Neurocircuitry of illness-induced hyperalgesia

BRAIN RESEARCH ELSEVIER

Brain Research 639 (1994) 283-299

Research Report

Neurocircuitry of illness-induced hyperalgesia L.R. Watkins a,,, E.P. Wiertelak b, L.E. Goehler c, K. MooneY-Heiberger a, j. Martinez a, L. Furness a, K.P. Smith a, S.F. Maier a " Department of Psychology, University of Colorado, Boulder, Boulder, CO 80309, USA b Department of Psychology, Macalester College, St. Paul, MN 55105, USA c Department of Cellular and Structural Biology, University of Colorado and Health Sciences Center, Denver, CO 80262, USA (Accepted 2 November 1993)

Abstract

We have previously demonstrated that illness-inducing agents such as lithium chloride (LiC1) and the bacterial cell wall endotoxin lipopolysaccharide (LPS) produce hyperalgesia on diverse pain measures. The present series of studies attempted to identify the neurocircuitry mediating these effects. These studies have demonstrated that illness-inducing agents produce hyperalgesia by activating: (a) peripheral nerves rather than by generating a blood-borne mediator (Expt. 1); (b) vagal afferents, specifically afferents within the hepatic branch of the vagus (Expt. 2); (c) as yet unidentified brain site(s) rostral to the mid-mesencephalon (Expt. 6); (d) a centrifugal pathway that arises from the nucleus raphe magnus, and not from the adjacent nucleus reticularis paragigantocellularis pars alpha (Expts. 4 and 5); (e) a centrifugal pathway in the dorsolateral funiculus of the spinal cord (Expt. 3); and (f) the same centrifugal pathways for diverse illness inducing agents (Expts. 3, 7 and 8). These data call for the re-evaluation of a number of assumptions inherent in previous studies of hyperalgesia. Key words: Hyperalgesia; Dorsolateral funiculus; Nucleus raphe magnus; Nucleus reticularis paragigantocellularis; Vagal afferent; Liver; Spinal cord; Lipopolysaccharide

I. Introduction

Pain is dynamically modulated by both pain inhibitory and pain facilitatory systems, rather than being passively received by the central nervous system. Pain inhibitory systems have been by far the most thoroughly investigated (for review, see [8,32,108]). Decades of study have provided detailed knowledge of how analgesia is produced by pharmacological agents, brain stimulation, and diverse environmental stimuli. This work has provided clear insights into the neurochemistry, neurophysiology, and neuroanatomy of these systems. In contrast, pain enhancement systems have only recently become a focus of study. By and large, such studies have examined hyperalgesia produced within or near a dermatome exposed to acute or chronic trauma. These studies have typically investigated hyperalgesia produced by painful events: by burns [40,65], by injec-

* Corresponding author. Fax: (1) (303) 492-2967. 0006-8993/94/$07.00 © 1994 Elsevier Science B.V. All rights reserved SSDI 0 0 0 6 - 8 9 9 3 ( 9 3 ) E 1 5 1 4 - 4

tion of irritating chemicals [18,29,39,66,73], by nerve damage [1,9,23,28,41,45,82,109,110], or by induction of joint inflammation [44,87]. Moreover, such work has primarily focused on the role of peripheral nerves and the spinal cord dorsal horn in pain enhancement. Perhaps the principal discovery of this work is that enduring pain can activate NMDA-linked nitric oxide a n d / o r prostaglandin cascades within the spinal cord, leading to increased pain responsivity by spinal cord dorsal horn cells [41,59,62]. This work has already yielded new possibilities for alleviation of neuropathic pain in m a n [19,41]. Although investigation has focused on sustained pain input as the trigger for pain enhancement systems, other input may also be able to engage pain facilitory circuitry. We have recently reported that agents that induce illness also produce hyperalgesia [56,57,104,105]. Both lipopolysaccharide (LPS; endotoxin derived from Gram-negative bacterial cell walls) and lithium chloride (LiCI) produce prolonged hyperalgesia. This illness-induced sensitization is generalizable across pain measures, and appears to be specific to pain [104,105].

284

LR. W)ltkins et al. /Brain Research 03 ~) (1994) 283-29~)

In addition, we have also shown that illness-induced hyperalgesia can be conditioned, such that cues (such as the taste of saccharine) that have been paired with an illness-inducing agent (such as LiCl) become sufficient to produce hyperalgesia [104,105]. These findings are consistent with current concepts of sickness behavior [22,47] and with classic conceptualizations of the role of recuperative behavior in pain [13,96]. That is, pain enhancement during a period of infection or illness would benefit the organism by minimizing activity, thus saving energetic resources for fighting infection, and for allowing recovery and healing to occur. The mechanisms by which illness-inducing agents produce pain enhancement are unknown. It is of interest to determine whether they produce hyperalgesia by the same or .by different mechanisms than are employed by prolonged pain input. At present, all that is known is that intraperitoneal administration of either LPS or LiCI enhances both supraspinally mediated pain responses (formalin test) and spinally mediated= pairL reflexes (tailflick test). This latter observation indicates that systemic illness-inducing agents must activate pain enhancement circuitry either at the level of the spinal cord or at brain sites, which modulate pain via centrifugal pathways to the spinal cord dorsal horn. The aim of the present series of experiments was to define the neurocircuitry of hyperalgesia produced by acute exposure to illness-inducing agents.

2. General methodology

2.1. Subjects Adult male Sprague-Dawley rats (Holtzman Lab.; 450-600 g) were used in all experiments. The animals were single housed with standard rodent chow and water available ad libitum. All testing was done during the light phase of the 12 h / 1 2 h l i g h t / d a r k cycle. All procedures were in accord with protocols approved by the University of Colorado Institutional Animal Care and Use Committee.

2.2. Pain testing Tailflick (TF) latencids to radiant heat were assessed using a modification [3] of the procedure developed by D'Amour and Smith [21]. All TF trials were conducted in a dimly lit experimental room maintained at 27°C. During testing, rats were loosely restrained in Plexiglas cylinders, designed such that their tails protruded from the rear so to allow TF testing to occur without disturbing the animals. The animals were placed in the cylinders approximately 30 min prior to testing. At this time, baseline TF latencies were assessed. The tail was placed over a radiant heat source

adjusted to provide baseline latencies m naive animals of approximately 6 s. This latency was chosen based on our previous studies, which demonstrated that it allows robust illness induced hyperalgesia to be observed under our present conditions [56,57]. Four TF trials were performed at 2 min intervals, with the last 3 being averaged to form the baseline (BL) measure. BL determination was followed by drug injection to induce hyperalgesia (see below). TF trials werc then performed once every 5 rain for 55 rain.

2.3. Induction of hyperalgesia All drugs were injected intraperitoneally (i.p.). Lipopolysaccharide (LPS; Sigma L8274, lot no. 51H4033) injections were 0.2 m g / k g , 20 m l / k g . Lithium chloride (LiC1; Sigma) injections were 20 r a g / kg, 0.15 M. Control animals received equivolume vehicle (0.9% saline).

2.4. Data analysis Observers were unaware of group membership and all data were analyzed by ANOVA. Post-hoc Scheffe tests were calculated, where appropriate. The alpha levels of multiple comparisons were adjusted using the Bonferroni method.

3. Experiment 1: effect of combined subdiaphragmatic vagotomy, celiac ganglionectomy and superior mesenteric ganglionectomy on hyperalgesia induced by LPS Hyperalgesia induced by LPS is mediated via release of interleukin-1 (IL-1), most likely from peripheral macrophages [56,57]. IL-1, or some substance released in response to IL-1, must ultimately influence pain modulation at the level of the spinal cord since LPS produces pain enhancement on the spinally-mediated TF test [56,57,104]. This could be accomplished in any of 3 ways. First, hyperalgesia might be produced by generation of a blood-borne mediator that activates brain a n d / o r spinal cord site(s) either directly (by crossing the blood-brain barrier) or indirectly (by binding to specialized receptors on cerebral blood vessels thereby inducing the generation of central neurochemical mediators). It is noteworthy that both of these possibilities have been forwarded to explain central actions of peripheral IL-1 [6,22,25,26,31,69,84,86,94]. Second, hyperalgesia might result from generation of a mediator that binds to and activates vagal sensory afferents. Such activation would then be directly relayed to higher centers via the nucleus tractus solitarius [4,49,50,93]. Indeed, vagal sensory afferents have been previously implicated as mediating hyperalgesia [34,81,92]. Third, hyperalgesia might be produced by

L.R. Watkins et al. / Brain Research 639 (1994) 283-299

generation of a mediator that binds to and activates spinal afferents in the splanchnic nerves. Such activation would be indirectly relayed to the brain, after coursing through the celiac and superior mesenteric ganglia and terminating within laminae I and V of the spinal cord. This is the major route of sensory information from the viscera directly to the spinal cord, and is generally considered to be the sole pathway for relaying painful visceral information to the central nervous system [16,17,37]. The present experiment sought to define which pathway(s) mediate LPS-induced hyperalgesia by examining the effect of subdiaphragmatic vagotomy (to disrupt abdominal sensory signals via the vagus) combined with celiac and superior mesenteric ganglionectomy (to disrupt spinal afferents carried by the splanchnic nerves) on LPS-induced hyperalgesia. Failure of this manipulation to block hyperalgesia would indicate a blood-borne mediator, whereas a successful blockade of hyperalgesia would indicate mediation by peripheral nerves.

3.1. Surgical procedure Following sodium pentobarbital anesthesia (Nembutal; Abbott Labs.; 55 m g / k g intraperitoneal [i.p.]; supplemented with methoxyflurane [Metofane, Pitman-Moore] as needed to maintain surgical plane), a midline abdominal incision was made. The stomach was gently retracted and covered with saline-moistened sterile gauze. The anterior and posterior trunks of the vagus nerve were transected, including the hepatic branch. The stomach was then carefully reflected upward to reveal the right and left celiac and superior mesenteric ganglia where they lie between the celiac and superior mesenteric arteries and the descending aorta. The ganglia and all associated nerve fibers,

11- • Vagus + Splanchnic Denervation 10 ~ A Sham

~ s£ ~

6-"

•_c

5 3

!

i

BL

-

,

5

-

,

l0

•

,

15

•

i

20

-

,

25

-

,

-

30

,

35

.

,

40

-

,

45

.

,

50

.

,

.

55

Timecourse (rain.) Fig. 1. Combined transection of the subdiaphramatic vagus and splanchnic nerves blocks i.p. LPS hyperalgesia. Compared to shamoperated controls (diamonds), denervated rats (squares) showed no hyperalgesia in response to i.p. injection of LPS. These data demonstrate that hyperalgesia is dependent upon a neural, rather than humoral, pathway to the central nervous system.

