Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Spinal dynorphin A(1-17) (Dyn) produces antianalgesia, wherein the analgesic action of morphine is attenuated in the mouse tail-flick test. Dyn administered intrathecally (i.t.) at a very small dose (5 fmol) or released endogenously in the spinal cord of the mouse activates an ascending neuronal circuit to the brain where benzodiazepine receptors are stimulated as evidenced by elimination of Dyn-induced antianalgesia by spinal transection and i.c.v. administration of flumazenil, a central benzodiazepine receptor antagonist (Wang et al., 1994; Rady et al., 1998a). Then, through a descending neuronal circuit the stimulus activates spinal cholecystokinin (CCK), leu-enkephalin (LE), and N-methyl-D-aspartate (NMDA) receptors in linear sequence (Rady et al., 2001b) as illustrated in Fig. 1 to produce antianalgesia. Among the many workers who have studied the action of spinal CCK in nociception, Watkins and colleagues linked peripheral interleukin (IL)₁ to spinal CCK release. First, systemic administration of lipopolysaccharide (LPS) stimulates peripheral afferent nerves in the vagus to the brain to activate a descending neural circuit to produce spinal CCK and NMDA receptor actions (Watkins et al., 1994a). Second, lesioning of the dorsolateral funiculus eliminates the descending effect (Watkins et al., 1994b). Third, systemically administered LPS produces a peripheral IL₁-mediated "illness"-induced hyperalgesia (Maier et al., 1993). The commonality of spinal CCK involvement in Dyn-induced antianalgesia and LPS-/IL₁-induced hyperalgesia made us wonder whether Dyn action was mediated by IL₁. Laughlin et al. (2000) have demonstrated that spinal cord damage produced by pretreatment with a large dose of Dyn i.t. releases IL₁ in the spinal cord that is responsible for hyperalgesic and allodynic responses. However, i.c.v. administration of IL₁ in rats has been shown to produce hyperalgesia (Yabuuchi et al., 1996; Oka and Hori, 1999). Thus, the purpose of our present study was to determine whether the antianalgesic action of spinal Dyn in mice was mediated by IL₁ in the brain. This Dyn action would not involve excitation of peripheral afferent nerves because the Dyn is given intrathecally but involves an ascending neuronal circuit from the spinal cord to the brain. A unique aspect of the present study compared with those described above for IL₁ is that spinal Dyn through central ascending neuronal action would be proposed to activate an IL₁ response in the brain, a site remote from the point of administration of Dyn. In this experimental design, Dyn was given i.t. and the mediation by IL₁ was assessed by administration of various antagonists (IL₁ antiserum, Lys¹⁹³-D-Pro-Thr¹⁹⁵) i.c.v. Lys-D-Pro-Thr is a tripeptide analog of IL₁ that antagonizes certain but not all actions of IL₁ (Ferriera et al., 1988; Poole et al., 1999). Also, an inbred strain of mice, C3H/HeJ, was evaluated for Dyn and IL₁ responsiveness. C3H/HeJ mice have a mutation in the LPS gene, LPS^d, which creates a deficiency in the release of IL₁ in the brain in response to LPS (Doolittle et al., 1995; Gabellec et al., 1995; Johnson et al., 1997). If i.t. Dyn-induced antianalgesia is mediated by IL₁ in the brain, C3H/HeJ mice should not give an antianalgesic response to Dyn.

View larger version (23K): [in this window] [in a new window]   Fig. 1.   Antianalgesic action of i.t. Dyn against i.t. morphine: known components (illustration as modified from Rady et al., 2001) along with new steps (2 and 3) proposed for this study. Previous work showed that the spinal action of Dyn (step 1) activates a neuronal pathway to the brain where benzodiazepine receptors (which are inhibited by flumazenil, a central type benzodiazepine receptor antagonist) are activated and lead to release in the spinal cord of CCK and the consequent sequence (steps 4-6). LE has a 2 opioid receptor inverse agonist action. The sequence of these steps is based on the selective inhibitory action of the antagonists enclosed in rectangles at each step. The present work proposes that release of IL1 is involved in step 2 and follows the benzodiazepine receptor step, and IL1 action is linked to the linear sequence of steps 4 to 6. The antianalgesic action of IL1 is shown to be inhibited by Lys-D-Pro-Thr, an IL1 antagonist, and IL1 antiserum and PK11195, a peripheral type benzodiazepine receptor antagonist but not by flumazenil. C3H/HeJ mice, which have a deficiency in release of IL1, are used to reinforce the role for the IL1.

