Adenosine analogs have been shown to produce antinociception after intrathecal administration. To determine the adenosine receptor subtype involved in spinal antinociception, the effects of selective agonists and an antagonist on the evoked potentials recorded from a neonatal rat spinal cord were studied. The measured potentials are a slow ventral root potential (slow VRP), which is the C-fiber-evoked excitatory response associated with nociceptive information; a monosynaptic reflex (MSR), which reflects a non-nociceptive transmission related to motor function; and a dorsal root potential (DRP), which reflects a gamma-aminobutyric acidA (GABA(A)) receptor-mediated presynaptic inhibition associated with analgesia.
The evoked potentials were recorded in response to electric stimulation of a lumbar dorsal root. Dose-response curves of agonists for these responses were obtained to determine their relative potency order. The antagonist dissociation constants (K(D) values) were estimated by Schild analysis.
Adenosine agonists dose dependently inhibited the slow VRP and the MSR. However, the slow VRP was five to eight times more sensitive to them than was the MSR. The rank order of agonist potency was N6-cyclohexyladenosine (CHA) = N6-(R)-phenylisopropyladenosine (R-PIA) > 5'-N-ethylcarboxamidoadenosine (NECA) > CGS 21680 in both responses. 8-Cyclopentyltheophylline produced dose-dependent parallel shifts to the right of NECA dose-response curves for these responses. Schild analysis gave linear plots with slopes near unity. The K(D) values of CPT for the MSR and the slow VRP were estimated to be 5.5 nM and 4.3 nM, respectively. The DRP was also depressed by adenosine agonists with potency order of CHA > NECA > CGS 21680. 8-Cyclopentyltheophylline antagonized the inhibitory effects of CHA on the DRP.
The results indicate that adenosine agonists inhibit spinal sensory transmission related to nociception by acting at the A1 receptors. The A1 receptor also seems to be involved in transmission related to the spinal motor system. Feedback inhibition mediated by GABA(A) receptors does not contribute to this antinociceptive action.
Key words: Receptors, adenosine, agonists: CGS 21680; N6-cyclohexyladenosine; 5'-N-ethylcarboxamidoadenosine; N6-(R)-phenylisopropyladenosine. Receptors, adenosine, antagonists: 8-cyclopentyltheophylline; theophylline. Spinal cord: nociception.
During the past decade, adenosine has been widely accepted as a neuromodulator in the peripheral and central nervous systems. In this regard, adenosine is considered to play an important role in modifying nociceptive information in the spinal cord. Several behavioral studies have shown that intrathecal administration of adenosine analogs produces antinociception. [3-5]In the central nervous system, two main types of receptors are believed to mediate the actions of adenosine: A1and A2types of adenosine receptors. They are differentiated by the rank order potency of adenosine agonists and the affinities for antagonists. [6,7]The A1and the A2receptors have been identified in the rat spinal cord using receptor autoradiography and receptor binding studies. [8,9]Studies with in vivo models of nociception have suggested that the A1receptor plays an important role in spinal antinociception. [10,11]Recently, the sub-classification has become more complex, with the acceptance of A sub 3 receptors and subtypes of A2receptors. [12-14]In addition to the lack of selective agonists and antagonists for those receptors, the effects of adenosine agonists on non-nociceptive transmission, such as impairment of motor function and locomotor depression, limited behavioral studies characterizing the type of adenosine receptors involved in antinociception and sometimes led to conflicting interpretations and conclusions. [3,4]
In response to the electric stimulation of a lumbar dorsal root, the neonatal rat spinal cord generates a slow ventral root potential (slow VRP) and a monosynaptic reflex (MSR)in the corresponding ipsilateral ventral root and a dorsal root potential (DRP) in the adjacent dorsal root. The slow VRP is a substance P and N-methyl-D-asparate (NMDA) receptor-mediated excitatory response of interneurons and motoneurons evoked by the activation of primary afferent C-fibers, whereas the MSR reflects a direct synaptic transmission between large-diameter primary afferents and alpha-motoneurons. Although primary afferent C-fibers are not exclusively nociceptive, many afferent fibers from nociceptors are included in this population. The DRP is a gamma-aminobutyric acidA(GABAA) receptor-mediated depolarization of primary afferent terminals, which reflects a feedback inhibition associated with analgesia. The slow VRP and the DRP have been shown to be related to the excitatory and the inhibitory transmission in nociceptive processing, respectively. [19-20]Therefore, examination of the actions of adenosine agonists on these responses and characterization of the adenosine receptors involved may provide further information on the mechanism of adenosine-induced antinociception.