285

adipose and lymphatic tissue were removed. The stomach was then returned to its normal position, theincision closed, and penicillin administered. Sham surgery was identical to that described above with the exception that no neural lesions were made following retraction of the stomach and exposure of the ganglia. Since subdiaphragmatic vagotomy can result in an initial difficulty in swallowing and decreased appetite, these animals were provided with highly palatable foods until they resumed normal eating (usually 1-2 days).

3.2. Behavioral testing Behavioral testing was delayed until 8-14 days after surgery. TF testing was as described in General Methodology, above. All rats received LPS i.p.

3.3. Results Abdominal denervation (n = 8,) had no effect on baseline TF latencies, compared to sham-operated controls (n = 8) (Fl,14 = 1.57, P > 0.05). Denervation of the peritoneal cavity abolished the hyperalgesic effects of LPS (Fig. 1). ANOVA comparing the effect of LPS in denervated vs. sham operated controls confirmed a reduction in hyperalgesia in the denervated group (F1,14 = 86.22, P < 0.001). These data demonstrate that LPS produces a neural, rather than blood-borne, signal that activates pain modulatory circuitry within the central nervous system.

4. Experiment 2: effect of either subdiaphragmatic vagotomy or combined celiac + superior mesenteric ganglionectomy on LPS-induced hyperalgesia

The results of Expt. 1 show that LPS induces neural activation of either the vagus or splanchnic nerves, but does not delineate which nerve is critical. This issue was addressed in the present experiment by examining the effect, in separate groups, of either cefiac + superior mesenteric ganglionectomy (to disrupt the splanchnic nerve) or subdiaphragmatic vagotomy on LPS-induced hyperalgesia. Two types of vagotomies were performed: (a) a total subdiaphragmatic vagotomy to completely block vagal sensory information from the peritoneal cavity, and (b) selective transection of the hepatic branch of the vagus to specifically denervate the liver. This latter manipulation was included due to the combined facts that i.p. administered substances such as LPS make a 'first pass' through the liver prior to entering the circulation [35], and because the liver contains large numbers of specialized resident macrophages (Kupffer cells) which generate IL-1 and other cytokines in response to LPS and other toxins/ antigens detected in the blood-lymph mixture that

286

L.R. Watkins et al. / Brain Research 639 (1994) 2~3- 299

flows through this organ [24,71]. IL-1 in the liver can, in turn, either directly or indirectly activate the hepatic branch of the vagus [72].

4.1. Surgical procedure Four groups of animals received surgery. The sham-operated control group, celiac + superior mesenteric ganglionectomy group, total subdiaphragmatic vagotomy group, and selective hepatic branch vagotomy group each received the appropriate portions of the surgical procedure described in Expt. 1, above.

4.2. Behavioral testing Behavioral testing was delayed until 8-14 days after surgery. TF testing was as described in General Methodology, above. All rats received LPS i.p.

4.3. Surgical verification Verification of the surgical procedures was necessary, because it was expected that at least one of the surgical interventions would fail to effect LPS induced hyperalgesia. Completeness of subdiaphragmatic vagotomy can be verified by assessing its effect on stomach size since this procedure disrupts opening of the pyloric sphincter [78]. This in turn results in stomach enlargement due to chronic retention of stomach contents. In contrast, successful transection of only the hepatic branch of the subdiaphragmatic vagus has no effect on stomach size since pyloric sphincter function is not compromised. Upon completion of testing, animals that received sham surgery (controls), subdiaphragmatic vagotomies, or transection of the hepatic branch of the subdiaphragmatic vagus were overdosed with ether. The stomachs of these animals were removed and weighed. Completeness of celiac + superior mesenteric ganglionectomy can be verified by assessing its effect on immunohistochemically detectable calcitonin gene-related peptide (CGRP) in the stomach. C G R P fibers in the stomach are entirely extrinsic in origin, and they are derived from spinal dorsal root ganglion cells whose processes travel with splanchnic.nerves. Thus they are destroyed by celiac + superior mesenteric ganglionectomy [90]. Therefore, upon completion of behavioral testing, 3 control animals and all animals that had received celiac + superior mesenteric ganglionectomy were overdosed with sodium pentobarbital and perfused transcardially with heparinized saline followed by 4% paraformaldehyde in phosphate buffered saline (PBS, pH 7.2). The stomachs were removed, rinsed of their contents, and post-fixed in 4% paraformaldehyde for 2 hrs. The stomachs were then transferred to 30% sucrose-PBS for 3 days. Following embedding in Tis-

sueTek O.C.T. compound (Miles Labs.). the stomachs were cut in a cryostat at 25 /zm, mounted on gelatincoated slides, briefly air-dried, and then stored at - 2 0 ° C until reacted. To test for CGRP-immunoreacrive fibers, stomach slices were brought up to room temperature in potassium phosphate buffered saline (KPS, pH 7.2). After drying on a 36°(7 slide warmer, the slides were immersed in 3 changes (10 rain each) of KPBS and then coated with 1 : 500 rabbit anti-rat C G R P (Cambridge Research Biochem.; lot No. 08926; diluted in 0.2% normal goat serum and 0.3% Triton X in KPBS, pH 7.2 [0.2% NGS-0.3% TX-KPBS]), covered with parafilm and placed in a humidified chamber at 5°C for 72 h. At this time, the slides were immersed in 3 changes (10 min each) of KPBS and then coated with 1 : 50 biotinylated goat anti-rabbit IgG (Vector; lot No. B1202; diluted in 0.2% NGS-0.3% TX-KPBS). The slides were placed in a humidified chamber at 5°C for an additional 96 h. At this time, they were washed in 3 changes (10 min each) of KPBS and then reacted according to the manufacturer's (Vector) standard instructions for the Avidin-Biotin-Peroxidase Complex procedure [42] and developed with cobalt chloride enhancement of the diaminobenzidine reaction [43]. Thc tissue was examined by light microscopy for the presence of C G R P containing fibers. All microscopic verification was performed by an author [L.E.G.] who has experience in gastrointestinal immunohistochemistry and was blind to group designation of the tissue. Data were included for analysis only from animals with verified lesions.

4.4. Resul~ Stomach weight and immunohistochemical analyses confirmed the efficacy of the subdiaphragmatic vagotomy and hepatic branch transections. With regard to C G R P immunoreactivity, fine varicose CGRP-immunoreactive nerve fibers were seen in sham operated animals running in fascicles in the stomach muscle layer and submucosa. C G R P immunoreactive fibers were especially prevalent in association with blood vessels. Single C G R P immunoreactive fibers were also seen running along the villi. In animals deemed to have received complete celiac and superior mesenteric ganglionectomy, CGRP immunoreactivity was absent from all regions of the stomach. The stomachs (mean weight in grams +_ S.E.M.) of subdiaphragmatic vagotomized rats (13.348 _+ 1.803 g; n : 7) were heavier than stomachs of either sham-operated animals (7.243 +_ 0.851 g; n = 7) or rats whose hepatic branch of the vagus was selectively lesioned (4.855 +_ 0.456 g; n = 8). These observations were borne out statistically in that a main effect of group was found (F2,19= 14.959, P<0.0001). Post-hoc Scheffe tests revealed that the subdiaphragmatic vagotomy

L.R. Watkins et al./ Brain Research 639 (1994) 283-299 Sham Splanchnic • Subdi.aphragm. Vagus &

9-

O

~dlt

T

~7-

.~ 5"~

[-

4.

3

,

.

,

.

,

.

,

.

,

.

,

.

,

.

,

.

,

.

,

.

,

.

,

10 15 20 25 30 35 40 45 50 55 Timecourse (rain) Fig. 2. Discrete peripheral nerve lesions identify the hepatic vagus as crucial to LPS hyperalgesia. Compared to sham-operated controls (diamonds), splanchnic denervation (accomplished by combined superior and celiac ganglionectomy;open circles) had no effect on the hyperalgesic effect of i.p. LPS. In contrast, LPS hyperalgesia was abolished by either total subdiaphragmatic vagotomy (squares) or discrete transection of the hepatic branch of the vagus (filled circles). These data demonstrate that sensory afferent signals carried by the hepatic vagus are critical for the the production of LPS hyperalgesia. BL

5

group reliably differed from both the sham group and hepatic nerve lesion group, but that the sham and hepatic lesion groups did not reliably differ from each other. None of the surgical procedures affected baseline TF latencies, compared to sham-operated controls (F3,25 0.706, P > 0.05). LPS-induced hyperalgesia was not blocked by combined celiac + superior mesenteric ganglionectomy but was abolished by both types of vagal transections (Fig. 2; n = 8 for all groups, except subdiaphragmatic vagotomy: n = 7). It is noteworthy that total subdiaphragmatic vagotomy (which includes transection of the hepatic branch, see above) had no greater effect than selective transection of the hepatic branch of the vagus. These data indicate that sensory signals arising from the liver a n d / o r liver vasculature can completely account for the induction of LPS hyperalgesia. These observations were supported by statistical analyses. A N O V A revealed a main effect of group (F3,25 = 17.123, P < 0.001). Post-hoc Scheffe tests revealed that the subdiaphragmatic vagotomy group and the hepatic vagus transection group were both reliably less hyperalgesic than either the c e l i a c + superior mesenteric ganglionectomy group or the sham group. Importantly, the total subdiaphragmatic vagotomy group and the selective hepatic branch transection group did not differ. Likewise, the sham group and the celiac + superior mesenteric ganglionectomy group did not differ. :

induced hyperalgesia. Since all vagal afferents synapse in the brain, rather than spinal cord [4,49,50], these data imply that the spinal cord has no apparent role in transmitting LPS-induced signals to the brain. These data demonstrate that LPS induces vagal activity which is transmitted directly to the brain. This, in turn, activates a centrifugal pathway responsible for producing hyperalgesia at the level of the spinal cord. The present study was designed to identify the spinal trajectory of this hyperalgesia pathway.

5.1. Surgical procedure The details of this procedure have been reported previously [99]. Briefly, rats were initially anesthetized with ether, and maintained at surgical plane using methyoxyflurane. Sham-operated animals received a laminectomy and dural reflection at the second thoracic vertebral level. The incision was then closed with no neural lesion being made. Dorsolateral funiculus (DLF)-lesioned rats and dorsal column (DC)-lesioned rats were similarly prepared, but in addition received either bilateral D L F lesions or bilateral DC lesions, respectively. DC lesions were included as a control for nonspecific neural damage. More ventrally placed lesions were not included, due to motor dysfunctions and severe urinary infections that would result. Involvement of ventral pathways would be implied by a failure of either D L F or DC lesions to prevent LPS hyperalgesia.

