BI-DIRECTIONAL CHANGES IN AFFECTIVE STATE ELICITED BYMANIPULATION OF MEDULLARY PAIN-MODULATORY CIRCUITRY

N. HIRAKAWA,* S. A. TERSHNER,† H. L. FIELDS and B. H. MANNING‡Departments of Neurology and Physiology, and the W. M. Keck Foundation Center for Integrative Neuroscience,

University of California at San Francisco, San Francisco, CA 94143-0453, USA

Abstract—The rostral ventromedial medulla contains three physiologically defined classes of pain-modulating neuron that projectto the spinal and trigeminal dorsal horns. OFF cells contribute to anti-nociceptive processes, ON cells contribute to pro-nociceptiveprocesses (i.e. hyperalgesia) and neutral cells tonically modulate spinal nociceptive responsiveness. In the setting of noxiousperipheral input, the different cell classes in this region permit bi-directional modulation of pain perception (analgesia vs hyper-algesia). It is unclear, however, whether changes in the activity of these neurons are relevant to the behaving animal in the absenceof a painful stimulus. Here, we pharmacologically manipulated neurons in the rostral ventromedial medulla and used the place-conditioning paradigm to assess changes in the affective state of the animal. Local microinjection of thea1-adrenoceptor agonistmethoxamine (50.0mg in 0.5ml; to activate ON cells, primarily), combined with local microinjection of thek-opioid receptoragonist U69,593 (0.178mg in 0.5ml; to inhibit OFF cells), produced an increase in spinal nociceptive reactivity (i.e. hyperalgesiaon the tail flick assay) and a negative affective state (as inferred from the production of conditioned place avoidance) in theconscious, freely moving rat. Additional microinjection experiments using various concentrations of methoxamine alone orU69,593 alone revealed that the rostral ventromedial medulla is capable of eliciting a range of affective changes resulting inconditioned place avoidance, no place-conditioning effect or conditioned place preference (reflecting production of a positiveaffective state). Overall, however, there was no consistent relationship between place-conditioning effects and changes in spinalnociceptive reactivity.

This is the first report of bi-directional changes in affective state (i.e. reward or aversion production) associated with pharma-cological manipulation of a brain region traditionally associated with bi-directional pain modulation. We conclude that, in additionto its well-described pain-modulating effects, the rostral ventromedial medulla is capable of modifying animal behavior in theabsence of a painful stimulus by bi-directionally influencing the animal’s affective state.q 2000 IBRO. Published by ElsevierScience Ltd. All rights reserved.

Key words:pain, affect, rostral ventromedial medulla, reward, aversion, hyperalgesia.

The mammalian brain contains a pain-modulating system thatcontrols transmission of nociceptive signals at the level of thespinal and trigeminal dorsal horns. This system partlycomprises neural circuits linking the neocortex, amygdala,periaqueductal gray matter (PAG), rostral ventromedialmedulla (RVM) and dorsal horn.15,16,28 Several differentclasses of drug, including opioids, cannabinoids and nicotiniccholinergic agonists, produce antinociception through actionson this circuitry.6,28,32,34

In the rat, the RVM consists of the nucleus raphe magnus,nucleus gigantocellularis pars alpha and adjacent reticularformation. Three physiological classes of neurons havebeen identified in this region based upon changes in theirfiring rate in relation to the noxious heat-evoked tail flick(TF) reflex. ON cells display an abrupt increase in firingjust prior to the TF.16,18 OFF cells show an abrupt pause infiring just prior to the TF. Neutral cells display no change in

firing prior to the TF. All three cell types project to the spinaldorsal horn,17 where they terminate in areas that receive noci-ceptive input from the periphery.

ON cells and OFF cells exert opposing actions on spinalnociceptive transmission. For example, systemic administra-tion of morphine18 or cannabinoids32 inhibits RVM ON cellsand activates OFF cells, resulting in decreased nociception(i.e. antinociception). By contrast, naloxone-precipitatedopioid withdrawal causes inhibition of OFF cells and activa-tion of ON cells,5 resulting in hyperresponsiveness to noxiousstimulation.5,26

It is clear that in the setting of noxious input from theperiphery, the different RVM cell classes permit bi-directionalmodulation of spinal nociceptive responsiveness (e.g. anal-gesia vs hyperalgesia). Under these circumstances, the RVMis capable of modifying animal behavior by increasing ordecreasing the influence the noxious stimulus has on theorganism as a whole. For example, the biological significanceof OFF cell activation in the normal, conscious animal can beappreciated in the context of predatorial conflicts and thephenomenon of “stress-induced analgesia”. Clearly, it isadvantageous under some conditions for an animal’s percep-tion of a noxious stimulus to be reduced so that appropriatedefensive responses can be engaged.29

What is unclear, however, is whether changes in RVMneuronal activity are relevant to the behaving animal in theabsence of significant nociceptive input from the periphery.This question is particularly relevant because the biologicalsignificance of ON cell activation is not well understood.