As mentioned above, spinal Dyn acting through an ascending circuit activates benzodiazepine receptors in the brain; flumazenil, a benzodiazepine receptor antagonist, eliminates this antianalgesic action (Rady et al., 1998a). The antianalgesic action of i.c.v. pentobarbital and neurotensin is also inhibited by i.c.v. flumazenil (Wang and Fujimoto, 1993; Wang et al., 1994; Holmes et al., 1999; Rady et al., 2001a). Thus, another aspect of the present study was to determine whether the benzodiazepine receptor and IL₁ responses could be shown to be separable steps. Flumazenil, Lys-D-Pro-Thr, and IL₁ antiserum act by different mechanisms and their use might selectively inhibit one but not the other step under investigation. Another consideration was whether an additional type of benzodiazepine receptor, the peripheral type, might be involved. Historically, the peripheral type of receptor was defined by the observation that they occurred in peripheral tissues; but, they are present in the central nervous system as well (Gavish et al., 1999). Relatively selective inhibition of peripheral and central type benzodiazepine receptors can be achieved by PK11195 and flumazenil, respectively, and their use in the present study demonstrated involvement of both types of benzodiazepine receptors in separable steps in the sequence of actions in the brain to produce antianalgesia.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Animals. Male CD-1 mice weighing 25 to 30 g were purchased from Charles River (Wilmington, MA). In one later set of experiments, 9-week-old male inbred C3H/HeJ mice, which have a mutant LPS gene to produce a deficiency in release of IL₁ by LPS, and C3H/HeOuJ mice, the normal counterpart to the C3H/HeJ mice, were obtained from Jackson Laboratories (Bar Harbor, ME). The mice were housed up to five per cage under 12-h light/dark cycle with food and water available ad libitum, including during the experiment. They were handled daily but not used for at least 2 days after arrival. Experiments were performed at the same time on each experimental day generally from 8:00 AM to 12:00 PM, but no extra care was given to minimize stress or diurnal fluctuations known to occur to IL₁ (Vitkovic et al., 2000). Each animal was used for only one experiment. In the single-dose experiments (in contrast with the ED₅₀ experiments below) with CD-1 mice 7 to 10 animals were used for each group; one exception was the dose range study for Lys-D-Pro-Thr where four to five animals were used per group for the preliminary study, but the experiment with the effective dose was replicated later using a larger number of animals. For the inbred C3H/HeJ and C3H/HeOuJ studies, usually five to six (with an exception of 10 for i.c.v. nociceptin; Fig. 5F) animals were used per group because of their high cost.

Measurement of Antinociception. The radiant heat tail-flick test was used to measure the antinociceptive response to morphine. This involved holding the mouse by hand with a towel draped over it while the tail was placed in the test apparatus. Immediately before the start of the experiment, the predrug control tail-flick latency (the average of two trials) of 2 to 4 s was obtained. Drugs were administered as described below and the latency redetermined as the postdrug latency. An automatic cut-off time set at 10 s prevented trauma to the tail and was considered to be the maximum response time. The percentage of maximum possible effect (% MPE) for each mouse was calculated from the tail-flick latencies as follows:
The mean % MPE values were determined for groups treated with i.t. morphine to produce analgesia and various antianalgesic agents by i.t. or i.c.v. routes along with agents to modify their actions (antagonists, antiserum).

Drug Administration. Drugs were administered free-hand i.t. between the 5th and 6th lumbar vertebrae by using a 30-gauge needle attached to a 50-µl Hamilton syringe by the method of Hylden and Wilcox (1980) in a volume of 5 µl generally 5 min (unless stated otherwise) before the postdrug tail-flick test. Other drugs were administered free-hand i.c.v. by using a 22-gauge removable needle attached to a 25-µl Hamilton syringe by the method of Haley and McCormick (1957) in a volume of 4 µl under light halothane anesthesia 10 min (unless stated otherwise) before the postdrug tail-flick test. The momentary halothane anesthesia did not last into the time when the tail-flick test was performed, and there was no residual effect of the halothane on the morphine-induced analgesic response. Peptides (Dyn, CCK8s, nociceptin, neurotensin, Lys-D-Pro-Thr) were dissolved in a 0.9% saline solution containing 0.01% Triton X-100. Nonpeptide drugs were dissolved in a 0.9% saline solution. Control serum, IL₁ antiserum, and CCK antiserum were diluted with 0.9% saline. When drugs were given together at the same site and time, the solutions were premixed so that the drugs were given in a single administration. As previously published (Rady et al., 1994, 1998a,b, 2001a,b) near maximal drug effects were obtained at 10 and 5 min for i.c.v. and i.t. administration, respectively. Administration of antiserum for IL₁ as a pretreatment for 1 h was successful as had been used for other antisera (Rady et al., 2001b). The doses and times of administration of drugs are given with each experiment. Appropriate vehicle solutions were administered to control groups to parallel each drug treatment. All studies were done in compliance with the Institutional Animal Care and Use Committee (Animal Studies Subcommittee).