5.2. Behavioral testing Behavioral testing was delayed until 8-14 days after surgery. TF testing was as described in General Methodology, above. All rats received LPS i.p.

5.3. Surgical verification A modification of the method described by Kluver and Barrera [48] was used to detect neural degenera-

The results of Expt. 2 demonstrate that vagal afferents from the liver participate in the mediation of LPS

/~ Sham @ DC • DLF

8-

~t ~=

T

" ±

x T .I..k

T

T

.£

=

6-

~d 5

,

5. E x p e r i m e n t 3: effect of spinal cord dorsal lateral funiculus (DLF) lesions and dorsal c o l u m n (DC) lesions on hyperalgesia induced by LPS

287

BL

-

, 5

.

, 10

.

, 15

.

, 20

.

, 25

.

, 30

.

, 35

.

, 40

.

, 45

-

, 50

.

,

.

55

Timecourse (rain) Fig. 3. Bilateral lesions of the spinal dorsal lateral funiculus (DLF) abolishes LPS hyperalgesia. Compared to sham-operated controls (diamonds), dorsal column (DC) lesions had no effect on LPS hyperalgesia. In contrast, DLF lesions (squares) abolished the ability of LPS to produce pain enhancement.

288

I..R. V(atkins et aL / Brain Research 639 (1994) 2&{-29q

tion resulting from dorsal lateral funiculus and dorsal column lesions. All animals were overdosed with sodium pentobarbital and perfused with heparinized saline followed by 10% formalin. The vertebral column surrounding the laminectomy site was removed and postfixed in 10% formalin for 2 weeks. Following dissection, the spinal cords were post-fixed an additional 2 days in 30% sucrose-10% formalin. Serial 25 > m sections were then cut in a cryostat and collected into PBS. After mounting every sixth section onto gelatinized slides, the tissue was then immersed in 70% alcohol for 10 rain and then stained for 24 h in 0.1% Luxol fast blue at 40°C. After rinsing in 95% alcohol and distilled water, the tissue was differentiated in fresh 0.05% lithium carbonate followed by several changes of 70% alcohol and distilled water. Staining for 20 sec in room temperature 0.25% Cresyl violet then occurred, followed by a distilled water wash and differentiation in 95% alcohol. Microscopic examination was used to determine the maximal extent of each lesion. Lesion extents were comparable to those reported previously [99].

5. 4. Results Baseline latencies did not differ between groups (F2.30 = 0.084, P > 0.05). LPS-induced hyperalgesia was abolished by bilateral lesions of the D L F (n = 11), compared to sham-operated controls (n = 10) (Fig. 3). In contrast, bilateral DC lesions (n = 12) had no effect on LPS hyperalgesia (Fig. 3). These observations were supported by statistical analyses. A reliable group difference was found (F2,30 = 9.26, P < 0.001). Post-hoc Scheffe tests revealed no difference between the DC and sham lesion groups. In contrast, reliable differences were observed between the D L F and DC groups, and between the D L F and sham groups. The lack of effect of DC lesions supports the conclusion that nonspecific effects secondary to general neural damage cannot account for the blockade of hyperalgesia produced by D L F lesions. The effect of D L F lesions on LPS hyperalgesia contrasts with the results of previous studies which have implicated the ventral funiculus in mediating the pain facilitory effects of cervical vagal afferent stimulation [34,112] and conditioned antianalgesia [101].

6. Experiment 4: effect of medullary nucleus raphe magnus (NRM) lesions and nucleus reticularis paragigantocellularis pars alpha (NRPgc) lesions on LPS-induced hyperaigesia

send axons through the DLF [97], the N R M and NRPgc have been repeatedly implicated in pain modulatory circuitry. Although this region is best known for its role in pain inhibitory pathways [8,32], recent work has suggested that it is involved in pain enhancement as well [32,100]. Therefore, the present experiment examined whether lesions of either the N R M or NRPgc would effect LPS-induced hyperalgesia.

6.1. Surgical procedure The procedure used has been described in detail previously [111]. Briefly, animals were anesthetized with sodium pentobarbital (55 m g / k g i.p.) supplemented with methoxyflurane as needed to maintain surgical plane. Electrolytic lesions were made by passing cathodal current (Stoelting Instr.) through a stereotaxically placed, single-pole, teflon-insulated stainless steel wire (Medwire, No. 361SS8T) which was cut flat to expose only the cross-sectional diameter of the wire. N R M and bilateral NRPgc lesions were made in separate groups of rats. N R M lesions were made using a 5 s, 0.8 m A current aimed at AP = - 1 . 8 m m from interaural zero, M L = 0, and D V = - 0 . 5 mm from interaural zero. Bilateral NRPgc lesions were made using a 3 sec, 0.8 mA current at each site, aimed at AP = - 1 . 8 mm from interaurai zero, ML = ± 1.0 mm, and DV = - 0.7 mm from interaural zero. Lesion coordinates were derived from the atlas of Paxinos and Watson [75]. In sham-operated controls, the electrode wire was lowered within 3 m m of either N R M or NRPgc. Upon closure of the incision, all animals received penicillin to prevent infection.

6.2. Behavioral testing Behavioral testing was delayed until 8-14 days after surgery. TF testing was as described in General Methodology, above. All rats received LPS i.p.

6.3. Surgical t:erification Following formalin fixation and sucrose-formalin post-fixation as described above, the brains of all animals were cut at 40 /,m and serial sections collected through the entire extent of the lesions. After mounting on gelatinized slides and air drying, the sections were stained with cresyl violet. Microscopic examination was used to determine the maximal extent of each lesion. Data were included for analysis only from animals with verified lesions (Figs. 4 and 5).

6.4. Results The results of Expt. 3 demonstrate that a centrifugal pain enhancement pathway is contained within the D L F of the spinal cord. Of the brain sites known to

Neither N R M (n = 7) nor NRPgc lesions (n = 7) altered baseline T F latencies compared to their respec-

L.R. Watkins et al. /Brain Research 639 (1994) 283-299 9-

289

,

•

8-

NRM

A Sham

.~ 5~

4-

BL

5

10

15

20

25

30

35

40

45

50

55

Timecourse (min)

Fig. 6. Lesions of the medullary nucleus raphe magnus (NRM) abolish LPS hyperalgesia. Compared to sham-operated controls (diamonds), N R M lesions (squares) prevented LPS from producing pain enhancement.

Fig. 4. Representative profile of the nucleus raphe m a n g u s lesions studied. The lesions typically extended from - 9 . 3 0 m m IA (top section containing a lesion) to - 1 1 . 3 0 m m IA (bottom section containing a lesion). The sections drawn represent - 9 . 3 0 , - 9 . 6 8 , - 1 0 . 3 0 , - 1 1 . 3 0 , - 12.30 and - 1 3 . 2 4 m m IA from Paxinos and Watson [75]. Although infringement of the nucleus raphe pontis occurred in some cases, this was not a consistent effect and so does not appear to account for the effects observed.

tive sham-operated controls (NRM shams = 7; NRPgc shams --- 8) (for NRM: F1,12= 0.00001, P > 0.05; for NRPgc: F1,13 = 0.07, P > 0.05). NRM lesions abolished LPS-induced hyperalgesia, compared to NRM-sham operated controls (FL~ z = 18.574, P < 0.001) (Fig. 6). In contrast, bilateral NRPgc lesions had no effect, in that rats with bilateral NRPgc lesions became as hyperalgesic as NRPGC-sham-operated animals following i.p. LPS (F1,13 = 2.714, P > 0.05) (Fig. 7).

7. Experiment 5: immunohistochemical assessment of LPS-induced c-fos activation in NRM and NRPgc

The blockade of LPS hyperalgesia observed in NRM-lesioned rats (Expt. 4, above) suggests that NRM neurons are activated by i.p. LPS. An alternative explanation would be that the electrolytic lesions disrupt hyperalgesia by damaging axons-of-passage rather than by destroying intrinsic neurons. One approach would be repeating Expt. 4 using excitotoxic, instead of electrolytic, lesions. However, given the viability problems associated with use of excitotoxic lesions in this brain

6.

A Sham • NRPgc

~ 5.

2 Fig. 5. Representative profile of the nucleus reticularis paragigantocellularis lesions studied. The lesions typically extended from - 10.30 m m IA (top section containing a lesion) to - 13.24 m m IA (bottom section containing a lesion). The sections drawn represent - 9 . 3 0 , - 9 . 6 8 , - 10.30, - 11.30, - 12.30 and - 13.24 m m IA from Paxinos and Watson [75].

BL

5

l0

15

20

25

30

35

40

45

50

55

T i m e c o u r s e (rain)

Fig. 7. Lesions of the the medullary nucleus reticularis paragigantocelluaris pars alpha (NRPgc) do not affect LPS hyperalgesia. Compared to s h a m oeprated controls (diamonds), NRPgc lesions (squares) had no effect on pain e n h a n c e m e n t produced by LPS.

2~1)

I..R. I}~nkins et al. / Brain Resear('h 030 ( l~)g4) 283 2UO

45] • NRM .o= [] NRPgc 3 0 ~

~•

g~

o z

0.

Vehl Veh2 LPSI LPS2 Fig. 8. LPS increases fos-like immunoreactivityin neurons within the nucleus raphe magnus (NRM), but not the nucleus reticularis paragigantocellularis pars alpha (NRPgc). As shown in Figs. 6 and 7, NRM lesions (but not NRPgc lesions) abolish LPS hyperalgesia. Similarly, LPS appears to activate the immediate-early protooncogene cfos in the NRM (but not in the NRPgc), since it differentially results in increased numbers of NRM neurons showing los-like immunoreactivity. The two animals tested using i.p. saline vehicle are labelled Vehl and Veh2. The two animals tested using i.p. LPS are labelled LPS1 and LPS2.

region [15], a different approach was employed. The present experiment attempted to verify whether N R M neurons become activated following i.p. LPS by seeking evidence of LPS-induced activation of the c-los immediate-early protooncogene. Fos-like immunoreactivity was used as an indirect assessment of c-los activation [27,89]. While one previous study has examined c-los activation in brain following systemic LPS, no examination was made of the ventral medial medulla [95]. Since NRPgc lesions failed to block LPS-induced hyperalgesia (Expt. 4, above), LPS-induced Fos-like immunoreactivity in this region was also examined to determine whether Fos-like protein specifically increases in areas mediating this hyperalgesic state.