We examined the effects of selective A1agonists, N6-cyclohexyladenosine (CHA) and N6(R)-phenylisopropyladenosine (R-PIA), and a less selective A1/A2agonist, 5'-N-ethylcarboxamidoadenosine (NECA), and a selective A2agonist, CGS 21680, on these evoked responses. In addition, a selective adenosine A1antagonist, 8-cyclopentyltheophylline (CPT), was tested alone and against the antinociceptive effects of agonist.
Materials and Methods
All protocols have been approved by the Hokkaido University School of Medicine Animal Care and Use Committee. Newborn Wistar rats (2-4 days old) of either sex were used in these experiments. They were anesthetized with halothane and decapitated. The vertebral column was rapidly removed and placed in a dish containing oxygenated artificial cerebrospinal fluid composed of 123 mM NaCl, 4 mM KCl, 2 mM CaCl2, 1.3 mM MgSO4, 26 mM NaHCO2, 1.2 mM NaHPO4, 10 mM glucose. After a dorsal laminectomy was performed, the spinal cord from the upper thoracic to the sacral level was dissected free with the spinal roots. The spinal cord preparation was mounted in a recording chamber and superfused at 2-3 ml/min with artificial cerebrospinal fluid bubbled with 95% oxygen and 5% carbon dioxide, maintaining a pH 7.3-7.4. The temperature was kept constant at 26-27 [degree sign] Celsius.
Stimulation and Recording
Suction electrodes were used for stimulation and recording. Stimuli were delivered to a lumbar dorsal root, and evoked potentials were recorded extracellulary from the corresponding ipsilateral ventral root or the adjacent dorsal root. Square wave pulses lasting 0.2 ms were used at more than 120-s intervals. The stimulus voltage adjusted to be supramaximal for evoking the slow VRP and the DRP ranged from 20-40 V. The intensity of the supramaximal stimuli for the MSR was 8-10 V. Potentials were displayed on an oscilloscope, digitized using an A-D converter (TEAC, Tokyo, Japan), and saved on the hard disk of a computer (Hewlett Packard, Palo Alto, CA) for subsequent analysis. The preparation was allowed to rest for about 90 min before experimentation began, when the evoked potential had fully recovered from surgical stress and stable responses could be evoked for at least 3 h. After three stable control responses (less than 10% variation) were obtained, drugs diluted to the desired concentrations in artificial cerebrospinal fluid were applied to the spinal cord for more than 15 min, when the drug effects reached steady state and did not change thereafter.
The positive peak amplitudes of the MSR and the DRP above baseline were measured, and the area under the curve of the slow VRP was calculated from stored records. Drug effects were expressed as a percentage of inhibition of the mean of control responses. Cumulative dose-response curves for adenosine agonists were obtained. Median inhibitory concentration (IC50) values were determined by the standard Hill plot, including only those concentrations that inhibit 10-90% of control values. The antagonist dissociation constant (KDvalue) was estimated using a Schild plot. Briefly, dose-response curves of NECA were generated in the absence and presence of a fixed concentration of an antagonist. The ratios of the IC50values (DR) in the presence and absence of the antagonist were used in the Schild equation log (DR - 1) = n log [B] - log KD, where [B] is the molar concentration of NECA and n is the slope of the regression of log (DR - 1) on log [B]. The pA2values were determined on the abscissae intercept of the regression line. Where the regression log (DR-1) line fitted by the least-squares method had a slope approximated to unity, the KDvalue was taken as the negative antilog of the pA sub 2 value.