RVM, pain and affect 861

861

NeuroscienceVol. 100, No. 4, pp. 861–871, 2000q 2000 IBRO. Published by Elsevier Science Ltd

Printed in Great Britain. All rights reserved0306-4522/00 $20.00+0.00PII: S0306-4522(00)00329-8

Pergamon

www.elsevier.com/locate/neuroscience

*Present address: Department of Anesthesiology and Critical Care Medi-cine, Saga Medical School, 5-1-1 Nabeshima, Saga 849-8501, Japan.

†Present address: Department of Psychology, Western New EnglandCollege, Box 5249, Springfield, MA 01040, USA.

‡To whom correspondence should be addressed at: Department of Neurol-ogy. Tel.:11-415-476-9379; fax:11-415-476-9386.

E-mail address:bart@phy.ucsf.edu (B. H. Manning).Abbreviations: CPA, conditioned place avoidance; CPP, conditioned

place preference; MTX, methoxamine hydrochloride; PAG, periaque-ductal gray matter; RVM, rostral ventromedial medulla; TF, tail flick;U69,593, ([1]-[5a,7a,8b]-N-methyl-N-[7-(1-pyrrolidinyl)-1-oxaspiro(4.5)dec-8-yl]benzenacetamide).

Although it is clear that ON cell activation increases spinalnociceptive responsiveness (see above), the issue of whetherON cell activation can have behavioral relevance in theabsence of significant noxious peripheral input has not beenaddressed.

In the present experiments we used the place-conditioningparadigm to determine whether pharmacological manipula-tion of RVM neurons is capable of influencing the affectivestate of the conscious, behaving animal. These investigationswere focused on manipulations that increase spinal nocicep-tive reactivity. We hypothesized that, in the absence of signi-ficant noxious input from the periphery, a pharmacologicalmanipulation of the RVM that normally results in increasedspinal nociceptive reactivity is sufficient to generate a “virtualpain” signal that produces a negative affective state and,consequently, conditioned place avoidance (CPA). In thisway, clues to the biological significance of ON cell activationmight be obtained.

EXPERIMENTAL PROCEDURES

Subjects

Subjects were male Long–Evans rats (Simonsen Laboratory, Gilroy,CA) weighing 275–350 g at the start of the experiment. Rats werehoused three per cage on a 12 h/12 h light–dark schedule with foodand water availablead libitum. All experiments were carried out withthe approval of the Institutional Animal Care and Use Committee at theUniversity of California, San Francisco. All efforts were made to mini-mize animal suffering and to reduce the number of animals used.

Surgery

Under ketamine hydrochloride (100 mg/kg, i.p.) and sodium pento-barbital (25 mg/kg, i.p.) anesthesia, a single stainless steel guidecannula (26-gauge) was stereotaxically implanted 1.5 mm above theRVM according to the atlas of Paxinos and Watson.38 A stainless steelstylet was inserted into the guide cannula to keep it free of debrisduring the surgical recovery period. After surgery, an antibiotic (enro-floxacin) was administered intramuscularly. Animals were left torecover for five days before behavioral testing began.

Apparatus

In the present experiments, the effect of an intra-RVM drug treat-ment was paired repeatedly with a distinct compartment in a place-conditioning apparatus. The ability of that treatment to produce CPA orconditioned place preference (CPP) was then assessed. A standardplace-conditioning apparatus, with one major modification (seebelow), was used. The apparatus consisted of three wooden compart-ments. Two of the three compartments were “conditioning” compart-ments (i.e. intra-RVM drug treatments were paired with one or theother of these compartments) and the third compartment was a“neutral” compartment. Each of the three compartments was charac-terized by distinct visual and olfactory stimuli. One conditioningcompartment had horizontal stripes on the walls and an odor of 1.0%acetic acid, while the other conditioning compartment had verticalstripes and an odor of Medi-aire Biological Odor Eliminator (C. R.Bard, Murray Hill, NJ; 100 nl dissolved in 1 l distilled water). Walls ofuniform color and no distinctive odor characterized the neutralcompartment. The neutral compartment had two doors, consisting ofan opening to each conditioning compartment. These openings couldbe closed with a partition during conditioning to constrain the animalwithin a single conditioning compartment. All three compartmentswere equipped with eight infrared photo beams aimed at infrareddetectors. The photo beams were positioned horizontally 1.5 cmabove the floor. On the test day (i.e. after the conditioning procedure;see below), signals from the photo beams were fed into a computerprogrammed to record the total time spent in each compartment.