first rinsed with 3 changes of KPBS (each 1(1 rain) to remove the cryoprotectant, and then exposed to 50();; ethyl alcohol for 10 mira 7()~ ethyl alcohol fl)r 15 rain, and 50% ethyl alcohol fi)r an additional 10 rain to suppress endogenous red blood cell peroxidase activity [64]. Following 2 changes of KPBS (5 rain each), the tissue was incubated in a 1% normal goat serum, 0.3% Triton-X KPBS (1% NGS-0.3% TX-KPBS) for 60 rain at room temperature, rinsed with two 5 min changes of KPBS and placed in rabbit anti-Fos IgG (generously supplied by M.J. Iadarola at 1 : 2000 in 5°C 0.2% N G S 0.3% TX-KPBS for 48 h. This rabbit anti-Fos is a polyclonal antisera made against a conserved region of Fos (amino acids 129-153; M-peptide, Peninsula Labs). in addition to Fos, this antisera also recognizes other Fos-like nuclear proteins or Fos-related antigens. The tissue was then rinsed with three 5 rain changes of 5°C 0.2% N G S - 0 . 3 % TX-KPBS and placed in room temperature 1:100 goat anti-rabbit lgG secondary antibody (Vector; lot No. B1202; diluted in 0.2% N G S 0.3% TX-KPBS) for 60 rain. After two 10 min washes in room temperature 0.2% N G S - 0 . 3 % TX-KPBS and one 10 rain wash in room temperature KPBS, the tissue was reacted according to standard manufacturer's

7.1. P r o c e d u r e

Rats were placed in Plexiglas tubes in the experimental environment 30 min prior to i.p. injection of either LPS (n = 2) or equivolume vehicle (n = 2). At 1 h post-injection, rats were euthanized with sodium pentobarbital. This timepoint was chosen based on pilot studies that indicated that LPS-induced Fos-like immunoreactivity was decreasing by 3 h post-injection (Watkins, Wiertelak and Goehler, unpublished data). The rats were perfused transcardially with 2.0% sodium nitrite in heparinized physiological saline, followed by 4% paraformaldehyde in KPBS (ph 7.2), and lastly by 100 ml saline. The brains were removed, blocked, immersed in 3 changes of KPBS (each 10 rain), and then placed in 5°C 30% sucrose-KPBS overnight. At this time, 25 mm serial sections were collected into a 5°C cryoprotectant solution [102]. Tissue from LPS- and vehicle-injected rats were always run simultaneously to ensure identical reaction times. To react the tissue, free-floating sections were

Fig. 9. Representative mid-saggital sections demonstrating the level and extent of decerebration lesions for each of the 12 rats tested (one section/subject).

291

L.R. Watkins et al. / Brain Research 639 (1994) 283-299

(Vector) instructions for the avidin-biotin-peroxidase complex procedure [42], with the exception that cobalt chloride was added to the diaminobenzidine solution to enhance the chromagen reaction [43]. Using these procedures, no background staining is observed when the primary antibody is replaced by control serum. 7.2. Results

•

Decerebrate/Vehicle

A

Decerebrate/LPS

[]

Sham/Vehicle

,

.

~7-

.~I

5-

•" ~

4-

[-

LPS differentially increased cfos-like immunoreactivity in the NRM. As seen in Fig. 8, the 2 rats injected with LPS exhibited almost twice the number of Fos-like immunoreactive neurons in the NRM than did the 2 saline-injected controls. Despite the small group sizes, this difference approached statistical significance (F = 14.872, P = 0.06). No such trends were observed in the NRPgc. As seen in Fig. 9, virtually identical numbers of Fos-like immunoreactive neurons were found in the NRPgc of LPS- and saline-injected animals (F = 0.03, P > 0.05). These data support the results of Expt. 5 in that they provide evidence that the neurons within NRM, but not NRPgc, become activated following LPS administration.

8. Experiment 6: effect of decerebration on LPS-induced hyperalgesia

As shown in Expt. 2, LPS induces hyperalgesia via activation of hepatic vagal afferents. In rat, vagal afferents from the liver terminate exclusively within the medullary nucleus tractus solitarius [4]. This nucleus, in turn, projects to at least 26 widely scattered brain sites [49,50,83]. Importantly, it does not directly project to the NRM, which was shown in Expts. 4 and 5 to be the likely source of the critical pain enhancement pathway in the spinal DLF (Expt. 3). A decerebration procedure was used as an initial attempt at identifying the level of the neuraxis critical for mediating LPS-induced hyeralgesia. Decerebration has previously been used to explore circuitry of stressinduced analgesia [61,98] and cervical vagal stimulation-induced hyperalgesia [34,80]. The use of decerebration in the present experiment allowed comparison to these previous studies as well as insight into the underlying circuitry of LPS-induced hyperalgesia.

3

, BL

.

, 5

•

, 10

•

, 15

.

, 20

.

, 25

.

30

,

.

35

, 40

.

, 45

.

, 50

.

,

.

55

Timecourse (min) Fig. 10. Decerebration blocks LPS hyperalgesia. LPS hyperalgesia was still observable in sham-operated rats injected with LPS (filled triangles), compared to vehicle controls (open squares). In contrast, decerebration prevented LPS from producing hyperalgesia (open triangles), compared to decerebrated vehicle-injected rats (filled squares).

Following incision of the dura, the wound was closed. For decerebrated rats, the same brain exposure was performed but, in addition, the brain was medio-laterally transected using suction. Following wound closure, all animals received penicillin.

8.2. Behavioral testing Behavioral testing was delayed until approximately 16-18 h after surgery, similar to previous reports [61,98]. TF testing was as described in General Methodology, above. All rats received either LPS or equivolume vehicle i.p.

8.3. Surgical verification Upon completion of behavioral testing, the animals were overdosed with ether, and their skulls removed and fixed in 10% formalin for a minimum of 2 weeks. At this time, the brain was dissected free and carefully examined to determine whether transection was complete. The approximate caudal extent of the transection was then determined by examination of mid-saggital sections. Data were included for analysis only from animals with lesions verified to completely transect the brain (Fig. 9).

8.4. Resul~ 8.1. Surgical procedure The details of this procedure have been presented in detail previously [98]. Briefly, the animals were initially anesthetized with ether and maintained on methoxyflurane. Sham-operated control animals received 2 rectangular openings in the skull which exposed the entire dorsal surface of the brain 1-3 mm rostral to lambda but avoided the mid-saggital sinus.

Decerebration had no effect on baseline TF latencies, compared to sham controls (F3,28 =0.518,P> 0.05). However, decerebration abolished the ability of i.p. LPS to induce hyperalgesia (Fig. 10). LPS still produced hyperalgesia in sham operated controls, compared to the vehicle-injected sham-operated group. These observations were supported by statistical analyses. ANOVA revealed a reliable group effect (Fa,E8=

') -9 t )-

L.R. Watkhts et a l . / B r a i n Research 030 (1994) 283.-299

10.160, P < 0.001). Post-hoc Scheffe tests revealed reliable differences between the decerebrate/vehicle (n = 8) and s h a m / L P S groups (n = 8), between the dec e r e b r a t c / L P S (n = 8) and the s h a m / L P S groups, and between the sham/vehicle (n = 5) and the s h a m / LPS groups.

9÷

2

9.1. Surgical procedure Rats were initially anesthetized with ether and then maintained at surgical plane using methoxyflurane (Metofane; Pitman-Moore). As described in detail elsewhere [99], spinal transection was performed at the second thoracic vertebral level using a heat cautery to seal major blood vessels and microdissecting scissors to cut the neural tissue. Sham operated control animals received a laminectomy and dural reflection, but no spinal cord damage. Gelfoam (Upjohn) was applied to aid coagulation. After closing the incision, the animals were treated with penicillin and housed in a 27°C room. Testing was delayed for approximately 16-18 h.

Spinal/LiCI

i

6543

¸

,

•

B L

Like LPS, LiCl produces both illness and hyperalgesia [56,57,104]. However, the mechanism of action of these two illness-inducing agents may be quite different. Unlike LPS, lithium salts cross the blood-brain barrier and exert direct neural effects [11,36]. The question arises whether LiC1 produces hyperalgesia via a direct blood-borne action on spinal circuitry or whether the action is indirect, via activation of centrifugal neurocircuitry originating within the brain. Therefore, the present experiment examined the effect of spinal transection on the ability of LiC1 to induce hyperalgesia on the spinally-mediated T F test [56,57,104]. Hyperalgesia would be unaffected by this manipulation if this pain enhancement results from direct effects at the level of the spinal cord, but would be abolished if activation of a centrifugal pathway is critical.

Spinal/Vehicle

8-

.d

9. Experiment 7: effect of spinal transection on LiCI induced hyperalgesia

'~J

•

i 5

•

, 10

.

, 15

.

, 20

•

, 25

,

, 30

•

,

.

35

, 40

•

, 45

•

, 50

•

f

•

55

Timecourse (min) Fig. 11. Spinal transection prevents LiCI hyperalgesia. LiCl produces hyperalgesia via activation of a centrifugal pain modulation circuit since spinal transection prevents LiC1 (open squares) from producing hyperalgesia relative to vehicle-spinal controls (open circles). LiCI hyperalgesia is still demonstrable in LiCl-injected sham-operated rats (triangles).

region of the spinal cord was exposed and visually inspected for any signs of damage (sham controls) and carefully probed with microdissecting forceps to identify any signs of incomplete spinal transection (experimentals). Data were included for analysis only from animals with verified surgical manipulations.

9.4. Results Spinal transection (n = 12) had no reliable effect on baseline T F latencies compared to sham-operated controls (n = 8) (Fz, L7= 0.521, P > 0.05). Spinal transection abolished LiCl-induced hyperalgesia (Fig. 11). That is, spinally transected animals that received LiC1 (n = 6) failed to become hyperalgesic relative to either spinally transected animals receiving vehicle (n = 6) or shamoperated animals injected with LiC1 (n = 8). These conclusions were supported by statistical analyses. A N O V A revealed a main effect of g r o u p (F2,17 = 7.924, P < 0.01). Post-hoc Scheffe tests revealed reliable differences between the spinal/saline vs. the s h a m / L i C I groups, and between the spinal/LiC1 and the s h a m / LiC1 groups. No reliable difference was found between spinal/LiC1 and spinal/saline.

9.2. Behauioral testing TF testing was as described in General Methodology, above. All rats received either LiC1 or equivolume vehicle i.p.

9.3. Verification of spinal transection Upon completion of testing, all animals were overdosed with ether. The vertebral column surrounding the surgical intervention was removed and fixed in 10% formalin for 2 weeks. At this time, the laminectomized

10. Experiment 8: effect of DLF and DC lesions on LiCI induced hyperalgesia

The results of Expt. 7 show that LiCl-induced hyperalgesia is mediated by activation of a centrifugal pathway from brain to spinal cord. In order to test for generality of neural pathways activated by illness-inducing agents, the present experiment examined the effect of DLF and DC lesions on LiCI induced hyperalgesia. This experiment provides parallel information to that of Expt. 3, above.