Data are expressed as mean +/- SEM, and n represents the number of experiments. The statistical significance of drug effects was assessed using a Student's paired t test. The IC50values between agonists were compared using analysis of variance with Scheffe's F test. P values < 0.05 were considered significant.
Drugs used were obtained from the following sources: CHA, R-PIA, NECA, CGS 21680, and CPT from Research Biochemicals (Natick, MA); and theophylline from Wako Pure Chemical (Osaka, Japan).
Electric stimulation of the dorsal root evoked characteristic potentials consisting of a short-latency MSR and a long-lasting slow VRP in the corresponding ventral root, and a short-duration, low-amplitude DRP in the adjacent dorsal root, as previously reported.
Effects of Adenosine Agonists on the Monosynaptic Reflex and the Slow Ventral Root Potential
Selective A1agonists CHA and R-PIA, and a less selective A1/A2agonist, NECA, produced dose-dependent inhibition of the MSR and the slow VRP (Figure 1and Figure 2). CHA and R-PIA were more potent than NECA in their inhibitory effects on both responses. All of these agonists more potently inhibited the slow VRP than the MSR. As shown in Table 1, IC50values of these agonists for the MSR were approximately five to eight times greater than those for the slow VRP. In contrast, a selective A2agonist CGS 21680 had a weak depressant effect on the slow VRP, with no effect on the MSR up to 10 micro Meter (Figure 2). The rank order of potency of adenosine agonists was CHA = R-PIA > NECA much > CGS 21680 in the MSR and the slow VRP. (Here and elsewhere, = indicates less than 2-fold difference in IC sub 50, > indicates a 2- to 10-fold difference, and much > indicates greater than 10-fold difference; Table 1). The effects of cumulative doses of CHA, R-PIA, and NECA were not fully reversed even after 60 min of washing but were reversed to the control levels within 10 min by applying 25 micro Meter theophylline.
Effects of an Adenosine Antagonist on the Monosynaptic Reflex and the Slow Ventral Root Potential
A selective A1antagonist CPT was tested alone and against the effects of NECA on the MSR and the slow VRP. CPT did not produce any significant effect of its own on the MSR or the slow VRP in the range of 1 nM to 10 micro Meter (data not shown). However, CPT at concentrations of 10 nM, 30 nM, and 100 nM produced dose-dependent parallel shifts to the right of the NECA dose-response curves for the MSR and for the slow VRP (Figure 3(A and B)). Schild analysis for the MSR and the slow VRP gave linear plots (r = 0.998 and r = 0.999) with slopes of 1.045 +/- 0.066 and 0.948 +/- 0.018, respectively (Figure 3(C)). The pA2 values were calculated as 8.26 and 8.37, equivalent to K sub D, values of 5.5 nM for the MSR and 4.3 nM for the slow VRP, respectively.
Effects of Adenosine Agonists on the Dorsal Root Potential
To know whether adenosine receptors are involved in GABAA-mediated presynaptic inhibition, effects of CHA, NECA, and CGS 21680 on the DRP were studied. CHA and NECA inhibited the DRP in a dose-dependent manner, with IC50values of 48.5 +/- 7.1 nM and 199 +/- 16 nM, respectively (Figure 4). Enhancement of the DRP amplitude by these drugs was not observed at any concentration. In contrast, CGS 21680 did not have any significant effect on the DRP in the range of 1 nM to 3 micro Meter. The agonist order of inhibitory effects on the DRP was CHA > NECA much > CGS 21680. The effects of cumulative doses of CHA and NECA were fully reversed by the application of theophylline 25 micro Meter. The effects of CHA were also studied in the presence of CPT. CPT at a concentration of 100 nM, which had no significant effect of its own on the DRP, shifted to the right the dose-response curve of CHA in a parallel way (Figure 4). The IC50, value of CHA in the presence of CPT was 2,232 +/- 799 nM.