ED₅₀ Determination for i.t. Morphine as Affected by i.t. Dyn and i.c.v. IL₁ Given Alone or with i.c.v. IL₁ Antiserum. In one set of experiments, dose-response relationships for i.t. morphine antinociception as affected by various treatments were established using the quantal response to morphine. Tail-flick latencies greater than 3 standard deviations over the mean predrug time were considered to be antinociceptive. At each dose of morphine the mice showing antinociception were expressed as a percentage of the number of animals (usually eight but in a few cases seven) in that group. Three or more dose levels were used to construct each of the dose-response curves. The percent responding values were transformed to probit values and plotted against the log of the morphine dose then ED₅₀ values with 95% confidence intervals were derived and comparisons of the slope and ED₅₀ values were made (Litchfield and Wilcoxon, 1949).

Statistical Analysis. Analyses of the single-dose results involving a comparison between the mean % MPE for two groups were by Student's t test. Those involving more than two groups were evaluated by analysis of variance and comparison of all the groups to each other by Newman-Keuls test or comparison of one control to several treatment groups by Dunnett's test. The results for only the main comparisons are given. A P  0.05 was taken to indicate a significant difference between groups in all analyses.

Source of Drugs. The drugs used and their sources were as follows: morphine sulfate · 5H₂O (Mallinckrodt, St. Louis, MO); Dyn and neurotensin (Peninsula Laboratories, Belmont, CA); nociceptin (Bachem, Torrence, CA); IL₁ and IL₁ antiserum (R & D Systems, Minneapolis, MN); Lys-D-Pro-Thr, MK801, PK11195, lipopolysaccharide, and pentobarbital sodium (Sigma, St. Louis, MO); CCK8 antiserum (Chemicon International, Temecula, CA); naltriben methane sulfate (Sigma/RBI, Natick, MA); and 7-benzylidenenaltrexone (BNTX) (Tocris Cookson, Ballwin, MO). The flumazenil was supplied by Hoffman-La Roche (New York, NY). The doses of the drugs, given with the experiments, were for the forms stated above.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Antianalgesic Action of i.c.v. IL₁. The results in Fig. 2A showed that the antinociceptive action of i.t morphine was attenuated by the i.c.v. administration of 5 to 10 ng of IL₁ but the effect was no greater at 20 ng. In Fig. 2B, the antianalgesic action for the 10-ng dose of IL₁ had a quick onset and lasted between 20 and 30 min.

View larger version (34K): [in this window] [in a new window]   Fig. 2.   Analgesic action of morphine (given as % MPE) and the antianalgesic action of IL1 in CD-1 mice. A, dose-response effect for IL1. The morphine was given i.t. 5 min before the tail-flick test. The IL1 was given i.c.v. at the stated dose 5 min before the tail-flick test. The administration of the drug vehicle is indicated by the "v" and the number of mice in each group is given at the top of the bar along with the S.E. The asterisk indicates a significant difference (P  0.05) by Dunnett's test. B, duration of action of IL1. The 10-ng dose of IL1 was given at the designated times before the tail-flick test, whereas the time of administration of the morphine was kept fixed at 5 min. The asterisk indicates significant difference (P  0.05) by Student's t test.