L.R. Watkins et aL / Brain Research 639 (1994) 283-299 7,

~6.

BL

5

10

15

20

:25

30

35

40

45

50

55

Timecourse (min) Fig. 12. Bilateral dorsolateral funiculus(DLF) lesions abolish LiCI hyperalgesia.Comparedto sham-operatedLPS-injectedcontrols(diamonds), rats with DLF lesionsstill showedcomparablehyperalgesia following i.p. LPS (squares). In contrast, rats with bilateral dorsal column lesionsstill developedLiCIhyperalgesia. 10.1. Procedures

Surgical procedures, behavioral testing, and lesion verifications were identical to those described in Expt. 3, above. 10.2. Results

LiCl-induced hyperalgesia was abolished by bilateral lesions of the DLF, compared to sham-operated controls (Fig. 12). In contrast, bilateral DC lesions had no effect on LiCI hyperalgesia (Fig. 12). These observations were supported by statistical analyses. A reliable group difference was found (F2.30 = 82.52, P < 0.0001). Post-hoc Scheffe tests revealed no difference between the DC and sham lesion groups. In contrast, reliable differences were observed between the DLF and DC groups, and betwen the DLF and sham groups. The lack of effect of DC lesions supports the conclusion that nonspecific effects secondary to general neural damage cannot account for the blockade of hyperalgesia produced by DLF lesions. The effect of DLF lesions on LiC1 hyperalgesia agrees with the blockade of LPS hyperalgesia by DLF lesions (Expt. 3) but contrasts with the results of previous studies which implicated the ventral funiculus in mediating the pain facilitory effects of cervical vagal afferent stimulation [34,79] and conditioned anti-analgesia [101].

11. Discussion The present series of studies represent the first investigation of neural pathways mediating illness-induced hyperalgesia. We have previously demonstrated that intraperitoneally injected illness-inducing agents such as LPS and LiC1 enhance response on both the tailflick and formalin tests, and that this sensitization appears to be specific to pain [104,105]. The present experiments have demonstrated that illness-inducing

293

agents produce hyperalgesia by activating: (a) peripheral nerves rather than by generating a blood-borne mediator (Expt. 1); (b) vagal afferents, specifically afferents within the hepatic branch of the vagus (Expt. 2); (c) as yet unidentified brain site(s) rostral to the mid-mesencephalon (Expt. 6); (d) a centrifugal pathway that arises from the nucleus raphe magnus, and not from the adjacent nucleus reticularis paragigantocellularis pars alpha (Expts. 4 and 5); (e) a centrifugal pathway in the dorsolateral funiculus of the spinal cord (Expt. 3); and (D the same centrifugal pathways for diverse illness inducing agents (Expts. 3, 7 and 8). One interesting aspect to the data presented in these experiments is that fact that abolishment of hyperalgesia often revealed a mild hypoalgesia; that is, a transient lengthening of the tail flick response. We have previously observed this phenomenon following blockade of LPS and LiC1 hyperalgesia using cytokine antagonists [57]. As discussed in this earlier paper [57], it appears that illness-inducing manipulations simultaneously activate both pain inhibitory and pain facilitatory mechanisms to some degree. As LiC1 and LPS can activate the pituitary-adrenal and sympathetic axes (for review, see refs. 57 and 105), perhaps some factor related to 'stress' underlies the mild pain inhibitory state induced. Given the strength of the hyperalgesic state produced by these manipulations, the behaviorally observable outcome is hyperalgesia. Thus pain responsivity at the behavioral level reflects the sum total of activity in these systems. The fact that i.p. LPS-induced hyperalgesia was completely abolished by subdiaphragmatic vagotomy, and more specifically by transection of only the hepatic branch of the subdiaphragmatic vagus, raises important questions as to the transduction mechanisms involved. The cell bodies of origin of the hepatic branch of the vagus are found in the left nodose ganglia and terminate predominantly in the left NTS [2,55]. Distally, the hepatic branch of the vagus ramifies from the anterior vagus [76]. The sensory afferent fibers terminate within the liver hilus, with endings found surrounding the peribiliary glands of the bile ducts, in paraganglia, in the portal vein adventitia, and in close apposition with the lymphatics [12,76]. No innervation of the liver parenchyma has been reported [12,76]. However, it is important to emphasize that as many as two-thirds of the vagal afferent fibers within the hepatic nerve actually do not terminate within the liver hilus, but rather follow the gastroduodenal a n d / o r common hepatic arteries [12]. Where these end is unknown. Thus, further work will be needed to clarify whether the liver is truly the source of the afferent signals which induce hyperalgesia. It is unlikely that LPS, which is comprised of fractured bacterial cell walls, could directly activate the vagus. The peripheral terminals of the rat vagus are

29-I

L.R. Watkins et al, / Brain Research t~39 (1994) 28.~-2~

known to express a number of receptors including muscarinic cholinergic, cholecystokinin, neurotensin, peptide YY, and opiate [67]. LPS is also unlikely to activate any of these receptors. Instead, it is more likely that the hepatic vagus expresses as-yet-unidentified receptors for substances released by hepatic immune cells in response to LPS. We have previously demonstrated that an antagonist for the cytokine IL-1 will abolish the hyperalgesic effects of either LPS or LiCI [56,57]. In the peritoneal cavity, where the LPS is injected, resident macrophages synthesize and release IL-I following LPS exposure [71]. This Ik-1 would be expected to travel to the liver since all gut circulation and lymphatics empty into the portal vein. The fact that lesions of the hepatic branch of the vagus were as effective as total subdiaphragmatic vagotomy in abolishing LPS hyperalgesia suggests that either a factor synthesized by peritoneal macrophages stimulates the hepatic vagus (either directly or via stimulated release of hepatic factors), or that LPS migrates to the liver, where it directly stimulates the release of a cytokine from hepatic macrophages (Kupffer cells) a n d / o r specialized hepatic endothelial cells [24]. This latter possibility is more likely for a number of reasons. First, the liver is the principal organ responsible for detecting and responding to toxins and b l o o d / lymph-borne antigens [24]. Second, the rapid onset of hyperalgesia after intraperitoneal LPS (within 5-10 rain) mandates that an equally rapid transduction mechanism must exist. The required rapidity of response strongly suggests m e d i a t i o n by liver macrophages. Unlike peritoneal macrophages which must upregulate synthetic pathways prior to IL-1 (or other cytokine) release [71], Kupffer cells and hepatic endothelial cells are tonically capable of releasing IL-1 [24], and thus can immediately respond to the presence of LPS. Third, Kupffer cells and hepatic endothelial cells can rapidly release a number of factors including prostaglandins, IL-1, IL-6, tumor necrosis factor (TNF), and nitric oxide (NO) [24], any of which could theoretically stimulate the hepatic vagus. Fourth, the Kupffer cells and endothelial cells are in intimate connection with both the b l o o d / l y m p h drainage through the liver as well as afferent vagal fibers [24]; in fact, afferent hepatic vagal fibers are found almost exclusively in the vicinity of Kupffer cells and endothelial cells [24]. Fifth, this contact implies that minute amounts of cytokines may be sufficient for activating the neural systems underlying the observed hyperalgesic state. And sixth, recent studies have shown that IL-1 injected into the hepatoportal system can activate the hepatic vagus for hours [72]. Although the fact that an IL-1 antagonist abolishes illness-induced hyperalgesia [56,57] implicates this cytokine as serving a critical role in the production of hyperalgesia, it needs to be recognized that this observation by no means implies that IL-1

directly activates the vagus. This cautional'y note i,~ necessitated by the fact that cytokincs such as IL-I typically act by triggering the generation of a cascade of subsequent cytokines, any of which might ultimately be found to be responsible for vagal activation. Studies are currently underway to determine the relative role of other products released by Kupffer and endothelial cells in producing hyperalgesia. Illness-induced activation of the hepatic vagus must result in signals to the nucleus tractus solitarius, since anatomical studies clearly show this to be the sole known termination of these fibers in rat [2,4]. Thus, the nucleus tractus solitarius, in addition to its role in producing analgesia [30,51,68,77], must now be considered an important site in pain facilitatory systems as well [34]. In turn, the nucleus tractus solitarius must somehow activate pain facilitatory neurons in the NRM, since lesions of this area (but not of the adjoining NRPgc) block illness-induced hyperalgesia (Expt. 4). However, a problem arises in that the nucleus tractus solitarius does not send an efferent projection to the NRM [50]. Thus a multisynaptic circuit must be proposed. Based on the fact that mid-mesencephalic transection blocked illness-induced hyperalgesia (Expt. 6), a critical relay must lie rostral to this level. Complex circuits could be easily proposed, but the simplest circuit would include a rostral brain site which receives a monosynaptic input from the nucleus tractus solitarius, and directly projects to the NRM but not to the NRPgc. Only one brain site appears to meet these requirements, this being the dorsomedial hypothalamic nucleus [14,50]. While the dorsomedial hypothalamus has been implicated in cardiovascular regulation, feeding/drinking regulation, inhibitory modulation of the lateral hypothalamic reward system, and opiate analgesia (for review, see [10]), the role of this nucleus in either taste aversion, illness/sickness responses, or pain enhancement has not, to our knowledge, been examined. Examination of LPS-induced Fos-like immunoreactivity (Expt. 5) confirmed the results of electrolytic lesions of ventral medullary sites (Expt. 4). That is, increases in Fos-like immunoreactivity (compared to vehicle-injected controls) were observed in the NRM of LPS-treated animals, but no comparable increase was observed in the adjacent NRPgc. These results extend the observations of Nance and colleagues [70,95]. Nance et al. reported that LPS results in c-Jbs activation in the paraventricular nucleus of the hypothalamus and A1 and A2 of the brainstem; NRM and NRPgc were not examined. The observed LPS effects, in agreement with the present report (Expts. 1 and 2), were mediated via the vagus since they were blocked by subdiaphragmatic vagotomy [70]. The involvement of the NRM in illness-induced pain facilitation compliments previous electrophysio-