Our results show that adenosine agonists inhibit sensory transmission related to nociceptive information at the spinal level and that the A1receptors are responsible for this effect. The rank order potencies of adenosine agonists in inhibiting the MSR and the slow VRP were highly consistent with the A1receptor order of CHA = R-PIA > NECA much > CGS 21680 (Table 2). In addition, a selective antagonist, CPT, competitively antagonized the inhibitory effects of NECA on the MSR and the slow VRP, with KDvalues of 5.5 nM and 4.3 nM, respectively. These values are close to that found by Bruns et al., who measured the displacement by CPT of [sup 3 H]NECA binding to A1receptors in rat brain membranes (10.9 nM; Table 2). The involvement of the A2receptors in these responses was discounted based on agonist order because CGS 21680, the most potent agonist at A sub 2 receptors, was much less potent than the other agonists. These results extend those of previous reports. Behavioral studies have shown that the antinociceptive activities of various adenosine agonists correlate with their affinities for the A1receptors. [3,10,23]The A1agonists have been shown to inhibit the C-fiber evoked activity of dorsal horn nociceptive neurons in the spinal cord. To characterize fully the receptor type involved, it is necessary to determine not only agonist potency order but also antagonist affinities with a range of antagonist doses; however, such studies have not been done in in vivo models of nociception.
The slow VRP is linked to nociception and selectively depressed by known analgesics, such as morphine and clonidine. This response is evoked by C-fiber activation and mediated by substance P and the NMDA-mediated polysynaptic pathway. On the other hand, the MSR is related to motor function, such as locomotion and posture, rather than to nociception. The MSR is an excitatory response of motoneurons evoked by the activation of large-diameter primary afferents. In the present study, high doses of adenosine agonists inhibited both responses by activating the A1receptors, suggesting that the A1receptors are involved in non-nociceptive and nociceptive neurotransmission. Most behavioral studies have reported some degree of motor impairment after intrathecal injection of adenosine agonists, [4,5,23]although they did not indicate the receptor type involved. Considering the involvement of the A1receptors in the inhibition of the MSR, it is likely that the inhibitory effects on spinal motor system also are mediated by the A1receptors.
Adenosine agonists were approximately five to eight times selective for inhibition of the slow VRP compared with the MSR, indicating that nociceptive transmission is more sensitive to adenosine agonists than is non-nociceptive transmission. Several lines of evidence support these findings. Autoradiographic studies show that the substantia gelatinosa contains the highest density of adenosine receptors in the spinal cord. The substantia gelatinosa serves as an important relay of nociceptive transmission. Reeve and Dickenson showed that adenosine agonists more effectively inhibited NMDA-mediated events such as postdischarge, wind-up, and formalin-induced response of nociceptive neurons than acute electrically evoked response in the spinal cord. Further, adenosine has been suggested to have presynaptic and postsynaptic effects on synaptic transmission from dorsal root fibers. Thus, the potent effect of adenosine agonists on polysynaptically evoked responses such as slow VRP and DRP may represent the sum of combined presynaptic and postsynaptic actions in a sequence of neurons.
The inhibitory effects of adenosine agonists were almost completely reversed by a prototypic adenosine receptor antagonist, theophylline. However, theophylline is not specific for subtypes (Table 2) and also has additional effects, such as releasing intracellular Ca sup 2+ and inhibiting phosphodiesterase at higher doses. Thus a potent and selective xanthine, CPT was chosen to antagonize the dose-response effects of NECA. On its own, CPT had no significant effect on the MSR or the slow VRP, indicating that endogenous adenosine is not involved in modulating these responses. In behavioral studies, some investigators have reported hyperalgesia after intrathecal injection of theophylline or other xanthines, but others have not observed any effect. [23,27]Further studies would be necessary to determine whether tonic release of adenosine regulates nociception.