Attenuation of Antianalgesia by i.c.v. Lys-D-Pro-Thr. In Fig. 3A, Lys-D-Pro-Thr, the presumptive IL₁ antagonist (Ferriera et al., 1988; Poole et al., 1999), given i.c.v. along with IL₁ inhibited the antianalgesic action. The results in Fig. 3B indicated that Lys-D-Pro-Thr also eliminated the antianalgesic action of i.t. Dyn. The next study used Lys-D-Pro-Thr to see whether it had any effect on the antianalgesic action of i.c.v. pentobarbital and neurotensin, which activate the same descending pathway as Dyn (Rady et al., 1998b, 2001b; Holmes et al., 1999) as well as i.c.v. nociceptin, which activates a different descending spinal prostaglandin-mediated pathway (Rady et al., 2001a). The i.t. Dyn result is shown again in Fig. 4A. In Fig. 4B administration of pentobarbital i.c.v. had the expected antianalgesic action against i.t. morphine (Wang and Fujimoto, 1993; Rady et al., 1998b). Lys-D-Pro-Thr given i.c.v. eliminated this antianalgesic action. The antianalgesic action of neurotensin was likewise attenuated by i.c.v. Lys-D-Pro-Thr (Fig. 4C). Lys-D-Pro-Thr was not effective against the antianalgesic action of CCK8s (Fig. 4D). The antianalgesic action of i.c.v. nociceptin, as expected, was not affected by i.c.v. Lys-D-Pro-Thr (Fig. 4E).

View larger version (30K): [in this window] [in a new window]   Fig. 3.   Ability of Lys-D-Pro-Thr to antagonize the antianalgesic action of IL1 and Dyn in CD-1 mice. A, antianalgesic action of i.c.v. IL1 was eliminated by Lys-D-Pro-Thr given with the IL1. B, antianalgesic action of i.t. Dyn was eliminated by Lys-D-Pro-Thr in a dose-dependent manner. Note that Dyn and Lys-D-Pro-Thr were given at different sites. In both panels, the asterisk indicates significant difference (P  0.05) by Dunnett's test.

View larger version (36K): [in this window] [in a new window]   Fig. 4.   Antianalgesic action of various agents and the effect of Lys-D-Pro-Thr in CD-1 mice. A, Dyn given i.t. produced antianalgesia that was antagonized by Lys-D-Pro-Thr given i.c.v. Antianalgesic actions of i.c.v. pentobarbital (B) and neurotensin (C) were eliminated by Lys-D-Pro-Thr. The antianalgesic action of i.t. CCK8s (D) and i.c.v. nociceptin (E) were not affected by Lys-D-Pro-Thr. The asterisk indicates significant difference (P  0.05) from groups without asterisk by Newman-Keuls test.

Antianalgesic Action of Various Agents in C3H/HeJ and C3H/HeOuJ Mice. C3H/HeJ mice have a mutation in the LPS gene, LPS^d, which creates a deficiency in the release of IL₁ in the brain in response to LPS stimulation (Gabellec et al., 1995; Johnson et al., 1997). Administration of i.t. Dyn, i.c.v. pentobarbital, and i.c.v. neurotensin did not produce antianalgesia against morphine in these mice, consistent with the gene mutation. On the other hand, the antianalgesic response to i.c.v. IL₁ was present (Fig. 5D), indicating that the IL receptor was present. The antianalgesic actions of i.t. CCK8s (Fig. 5E) and i.c.v. nociceptin (Fig. 5F) were also present in C3H/HeJ and did not depend on IL₁ release in the brain.

View larger version (56K): [in this window] [in a new window]   Fig. 5.   Testing for antianalgesic responses in C3H/HeJ mice that have a deficiency in release of IL1 because of a mutant LPS gene. No antianalgesic response occurred for Dyn i.t. (A), pentobarbital i.c.v. (B), and neurotensin i.c.v. (C). On the other hand, antianalgesic responses were obtained for IL1 i.c.v. (D), CCK8s i.t. (E), and nociceptin i.c.v. (F). The asterisk indicates significant difference (P  0.05) by Student's t test.

View larger version (28K): [in this window] [in a new window]   Fig. 6.   Testing for antianalgesic responses in C3H/HeOuJ mice that have a normal LPS gene. Antianalgesic responses were found for Dyn i.t. (A), pentobarbital i.c.v. (B), and neurotensin i.c.v. (C) and eliminated by Lys-D-Pro-Thr i.c.v. The asterisk indicated significant difference (P  0.05) from groups without asterisk by Newman-Keuls test.