L.R. Watkins et al. / Brain Research 639 (1994) 283-299

logical and behavioral work [32,46,52,53,85,100]. Although the NRM has long been implicated in pain inhibition [8], growing evidence suggests that this view underestimates its role in pain modulation. Electrical stimulation of this area has been observed, by some investigators, to excite dorsal horn nocisponsive neurons [52,53,85], and NRM (but not NRPgc) lesions abolish conditioned anti-analgesia [100]. The existence of a special class of NRM neuron (the 'ON' cell) with a pain facilitatory function has been proposed by Fields and colleagues [32,46], although its role in pain modulation is in dispute [74,91]. Although the NRM has a major projection to the spinal cord via the dorsolateral funiculus, previous studies have repeatedly implicated the ventral funiculus, not the dorsolateral funiculus, in centrifugal pain facilitation [79,100,101]. The present finding (Expts. 3 and 8) that dorsolateral funiculus lesions abolished b o t h LPS- and LiCl-induced hyperalgesias suggests that a unique centrifugal pain facilitatory pathway is activated by illness. As discussed above, pain enhancement has typically been studied as a 'pain-begets-pain' phenomenon. That is, pain facilitation has been assumed to occur due to traumatic events such as burns, nerve damage, visceral organ over-distention, joint inflammation, etc. In fact, such studies led to the proposal of a spinal-bulbo-spinal positive feedback loop to account for the hyperalgesia observed [16,17]. According to this model [16,17], pain signals from the viscera activate ascending pathways to the brainstem (possibly the NRM; see, [17,53]) which in turn, facilitates subsequent response of spinal cord neurones, via an as yet undefined spinal trajectory. This model is reminiscent of earlier, and more abstract, views of pain as a facilitator of recuperative behavior [13,96]. If the centrifugal pathway proposed by Cervero and colleagues serves this function, then perhaps illness is simply another way of accessing this same medullospinal circuit. This raises the issue of whether LiCI and LPS produce hyperalgesia because they activate viscera-to-spinal cord pain pathways. The focus on the spinal cord in the afferent portion of the circuit proposed by Cervero [16,17] derives from the fact that, for all of the trauma paradigms referred to above, the afferent pain signals are known (or assumed) to be directly relayed to the spinal cord. For visceral organs, these afferent signals would arrive via the splanchnic nerve. The splanchnic nerve is accepted to be the route of painful visceral stimuli to the central nervous system, while the vagus is thought to only carry non-noxious information [17,37]. Since transection of the splanchnic nerve failed to block illness-induced hyperalgesia, and transection of the hepatic vagus abolished illness-induced hyperalgesia, it is clear that illness-induced afferent information does not arrive at the central nervous system via the accepted route for visceral pain information.

295

However, such anatomical evidence may still beg the question of whether pain afferents are being activated by LiCI a n d / o r LPS. While it is true that rats injected with LPS or LiCI give no traditional behavioral indication of pain (that is, no writhing, vocalization, escape behaviors, etc. are ever observed), they do frequently assume a characteristic posture [60], lying either bellydown or on their sides. What class(es) of sensory fibers responds to these substances is unknown. It is clear that signals carried by the vagus are different from those of the splanchnic, at least insofar as electrical stimulation of the vagus, but not of the splanchnic, can cause emesis [5], and illness/emesis from food poisoning [5] and copper sulfate [20] are blocked by vagotomy. It has also never been determined whether 'silent' nociceptors might be involved. 'Silent' nociceptors, which become responsive only after inflammatory processes occur [38,54], would have been overlooked by the studies which led to the conclusion that the vagus only carries non-noxious information. A second issue raised by previous studies of trauma models regards the role of spinal cord substance P in illness-induced hyperalgesia. Given that substance P is a principle neurotransmitter of pain afferents to the spinal cord dorsal horn [7,88], and that substance P is known to be involved in the generation of NMDAmediated pain facilitory processes in spinal cord dorsal horn neurons [19,62], the assumption has naturally been made that the substance P involved in hyperalgesia derives from primary afferents. Our data suggests an alternative explanation. Since (a) NRM lesions blocked LPS-induced hyperalgesia (Expt. 4), (b) substance P-containing NRM neurons project to the spinal cord [63], (c) LPS-hyperalgesia is n o t affected by transection of the substance-P containing [7,88] splanchnic nerve (Expt. 2), and (d) intrathecal injections of substance P antagonist abolishes LPS-induced hyperalgesia [33], the NRM may well be an important source of substance P involved in pain facilitatory processes. One thing that is clear, is that multiple pain facilitatory circuits must exist. This parallels the situation for pain inhibitory circuitry, which is now well accepted to be composed of many complex circuits differentially activated under different circumstances. Although the study of pain enhancement systems is still in its infancy, the need to propose multiple pathways is obvious. One pain facilitatory circuit has been systematically defined by Gebhart, Randich and colleagues. Their work focuses on a pain enhancement system activated by low intensity cervical (presumed baroreceptor sensitive) vagal afferent stimulation [79,80]. This facilitation is blocked by decerebration [80] and is mediated by a centrifugal ventral funiculus pathway [79]. The effects are mediated by serotonin (suggested to derive from the nucleus raphe magnus [34]) and dynorphin at the level of the spinal cord [81].

296

L.R. Watkins et al. / Brain Research 639 (1094) 283-299

We have recently defined a similar pain facilitatory neural circuit mediating conditioned anti-analgesia [58,100,101,103]. This circuitry does not produce hyperalgesia, but rather abolishes the ability of opiates to produce analgesia [58,103]. In response to learned safety signals (cues that reliably predict that an aversive event will n o t occur), a neural circuit is activated which abolishes conditioned opiate analgesia, systemic morphine analgesia, and analgesia produced by IT and ~ opiate agonists [58,103,106,107]. The underlying neural circuitry includes the dorsal raphe nucleus, the nucleus raphe magnus (but not the nucleus reticularis paragigantocellularis), and the ventral funiculus of the spinal cord [100,101]. At the level of the spinal cord, cholecystokinin (but not neuropeptide FF or dynorphin) selectively mediates this effect [100,101,106,107]. Yet a third pain facilitatory system is apparently activated by illness-inducing agents. Illness-inducing agents injected intraperitoneally activate vagal afferents in the hepatic branch (presumably by binding IL-1 or other cytokine released in response to IL-1) which must signal the nucleus tractus solitarius. The nucleus tractus solitarius, in turn, must multi-synaptically activate the nucleus raphe magnus (but not the nucleus reticularis paragigantocellularis). This area then sends a critical pain facilitatory pathway via the dorsolateral funiculus to the spinal cord. At the level of the spinal cord, illness-induced hyperalgesia is mediated by substance P, cholecystokinin, NMDA, and NO, but does not appear to involve opiates, prostaglandins, leucotreines, or serotonin [33]. Thus it is clear, from even these early investigations, that pain facilitation will be as complex as pain inhibition. Acknowledgements. We would like to thank Kyra Simmons, Tim Koetzlow, Alicia Peticolas, and Lee H. Silbert for their outstanding technical assistance with various aspects of this project. We would also like to thank Dr. Dwight Nance at the University of Manitoba for pointing us in the direction of the liver, Dr. Tom Insel at LCS/NIMH for his patience in teaching us Fos immunohistochemistry, and Dr. Michael Iadarola at N I D R / N I H for both his advice on Fos immunohistochemistry and for his generous gift of Fos antibody. This work was supported by the Hughes Undergraduate Initiative, Undergraduate Research Opportunities Program, NIMH MH14617, and NIH NS31569.

12. References [1] Aanonsen, L.M., Kajander, K.C., Bennett, G.J. and Seybold, V.S., Autoradiographic analysis of 125I-substance P binding in rat spinal cord following chronic constriction injury of the sciatic nerve, Brain Res., 596 (1992) 259-268. [2] Adachi, A., Projection of the hepatic vagal nerve in the medulla oblongata, J. Auton. Nerv. Syst., 10 (1984) 287-293. [3] Akil, H. and Mayer, D.J., Antagonism of stimulation produced analgesia by p-CPA, a serotonin synthesis inhibitor, Brain Res., 44 (1982) 692-697. [4] Altschuler, S.M., Rinaman, L. and Miselis, R.R., Viscerotopic

representation of the alimentary tract in the dorsal and ventral vagal complexes in the rat. In S. Ritter, R.C. Ritter and C.D. Barnes (Eds.), Neuroanatomy and Physiolo~, of Abdominal Vagal Afferents, CRC Press, New York, 1992, pp. 21-54. [5] Andrews, P.L.R. and Lawes, I.N.C., A protective role for vagal afferents: an hypothesis. In: S. Ritter, R.C. Ritter and C.D. Barnes (Eds.), Neuroanatomy and Physiology of Abdominal Vagal Afferents, CRC Press. New York, 1992, pp. 279-302. [6] Banks, W.A., Kastin, A.J. and Durham, D.A., Bidirectional transport of interleukin-1 alpha across the blood brain barrier, Brain Res. Bull., 23 (1989) 433-437. [7] Barja, F. and Mathison, R., Sensory innervation of the rat portal vein and the hepatic artery, Z Auton. Nert. Syst., 10 (1984) 117-125. [8] Basbaum, A.I. and Fields, H.L., Endogenous pain control systems: brainstem spinal pathways and endorphin circuitry, Annu. Ret:. Neurosci., 7 (1984) 309-338. [9] Bennett, G.J. and Xie, Y.-K., A peripheral mononeuropathy in rat that produces disorders of pain sensation like those seen in man, Pain, 33 (1988) 87-107. [10] Bernardis, L.L. and Bellinger, L.L., The dorsomedial hypothalamic nucleus revisited: 1986 update, Brain Res. Ret.'., 12 (1987) 321-381. [11] Bernstein, 1.I., Chavez, M., Allen, D. and Taylor, E.M., Area postrema mediation of physiological and behavioral effects of lithium chloride in the rat, Brain Res., 575 (1992) 132-137. [12] Berthoud, H.R., Kressel, M. and Neuhuber, W.L., An anterograde tracing study of the vagal innervation of rat liver, portal vein and biliary system, Anat. Embryol., 176 (1992) 431-442. [13] Bolles, R.C. and Fanselow, M.S., A perceptual-defensive-recuperative model of fear and pain, Behat,. Brain Sci., 3 (1980) 291-301. [14] Carlton, S.M., Leichnetz, G.R., Young, E.G. and Mayer, D.J., Supramedullary afferents of the nucleus raphe magnus in the rat: a study using the transcannula HRP gel and autoradiographic techniques, J. Comp. Neurol., 214 (1983) 43-58. [15] Carruth, M.L., Fowler, A.A., Fairman, R.P., Mayer, D.J. and Liechnetz, G.R., Respiratory failure without pulmonary edema following injection of a glutamate agonist into the ventral medullary raphe of the rat, Brain Res. Bull., 28 (1992) 365-378. [16] Cervero, F., Laird, J.M.A. and Pozo, M.A., Selective changes of receptive field properties of spinal nociceptive neurones induced by noxious visceral stimulation in the cat, Pain, 51 (1992) 335-342. [17] Cervero, F. and Tattersall, J.E.H., Somatic and visceral sensory integration in the thoracic spinal cord. In F. Cervero and J.F.B. Morrison, Progress in Brain Research, Vol. 67, Elsevier, Amsterdam, 1986, pp. 189-202. [18] Chapman, V. and Dickenson, A.H., The spinal and peripheral roles of bradykinin and prostaglandins in nociceptive processing in the rat, Eur. J. Pharmacol., 219 (1992) 427-433. [19] Coderre, T.J., Katz, J., Vaccarino, A.L. and Melzack, R., Contribution of central neuroplasticity to pathological pain: review of clinical and experimental evidence, Pain, 52 (1993) 259-285. [20] Coil, J.D., Rogers, R.C., Garcia, J. and Novin, D., Conditioned taste aversions: vagal and circulatory mediation of the toxic unconditioned stimulus, Behav. Biol, 24 (1978) 509-519. [21] D'Amour, F.E. and Smith, D.L., A method for determining loss of pain sensation, J. Pharrnacol. Exp. Ther., 72 (1941) 74-79. [22] Dantzer, R., Bluteh, R.-M., Kent, S. and Goodall, G., Behavioral effects of cytokines: an insight into mechanisms of sickness behavior. In E. DeSouza (Ed.), Neurobiology of Cytokines, Academic Press, New York, in press. [23] Davar, G., Hama, A., Deykin, A., Vos, B. and Maciewicz, R., MK-801 blocks the development of thermal hyperalgesia in a

