of adenosine. Essays Biochem 1978; 14: 82–123
Lavand'homme PM, Eisenach JC: Exogenous and endogenous adenosine enhance the spinal antiallodynic effects of morphine in a rat model of neuropathic pain. Pain 1999; 80: 31–6
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
Chiari AI, Eisenach JC: Intrathecal adenosine: Interactions with spinal clonidine and neostigmine in rat models of acute nociception and postoperative hypersensitivity. A nesthesiology 1999; 90: 1413–21
Kim SH, Chung JM: An experimental model for peripheral neuropathy produced by segmental spinal nerve ligation in the rat. Pain 1992; 50: 355–63
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
Yaksh TL, Rudy TA: Chronic catheterization of the spinal subarachnoid space. Physiol Behav 1976; 7: 1032–6
Eisenach JC, Hood DD, Curry R: Preliminary efficacy assessment of intrathecal injection of an American formulation of adenosine in humans. A nesthesiology 2002; 96: 29–34
Shafer SL, Eisenach JC, Hood DD, Tong CY: Cerebrospinal fluid pharmacokinetics and pharmacodynamics of intrathecal neostigmine methylsulfate in humans. A nesthesiology 1998; 89: 1074–88
Sauernheimer C, Williams KM, Brune K, Geisslinger G: Application of microdialysis to the pharmacokinetics of analgesics: Problems with reduction of dialysis efficiency in vivo. J Pharmacol Toxicol Methods 1994; 32: 149–54
Gomes JA, Li XH, Pan HL, Eisenach JC: Intrathecal adenosine interacts with a spinal noradrenergic system to produce antinociception in nerve-injured rats. A nesthesiology 1999; 91: 1072–9
Sinclair CJ, Shepel PN, Geiger JD, Parkinson FE: Stimulation of nucleoside efflux and inhibition of adenosine kinase by A1 adenosine receptor activation. Biochem Pharmacol 2000; 59: 477–83
Sweeney MI: Adenosine release and uptake in cerebellar granule neurons both occur via an equilibrative nucleoside carrier that is modulated by G proteins. J Neurochem 1996; 67: 81–8
Moore RJ, Xiao R, Sim-Selley LJ, Childers SR: Agonist-stimulated [35S]GTPgammaS binding in brain modulation by endogenous adenosine. Neuropharmacology 2000; 39: 282–9
Guieu R, Peragut JC, Roussel P, Hassani H, Sampieri F, Bechis G, Gola R, Rochat H: Adenosine and neuropathic pain. Pain 1996; 68: 271–4