Antianalgesic Response Attenuated by i.c.v. Administration of IL₁ Antiserum. Results in Table 1 showed that the antianalgesic action of i.c.v. IL₁ produced a significant increase in the ED₅₀ of i.t. morphine. This increase was attenuated by a 1-h pretreatment with i.c.v. IL₁ antibody. Similarly, i.t. Dyn increased the ED₅₀ value of i.t morphine and this increase was also eliminated by i.c.v. administration of IL₁ antiserum. All the dose-response curves for morphine were parallel and indicated homogeneous effects of the treatments over the whole dose range of morphine.

View this table: [in this window] [in a new window]   TABLE 1 ED50 (95% CI) values for i.t. morphine in the tail flick test in CD-1 mice Morphine was given i.t. at four or more dose levels to all groups (n = 7-8 mice) 5 min before the tail flick test in control mice or those treated with IL1 (i.c.v., 10 ng, 5 min) or Dyn (i.t., 10 pg, 5 min) alone or along with IL1 antibody (i.c.v., 4 ng, 1 h).

Linking IL₁ Action to Spinal Antianalgesic Pathway. The steps (4-6) and the sequence of the spinal components for antianalgesic action are presented in Fig. 1. The need now was to establish a link between the antianalgesic action of IL₁ in the brain to the steps in the spinal cord. The IL₁ step might connect to the spinal sequence at any point. Figure 7A shows that the antianalgesic action of IL₁ was inhibited by i.t. administration of MK801, a noncompetitive NMDA receptor antagonist, indicating that the antianalgesic response produced by IL₁ acting in the brain was mediated at the spinal cord level by NMDA receptors. The antianalgesic action of IL₁ was also eliminated by i.t. naltriben, a ₂ opioid receptor antagonist, but not by i.t. BNTX, a ₁ opioid receptor antagonist (Fig. 7B). This latter specificity coincides with the inverse agonist action step of leu-enkephalin, a step that precedes the NMDA step (that is not sensitive to naltriben). Furthermore, the antianalgesic action of IL₁ was attenuated by CCK antiserum given as a pretreatment 1 h before the tail-flick test (Fig. 7C). Thus, steps 3 to 6 pictured in Fig. 1 were involved in IL₁ action and the IL₁ entered the sequence at the CCK point.

View larger version (28K): [in this window] [in a new window]   Fig. 7.   Determination of the point of linkage of IL1 to the linear sequence of the antianalgesic components in the spinal cord of CD-1 mice. The antianalgesic action of IL1 i.c.v. was inhibited by MK801 i.t. (A), naltriben i.t. but not BNTX i.t. (B), and 1-h pretreatment with CCK antiserum i.t. (C). Because of the linear sequence defined by earlier work (Rady et al., 2001b), this present result indicates that the antianalgesic action of IL1 made a link at the CCK step. The asterisk indicates significant difference (P  0.05) from groups without asterisk by Newman-Keuls test.

Studies Implicating Central and Peripheral Types of Benzodiazepine Receptors in Brain. Flumazenil given i.c.v. at a dose previously shown to inhibit i.t. Dyn and i.c.v. pentobarbital and neurotensin antianalgesia (Wang and Fujimoto, 1993; Rady et al., 1998a) did not inhibit the antianalgesic action of i.c.v. IL₁ (Fig. 8A). Administration of LPS i.c.v. inhibited morphine-induced analgesia (Fig. 8B), suggesting an antianalgesic action that is consistent with its ability to release IL₁ in the brain (Gabellec et al., 1995). This LPS-induced antianalgesia, like that of IL₁, was not inhibited by flumazenil. An initial experiment determined the time course of antianalgesic action of LPS. LPS (10 ng) was given i.c.v. at 1 to 4 h before the i.t. morphine analgesia tested at 5 min. LPS had no effect at 1 h, a small but significant antianalgesic effect at 1.5 h, a maximal effect at 2 h (50% reduction in MPE), and less but still significant effect of 4 h (data not shown).

View larger version (33K): [in this window] [in a new window]   Fig. 8.   Lack of involvement of the central type of benzodiazepine receptor in the antianalgesic action of i.c.v. IL1 and LPS as indicated by absence of effect of flumazenil i.c.v. IL1 (A) and LPS (B) were given i.c.v., 10 min and 2 h, respectively, before the tail-flick test and flumazenil was given i.c.v. 10 min before the tail-flick test. The asterisk indicates a significant difference (P  0.05) from other groups not similarly marked by Newman-Keuls test.