L.R. Watkins et al./ Brain Research 639 (1994) 283-299 rat model of experimental painful neuropathy, Brain Res., 553 (1991) 327-330. [24] Decker, K., Biologically active products of stimulated liver macrophages (Kupffer cells), Eur. J. Biochem., 192 (1990) 245261. [25] Dinarello, C.A., Biology of interleukin-1, FASEB J., 2 (1988) 108-115. [26] Dinarello, C.A., Interleukin-1 and interleukin-1 antagonism, Blood, 77 (1991) 1627-1652. [27] Dragunow, M. and Faull, R., The use of c-los as a metabolic marker in neuronal pathway tracing, J. Neurosci. Methods, 29 (1989) 261-265. [28] Draisci, G., Kajander, K.C., Dubner, R., Bennett, G.J. and Iadarola, M.J., Up-regulation of opioid gene expression in spinal cord evoked by experimental nerve injuries and inflammation, Brain Res., 560 (1991) 186-192. [29] Dray, A. and Perkins, M., Bradykinin and inflammatory pain, Trends Neurosci., 16 (1993) 99-104. [30] Du, H.-J. and Zhou, S.-Y., Involvement of solitary tract nucleus in control of nociceptive transmission in cat spinal cord neurons, Pain, 40 (1990) 323-331. [31] Dunn, A.J., Endotoxin-induced activation of cerebral catecholamine and serotonin metabolism: comparison with interleukin-1, J. Pharmacol. Exp. Ther., 261 (1992) 964-969. [32] Fields, H.L., Heinricher, M.M. and Mason, P., Neurotransmitters in nociceptive modulatory circuits, Annu. Rev. Neurosci., 14 (1991) 219-245. [33] Furness, L., Wiertelak, E.P., Maier, S.F. and Watkins, L.R., Illness-induced hyperalgesia: investigations of spinal cord neurochemistry, Proc. Soc. Neurosci., 19 (1993) in press. [34] Gebhart, G.F. and Randich, A., Vagal modulation of nociception, Am. Pain Soc. J., 1 (1992) 26-32. [35] Gilman, A.G., Goodman, L.S., Rail, T.W. and Murad, F., The Pharmacological Basis of Therapeutics, 1985. [36] Gottberg, E., Grondin, L. and Reader, T.A., Acute effects of lithium on catecholamines, serotonin, and their major metabolites in discrete brain regions, J. Neurosci. Res., 22 (1989) 338-345. [37] Grundy, D., Speculations on the structure/function relationship for vagal and splanchnic afferent endings supplying the gastrointestinal tract, J. Auton. Neru. Syst., 22 (1988) 175-180. [38] Habler, H.J., Janig, W. and Koltzenburg, M., A novel type of unmyelinated chemosensitive nociceptor in the acutely inflamed urinary bladder, Agents Actions, 25 (1988) 219-221. [39] Haley, J.E., Sullivan, A.F. and Dickenson, A.H., Evidence for spinal N-methyl-D-aspartate receptor involvement in prolonged chemical nociception in the rat, Brain Res., 518 (1990) 218-226. [40] Hardy, J.D., Wolff, H.G. and Goodell, H., Pain Sensation and Reactions, Williams and Wilkins, Baltimore, 1952. [41] Hayes, R.L., Mao, J., Price, D.D., Germano, A., d'Avella, D., Fiori, M. and Mayer, D.J., Pretreatment with gangliosides reduces abnormal nociceptive responses associated with a rodent peripheral mononeuropathy, Pain, 48 (1992) 391-396. [42] Hsu, S.-M., Raine, L. and Fanger, H., Use of avidin-biotin-peroxidase complex (ABC) in immunoperoxidase techniques: a comparison between ABC and unlabeled antibody (PAP) procedures, J. Histochem. Cytochem., 29 (1981) 577-580. [43] Hsu, S.-M. and Soban, E., Color modification of diaminobenzidine (DAB) precipitation by metallic ions and its application for double immunohistochemistry, J. Histochem. Cytochem., 30 (1982) 1079-1082. [44] Iadarola, M.J., Brady, L.S., Draisci, G. and Dubner, R., Enhancement of dynorphin gene expression in spinal cord following experimental inflammation: stimulus specificity, behavioral parameters and opioid receptor binding, Pain, 35 (1988) 313325.

297

[45] Kajander, K.C., Sahara, Y. and Iadarola, M.J., Dynorphin increases in the dorsal spinal cord in rats with a painful peripheral neuropathy, Peptides, 11 (1990) 719-728. [46] Kaplan, H. and Fields, H.L., Hyperalgesia during acute opioid abstinence: evidence for a nociceptive function of the rostral ventromedial medulla, J. Neurosci., 11 (1991) 1433-1439. [47] Kent, S., Bluthe, R.-M., Kelley, K.W. and Dantzer, R., Sickness behavior as a new target for drug development, Trends Pharmacol. Sci., 13 (1992) 24-28. [48] Kluver, H. and Barrera, E., A method for the combined staining of cells and fibers in the nervous system, J. Neuropathol. Exp. Neurol., 12 (1953) 400-403. [49] Leslie, R.A., Gwyn, D.G. and Hopkins, D.A., The central distribution of the cervical vagus nerve and gastric afferent and efferent projections in the rat, Brain Res. Bull., 424 (1982) 65-70. [50] Leslie, R.A., Reynolds, J.M. and Lawes, I.N.C., Central connections of the nuclei of the vagus nerve. In S. Ritter, S.C. Ritter and C.D. Barnes (Eds.), Neuroanatomy and Physiology of Abdominal Vagal Afferents, CRC Press, Ann Arbor, 1992, pp. 81-98. [51] Lewis, J.W., Baldrighi, G. and Akil, H., A possible interface between autonomic function and paincontrol opioid analgesia and the nucleus tractus solitarius, Brain Res., 424 (1987) 65-70. [52] Light, A.R., Casale, E.J. and Menetrey, D.M., The effects of focal stimulation in nucleus raphe magnus and periaquaductal gray in intracellularly recorded neurons in spinal laminae I and II, 3. Neurophysiol., 56 (1986) 555-571. [53] Lumb, B.M., Brainstem control of visceral afferent pathways in the spinal cord. In F. Cervero and J.F.B. Morrison, Progress in Brain Research, Vol. 67, Elsevier, Amsterdam, 1986, pp. 279293. [54] Lynn, B., 'Silent' nociceptors in the skin, Trends Neurosci., 14 (1991) 95. [55] Magni, F. and Carobi, C., The afferent and preganglionic parasympathetic innervation of the rat liver, demonstrated by the retrograde transport of horseradish peroxidase, J. Auton. Nerv. Syst., 8 (1983) 237-260. [56] Maier, S.F., Wiertelak, E.P., Goehler, L., Martin, D. and Watkins, L.R., Illness-induced hyperalgesia: effect of interleukin lfl (IL-1) and IL-1 antagonist, Proc. Soc. Neurosci., 19 (1993) in press. [57] Maier, S.F., Wiertelak, E.P., Martin, D. and Watkins, L.R., Interleukin-1 mediates the behavioral hyperalgesia produced by lithium chloride and endotoxin, Brain Res., 623 (1993) 321-325. [58] Maier, S.F., Wiertelak, E.P. and Watkins, L.R., Endogenous pain facilitatory systems: anti-analgesia and hyperalgesia, Am. Pain Soc. J., 1 (1992) 191-198. [59] Malmberg, A.B. and Yaksh, T.L., Hyperalgesia mediated by spinal glutamate or substance P receptor blocked by spinal cyclooxygenase inhibition, Science, 257 (1992) 1276-1279. [60] Meachum, C.L. and Bernstein, I.L., Conditioned responses to a taste CS paired with LiCl administration, Behav. Neurosci., 104 (1990) 711-715. [61] Meagher, M.W., Grau, J.W. and Kind, R.A., Role of supraspinal systems in environmentally induced antinociception: effect of spinalization and decerebration on brief shockinduced and long shock-induced antinociception, Behav. Neurosci., 104 (1990) 328-338. [62] Meller, S.T. and Gebhart, G.F., Nitric oxide (NO) and nociceptive processing in the spinal cord, Pain, 52 (1993) 127-136. [63] Menetrey, D. and Basbaum, A.I., The distribution of substance P-, enkephalin-, and dynorphin-immunoreactive neurons in the medulla of the rat and their contribution to bulbospinal pathways, Neuroscience, 23 (1987) 173-187. [64] Metz, C.B., Schneider, S.P. and Fyffe, R.E.W., Selective sup-