CPT competitively antagonized the effects of NECA on the MSR and the slow VRP. The Schild analysis for these responses gave slopes that were not different from unity, suggesting that NECA was acting on one receptor type. Because the KDvalues of CPT (5.5 nM and 4.3 nM) were close to the reported Kivalue for the A1receptors (10.9 nM), this confirms that NECA was acting via the A1receptors. Recently, the existence of A2receptor subclasses (A2a, which is usually called the A2receptor, and A2b) was proposed. [13,14]There is little possibility of the involvement of the A2areceptors because CGS 21680, the most potent selective A2aagonist, was much less potent than the other agonists. Considering that the A2breceptors are found widely distributed throughout the central nervous system, the consequences of the activation of the A2breceptors in the spinal cord deserve further exploration. However, the lack of selective agonists and antagonists for the A2breceptors hampers further studies. The possibility of the involvement of the A3receptors in these responses was also discounted because two of the xanthine derivatives used (theophylline and CPT) antagonized these effects, and the A3receptors have been reported to be insensitive to xanthine antagonists at doses many times greater than those used in these experiments. 
The mechanism by which adenosine and its analogs produce antinociception in the spinal cord is not fully understood. Generally they have inhibitory actions on neurotransmission in the central nervous system, which are mainly due to the activation of the A1receptors. [1,29,30]At the cellular level, activation of the A1receptors are associated with inhibition of transmitter release [31-33]and hyperpolarization of postsynaptic membrane mediated by increases in K sup + -conductance. [25,31,34]Of note, a recent study using intracellular recordings from dorsal horn neurons showed that inhibitory postsynaptic potentials evoked by low-threshold mechanical stimulation are mediated by two different components; GABAAreceptors and adenosine receptors. [26,35]Adenosine-mediated inhibitory postsynaptic potentials are mediated by increased K sup + conductance, whereas GABAA-mediated inhibitory postsynaptic potentials are mediated by increased Cl conductance. Thus it is possible that not only GABA but also adenosine might play a role as a feedback inhibitory transmitter in nociceptive transmission in the spinal cord. In this regard, we were interested in the effects of adenosine agonists on the DRP. The DRP is a GABAA-mediated depolarization of primary afferent terminals, which reduces the effectiveness of incoming sensory stimuli. The DRP thus represents a form of negative feedback causing presynaptic inhibition. Drugs such as barbiturates and benzodiazepines that act on GABAAreceptors have been shown to enhance the DRP. [18,20]We expected the enhancement of the DRP by adenosine agonists because of its association with analgesia. In our study, however, adenosine agonists did not enhance the DRP but produced dose-dependent inhibition. The agonist potency order of CHA > NECA much > CGS 21680 and the antagonism of the inhibitory effects of CHA by CPT suggest the involvement of the A1receptors. The pathway of the DRP includes glutamate NMDA or non-NMDA receptor-mediated interneurons, which release GABA onto primary afferents. Inhibition of glutamate release by adenosine has been shown in various brain regions, such as the cerebellum, hippocampus, and olfactory cortex. Uchiyama and North showed that adenosine equally inhibited the release of GABA as well as glutamate in the rat nucleus accumbens. Thus it is likely that the antinociceptive actions of adenosine and its analogs are mainly due to the inhibition of excitatory neuron such as substance P or NMDA receptor-mediated pathway, and are not due to the enhancement of GABAA-mediated feedback inhibition.
In conclusion, the present results indicate that adenosine agonists potently inhibit sensory transmission related to nociception at the spinal level. However, they also produce depressant effects on GABA sub A-mediated feedback inhibition and transmission related to motor function at high doses. The agonist order and the antagonist affinity suggest that all of these effects are brought about by the activation of the A1receptors. The antinociceptive action seems to be mediated by mechanisms other than the GABAA-mediated inhibitory system.
The authors thank Mervyn Maze, M.B., Ch.B., Department of Anesthesiology, Stanford University School of Medicine, for helpful comments on this manuscript; and Mitsue Azuma, Ph.D., and Masakiyo Ishikawa, for technical assistance.