1

View larger version (30K): [in this window] [in a new window]   Fig. 9.   Involvement of peripheral type benzodiazepine receptors in i.t. Dyn- and i.c.v. IL1-induced antianalgesia in CD-1 mice. A, peripheral type benzodiazepine receptor antagonist PK11195 given i.c.v. 10 min before the tail-flick test eliminated i.t. Dyn-induced antianalgesia. (B) PK11195 likewise eliminated the antianalgesia induced by IL1 given i.c.v. 10 min before the tail-flick test. The asterisk indicates significant difference (P  0.05) from groups without asterisk by Newman-Keuls test.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The present results implicate a role for IL₁ in the brain to produce antianalgesia as follows. 1) The IL₁ antagonist Lys-D-Pro-Thr inhibited the antianalgesic action of i.t. Dyn, i.c.v. pentobarbital, and i.c.v. neurotensin. 2) The i.c.v. administration of IL₁ produced antianalgesia. 3) The antianalgesic action of i.c.v. IL₁ and i.t. Dyn were eliminated by i.c.v. pretreatment with IL₁ antiserum. 4) In C3H/HeJ mice, which are deficient in release of IL₁ in the brain (Gabellec et al., 1995; Johnson et al., 1997) because of a mutant LPS gene (Doolittle et al., 1995), i.t. Dyn, i.c.v. pentobarbital, and i.c.v. neurotensin did not produce antianalgesia, whereas i.c.v. IL₁ still produced antianalgesia. 5) The antianalgesic action of IL₁ involves activation of spinal CCK, ₂, and NMDA receptors as does Dyn, pentobarbital, and neurotensin.

Flumazenil did not inhibit the antianalgesic action of i.c.v. IL₁, which placed the IL₁ step after the flumazenil-sensitive benzodiazepine receptor step (Fig. 1, step 2). Likewise, the antianalgesic action of i.c.v. LPS was not affected by i.c.v. flumazenil, indicating that the ability of LPS to release IL₁ was not affected by flumazenil. This lack of sensitivity to flumazenil indicated that LPS-induced release of IL1 had a separate mode of action from that through the central benzodiazepine receptor.

A 10-ng dose of IL₁ given i.c.v. produces hyperalgesia in the mouse (Gul et al., 2000). Thus, it is possible that a hyperalgesic action produces a functional antagonism of the analgesic action of i.t. morphine. No attempt was made in the present study to measure hyperalgesia to IL₁ because the tail-flick test parameters as set on the apparatus are not able to quantify hyperalgesia. There is no reduction in the tail-flick latency in controls. Thus, the caveat exists that what is called antianalgesia here might be a functional effect of hyperalgesia after IL₁. In the present study, the i.c.v. Lys-D-Pro-Thr antagonized the antianalgesic action of IL₁ as it does the hyperalgesic action (Ferriera et al., 1988). Also, the resistance to morphine analgesia produced in streptozotocin diabetic mice has been proposed to be due to increased IL₁ in the brain (Gul et al., 2000). To complicate matters, there appears to be a difference in the response to i.c.v. IL₁: in the mouse, a 10-ng dose produced antianalgesia (present study) and hyperalgesia (Gul et al., 2000), whereas 0.1 ng may produce slight analgesia (Gul et al., 2000); in the rat, a low i.c.v. dose produces hyperalgesia and a high dose may produce analgesia (Yabuuchi et al., 1996; Oka and Hori, 1999). This species difference was not examined in the current study.

Hyperalgesia and allodynia are also produced by spinal Dyn (Laughlin et al., 1997; Nichols et al., 1997). The allodynic but not the hyperalgesic action is eliminated by spinal transection in the rat (Bian et al., 1998). Because the allodynic (Bian et al., 1998) and antianalgesic action of spinal Dyn (Wang et al., 1994) is eliminated by spinal transection, these actions might be related. However, generally, antianalgesia is produced by a small dose, whereas much higher doses are necessary to produce hyperalgesia. It is not known whether Lys-D-Pro-Thr antagonizes allodynia.