298

L.R. Watkins et al. /Brain Research 639 (19941 283-299

pression of endogenous peroxidase activity: application for enhancing appearance of HRP-labeled neurons in vitro, .1. Neurosci. Methods, 26 (1989) 181-188. [65] Meyer, R.A., Campbell, J.N. and Raja, S.N., Peripheral neural mechanisms of cutaneous hyperalgesia. In H.L. Fields, R. Dubner and F. Cervero (Eds.), Adt:ances in Pain Research and Therapy, Vol. 9, Raven, New York, 1985, pp. 53-71. [66] Millan, M.J., Czlonkowski, A., Morris, B., Stein, C., Arendt, R., Huber, A., Holtt, V. and Herz, A., Inflammation of the hind limb as a model of unilateral, localized pain: influence on multiple opioid systems in the spinal cord of the rat, Pain, 35 (19881 299-312. [67] Moran, T.H. and McHugh, P.R., Vagal receptor transport. In S. Ritter, R.C. Ritter and C.D. Barnes (Eds.), Neuroanatomy and Physiology of Abdominal Vagal Afferents, CRC Press, New York, 1992, pp. 157-178. [68] Morgan, M.M., Sohn, J.-H., Lohof, A.M., Ben-Eliyahu, S. and Liebeskind, J.C.. Characterization of stimulation-produced analgesia from the nucleus tractus solitarius in the rat, Brain Res., 486 (19891 175-180. [69] Nakamura, H., Nakanishi, K., Kita, A. and Kadokawa, T., lnterleukin-1 induces analgesia in mice by a central action, Eur. J. PharmacoL, 149 (1988) 49-54. [70] Nance, D.M., Wan, W., Wetmore, L. and Greenberg, A.H. Biochemical and neural mediation of endotoxin induced c-los protein in the brain, Presentation delivered at Research Perspectices in Psychoneuroimmunoloty IV, May, 1993. [71] Nathan, C.F., Secretory products of macrophages, J. Clin. Invest., 79 (1987) 319-326. [72] Niijima, A., The afferent discharges from sensors for interleukin-l-beta in hepatoportal system in the anesthetized rat, J. Physiol., 446 (1992) 236P. [73] Noguichi, K., Kowalsi, K., Traub, R., Solodkin, A., ladarola, M.J. and Ruda, M.A., Dynorphin expression and los-like immunoreactivity following inflammation induced hyperalgesia are colocalized in spinal cord neurons, Mol. Brain Res., 10 (19911 227-233. [74] Oliveras, J.L., Martin, G. and Montagne, J., Single-unit activity at ventromedial medulla level in the awake, freely moving rat: effects of noxious heat and light tactile stimuli onto convergent neurons, Brain Res., 506 (1990) 19-30. [75] Paxinos, G. and Watson, C., The Rat Brain in Stereotaxic Coordinates, 2nd edn., Academic Press, New York, 1986. [76] Prechtl, J.C. and Powley, T.L., A light and electron microscopic examination of the vagal hepatic branch of the rat, Anat. Embryol., 176 (1987) 115-126. [77] Randich, A., Roose, M.G. and Gebhart, G.F., Characterization of antinociception produced by glutamate microinjection in the nucleus tractus solitarius and the nucleus reticularis ventralis, J. Neurosci., 8 (1988) 4675-4684. [78] Raybould, H.E., Vagal afferent innervation and the regulation of gastric motor function. In S. Ritter, R.C. Ritter and C.D. Barnes (Eds.), Neuroanatomy and Physiology of Abdominal Vagal Afferents, CRC Press, New York, 1992, pp. 193-220. [79] Ren, K., Randich, A. and Gebhart, G.F., Vagal afferent modulation of spinal nociceptive transmission in the rat, J. Neurophysiol., 62 (1989) 401-415. [80] Ren, K., Randich, A. and Gebhart, G.F., Electrical stimulation of cervical vagal afferents. I. Central relays for modulation of spinal nociceptive transmission, J. Neurophysiol., 64 (1990) 1098-1114. [81] Ren, K., Randich, A. and Gebhart, G.F., Spinal serotonergic and kappa opioid receptors mediate facilitation of the tail flick reflex produced by vagal afferent stimulation, Pain, 45 (1991) 321-329. [82] Ren, K., Williams, G.M., Hylden, L.L., Ruda, M.A. and Dub-

ner, R., The intrathecal administration ot excitatory amino aci~ receptor antagonists selectively attenuated carrageenan-m duced behavioral hyperalgesia in rats, Eur..I. Pharmacol. 219 (1992) 235-243. [83] Ricardo, J.A. and Koh, E.T., Anatomical evidence of direcl projections from the nucleus of the solitary tract to the hypothalamus, amygdala, and other forebrain structures in the rat, Brain Res., 153 (19781 1 26. [84] Rivest, S., Torres, G. and Rivier, C., Differential effects ol central and peripheral injection of interleukin-l/3 on brain c-fos expression and neuroendocrine functions, Brain Res., 587 (1992) 13-23. [85] Rivot, J.P., Chauouch, A. and Besson, J.M., Nucleus raphe magnus modulation of response of rat dorsal horn neurons to unmyelinated fiber inputs, J. Neurophysiol., 44 (19801 1(/3911157. [86] Rothwell, N.J., Function and mechanisms of interluekin-1 in the brain, Trends Pharmacol. Sci, 12 (19911 430-436. [87] Ruda, M.A., Iadarola, M.J., Cohen, L.V. and Young, W.S.I., In situ hybridization histochemistry and immunocytochemistry reveal an increase in spinal dynorphin biosynthesis in a rat model of peripheral inflammation and hyperalgesia, Proc. NatL Aead. Sci. USA, 85 (1988) 622-626. [88] Sharkey, K.A., Sobrino, J.A. and Cervero, F., Evidence for a visceral afferent origin of substance P-like immunoreactivity in lamina V of the rat thoracic cord, Neuroscience, 22 (1987) 11177-1083. [89] Sheng, M. and Greenberg, M.E., The regulation and function of c-los and other immediate early genes in the nervous system, Neuron, 4 (1990) 477-485. [90] Sternini, C., Vagal afferent innervation of the enteric nervous system. In S. Ritter, R.C. Ritter and C.D. Barnes (Eds.), Neuroanatomy and Physiology of Abdominal Vagal Afferents, CRC Press, Ann Arbor, 1992, pp. 135-156. [91] Thurston, C.L. and Randich, A., Effects of vagal afferent stimulation on ON and OFF cells inthe rostroventral medulla: relationships to nociception and arterial blood pressure, J. Neurophysiol., 67 (1992) 180-196. [92] Thurston, C.L. and Randich, A., Electrical stimulation of the subdiaphragmatic vagus in rats: inhibition of heat-evoked responses of spinal dorsal horn neurons and central substrates mediating inhibition of the nociceptive tail flick reflex, Pain, 51 (1992) 349-365. [93] Van Giersbergen, P.L.M., Palkovits, M. and De Jong, W., Involvement of neurotransmitters in the nucleus tractus solitarii in cardiovascular regulation, PhysioL Rec., 72 (1992) 789815. [94] vanDam, A.-M., Brouns, M., Louisse, S. and Berkenbosch, F., Appearance of interleukin-1 in macrophages and in ramified microglia in the brain of endotoxin-treated rats: a pathway for the induction of non-specific symptoms of sickness'?, Brain Res., 588 (1992) 291-296. [95] Vriend, C.Y., Wan, W., Janz, L., Sorvillo, J.M., Greenberg, A.H. and Nance, D.M., Effects of endotoxin on the induction of c-los protein in the brain, plasma levels of corticosterone, and norepinephrine and VIP levels in the spleen of the rat, Proc. Neurosci. Soc., 18 (1992) 1010. [96] Wall, P.D., On the relation of injury to pain, Pain, 6 (1979) 253-264. [97] Watkins, L.R., Griffin, G., Leichnetz, G.R. and Mayer, D.J., Identification and somatotopic organization of nuclei projecting via the dorsolateral funiculus in rats: a retrograde tracing study using HRP slow-release gels, Brain Res., 181 (1980) 1-15. [98] Watkins, L.R., Kinseheck, I.B. and Mayer, D.J., The neural basis of footshock analgesia: the effect of periaqueductal gray lesions and decerebration, Brain Res., 276 (1983) 317-324.

L.R. Watkins et al. / Brain Research 639 (1994) 283-299

[99] Watkins, L.R., Thurston, C.L. and Fleshner, M., Phenylephrine-induced antinociception: investigations of potential neural and endocrine bases, Brain Res., 528 (1990) 273-284. [100] Watkins, L.R., Wiertelak, E.P., Furness, L., Smith, K., Martinez, J. and Maier, S.F., Neurocircuitry of centrifugal pain facilitatory systems, Proc. Soc. Neurosci., 19 (1993) in press. [101] Watkins, L.R., Wiertelak, E.P., Mooney-Heiberger, K., Grahn, R. and Maier, S.F., Endogenous anti-analgesia: effects of amygdala, dorsal raphe and spinal lesions, Proc. Soc. Neurosci., 18 (1992) 1025. [102] Watson, R.E., Wiegand, S.J., Clough, R.W. and Hoffman, G.E., Use of cryoprotectant to maintain long-term peptide immunoreactivity and tissue morphology, Peptides, 7 (1986) 155-159. [103] Wiertelak, E.P., Maier, S.F. and Watkins, L.R., Learning to activate CCK anti-analgesia: light cues abolish morphine analgesia, Science, 256 (1992) 830-833. [104] Wiertelak, E.P., Mooney-Heiberger, K., VanWyhe, W., FitzPatrick II, G.R., Maier, S.F. and Watkins, L.R., Internal aversive stimuli elicit acute and conditioned hyperalgesia, Soc. Neurosci. Abstr., 18 (1992) 1026. [105] Wiertelak, E.P., Smith, K.P., Furness, L., Mooney-Heiberger, K., Mayr, T., Maier, S.F. and Watkins, L.R., Acute and conditioned hyperalgesic responses to illness, Pain, (1993) in press. [106] Wiertelak, E.P., Yang, H.Y.-T., Mooney-Heiberger, K., Maier,

[107]

[108]

[109]

[110]

[111]

[112]

299

S.F. and Watkins, L.R., Conditioned anti-analgesia: spinal opiate and anti-opiate mechanisms, Soc. Neurosci. Abstr., 19 (1993) in press. Wiertelak, E.P., Yang, H.Y.-T., Mooney-Heiberger, K., Maier, S.F. and Watkins, L.R., Conditioned anti-analgesia: spinal opiate and anti-opiate mechanisms, Brain Res., submitted. Yaksh, T.L., Al-Rodhan, N.R.F. and Jensen, T.S., Sites of action of opiates in production of analgesia. In H.L. Fields and J.-M. Besson, Progress in Brain Research, l/ol. 77, Elsevier, Amsterdam, 1988, pp. 371-393. Yamamoto, T. and Yaksh, T.L., Spinal pharmacology of thermal hyperesthesia induced by incomplete ligation of sciatic nerve, Anesthesiology, 75 (1991) 817-826. Yamamoto, T. and Yaksh, T.L., Spinal pharmacology of thermal hyperesthesia induced by constriction injury of sciatic nerve. II. Excitatory amino acid antagonists, Pain, 49 (1992) 121-128. Young, E.G., Watkins, L.R. and Mayer, D.J., Comparison of the effects of ventral medullary lesions on systemic and microinjection morphine analgesia, Brain Res., 290 (1984) 119129. Zhuo, M. and Gebhart, G.F., Spinal serotonin receptors mediate descending facilitation of a nociceptive reflex from the nuclei reticularis gigantocellularis and gigantocellularis pars alpha in the rat, Brain Res., 550 (1991) 35-48.