Our finding that the antianalgesic action of i.c.v. LPS has a slow onset and reaches a peak at 2 h requires further comment. This time course of action corresponds to findings in the literature for the slow onset and release of IL₁ (Gabellec et al., 1995). But, the present results show that the action of i.c.v. IL₁ is near maximal at 5 min. Similarly, all our previous work indicates that i.t. Dyn-induced antianalgesia is nearly maximal at 5 min. Thus, it would appear that the mode of release of IL₁ by i.c.v. LPS appears to be much slower than that attained through activation of a neuronal circuit produced by i.t. Dyn. Having made this conclusion about the slow onset of action of LPS, a contrasting situation must be mentioned. In the Introduction, we noted that Watkins et al. (1994) indicated that the quick onset of action of systemically administered LPS was due to its excitation of afferent nerves from the liver running in the vagus nerve sheath to the brain and not due to LPS acting in the brain. If as another of their articles (Maier et al., 1993) demonstrates IL₁ is responsible for producing hyperalgesia then it would seem that in their case, peripheral LPS must be releasing IL₁ very quickly or LPS-induced hyperalgesia is not mediated by IL₁ but by some other mechanism.

Even though the present work deals primarily with the antianalgesic action of IL₁ acting in the brain after Dyn administration in the spinal cord, others have linked spinal Dyn action to spinal IL₁ activity. Laughlin et al. (2000) link the hyperalgesic and allodynic actions of a large dose of Dyn i.t. with spinal IL₁. Both allodynia and hyperalgesia are attenuated by the i.t. administration of an IL₁ receptor antagonist. Also, spinal Dyn concentrations are enhanced in a number of chronic pain models (Draisci et al., 1991) and increased levels of IL₁ are found in the spinal cord in a sciatic cryoneurolysis neuropathic pain model (DeLeo and Colburn, 1999). Whether any of the steps we have proposed in the brain for IL₁ are similar to those acting in the spinal cord remains to be established. Because our model may more nearly reflect physiological mechanisms in the absence of gross pathological conditions, it may provide future insight into initial steps before such profound changes such as neuropathic pain develop.

Another notable point is that in our model, the administration of Dyn at the spinal cord leads to the activation of IL₁ in the brain, a function elicited by distal activation. This physical separation indicates that release of IL₁ was produced through nerve stimulation. Similarly, the effect of IL₁ in the brain was transmitted to the spinal cord through nerve action, which leads to the activation of CCK, LE, and NMDA receptors. The first assumption is that the flumazenil-sensitive benzodiazepine receptors would be on neurons. The central type of benzodiazepine receptor is an allosteric site for the binding of benzodiazepines to modulate -aminobutyric acid_A receptor function in neurons (Olsen, 1982). Because there seems to be very little IL₁ in neurons except at specific sites such as the hypothalamus (Breder et al., 1988), astrocytes would be a more likely source of IL₁ (Sairanen et al., 1997; Oka and Hori, 1999). Brain astrocytes also have peripheral type benzodiazepine receptors (Butterworth, 2000) and IL₁ released from astrocytes can act on IL₁ receptors present on neurons (Sairanen et al., 1997). The site of action of PK11195 might be on astrocytes but the fact that PK11195 inhibited the antianalgesic action of i.c.v. IL₁ would indicate that PK11195 interfered with the ability of IL₁ to activate the IL receptors on neurons (Ericsson et al., 1995; Sairanen et al., 1997), which are linked to spinal CCK receptors. The antagonistic action of Lys-D-Pro-Thr should be at IL receptors. However, a site of action of PK11195 in the astrocyte would not be compatible with our finding that PK11195 inhibited the antianalgesic action of IL₁ given i.c.v., which would bypass the need for release of IL₁ by the astrocyte.

Because spinal Dyn is involved in the resistance to morphine manifested in the sciatic nerve ligation neuropathic pain model (Gardell et al., 2000) and spinal Dyn activates IL₁ release in the brain (present study) and spinal cord (Laughlin et al., 1999), it is conceivable that tolerance to morphine can involve the antiopioid effect of Dyn through IL₁ action (in analogy to resistance in neuropathic pain as suggested by Gardell et al., 2000). Morphine tolerance and neuropathic pain are reversed by the administration of dynorphin antiserum (Gardell et al., 2000).

In summary, the results presented here indicated that IL₁ is involved in the antianalgesic actions of i.t. Dyn and i.c.v. pentobarbital and neurotensin. This connection is further evidenced by the results showing that administration of IL₁ i.c.v. produced antianalgesia through the CCK, LE, and NMDA receptor involvement. The present study further demonstrated that the central flumazenil-sensitive benzodiazepine receptor step occurs before the IL₁ step and the IL₁ step appears to involve a peripheral benzodiazepine receptor. Perhaps these present findings are indicative of the possible physiological neuromodulatory role played by IL₁ in the nervous system (Vitkovic et al., 2000; Araque et al., 2001).