Importantly, our place-conditioning apparatus was designed with anadditional feature. Normally, in addition to having distinct olfactoryand visual cues, the conditioning compartments of standard place-conditioning apparatuses differ with respect to somatosensory cues

(e.g., wood chips on the floor of one compartment vs wire mesh onthe floor of the other). We chose, however, to forego differences insomatosensory cues so that we could incorporate glass floors into thedesign. In this way, a Hargreaves thermal stimulator21,33could be posi-tioned under the floor and TF latencies obtained from an animal duringthe first conditioning session for each intra-RVM drug or vehicle treat-ment (see below).

General method of training and testing in the place-conditioningapparatus

Pre-conditioning day (day1). Pre-conditioning testing was done inorder to assess baseline preferences to each compartment in the appar-atus. On day 1, the entrance connected to each compartment wasopened. Each rat was allowed to move freely throughout the apparatusfor 20 min. The time spent by the rat in each compartment was recorded.

Conditioning days (days2–10). The conditioning phase of allexperiments consisted of nine days, with three days of treatmentbeing paired with each of the three compartments in the place-conditioning apparatus. For the conditioning compartments (A andB), rats received an intra-RVM microinjection (drug or vehicle) andwere then allowed to freely explore one of the compartments for30 min. For the neutral compartment (N), rats were simply placed inthe compartment (i.e. they did not receive a microinjection) andallowed to freely explore the compartment for 30 min. For eachcompartment, each of the three conditioning days was separated bythree days, but the order in which the rat received training in eachcompartment was counterbalanced (one rat received training in theorder ABNABNABN, another in the order ANBANBANB, etc.).The conditioning compartment to be paired with an intra-RVM treat-ment (i.e. drug or vehicle) was also counterbalanced within experi-mental groups.

In order to assess spinal nociceptive reactivity, TF tests were carriedout during the first day only of the conditioning phase for each condi-tioning compartment (A or B). In other words, each rat underwent TFtesting twice, during the first conditioning trial in compartment A andduring the first conditioning trial in compartment B. No TF tests werecarried out in the neutral compartment. Three TF tests were performedat 10, 15 and 20 min after rats were placed in each compartment, andthe average of the results of the three trials was used for statisticalanalysis.

Post-conditioning day (day11, test day).One day following theconditioning phase, each rat was allowed to move freely throughoutthree compartments for 20 min in an undrugged condition. The timespent in each compartment was recorded.

Materials

Methoxamine hydrochloride (MTX; Research Biochemicals Inter-national) was dissolved in 0.9% physiological saline to concentrationsof 25 and 100mg/ml. U69,593 (Research Biochemicals International)was dissolved in 100% ethyl alcohol (50ml alcohol for every mg ofdrug) and diluted to final concentrations of 0.356 and 0.712mg/ml in0.9% saline.

The drugs or their corresponding vehicles were injected in a volumeof 0.5ml over a period of 60 s through a 33-gauge stainless steelinjection cannula extending 1.5 mm past the tip of the guide cannula.The injection cannula was left in place for an additional minute toallow absorption of the injectate and to prevent reflux. The use of afine-gauge injection cannula ensured minimal physical trauma to braintissue despite repeated insertion (insertion of the injection cannula intothe RVM once per day for several days was necessary for the condi-tioning procedure; see above).

Histology

After all of the behavioral experiments, rats were deeply anesthe-tized with sodium pentobarbital and perfused through the heart with0.9% saline followed by 10% formalin. Brains were removed andimmersed in 0.1 M phosphate buffer/30% sucrose until they sank.Tissues were cut into 50mm sections and mounted on gelatin-coatedslides. The sections were stained with 0.1% Cresyl Violet and cover-slipped with medium. The cannula tracts and injection sites wereverified under a light microscope.

N. Hirakawaet al.862

Statistical analyses

Time spent in each compartment on the pre-conditioning and finaltest days was converted to percentage of total time (30 min) spentin each compartment. For the pre-conditioning day, possible pre-conditioning bias toward favoring one compartment over the otherswas evaluated using one-way repeated measures ANOVA. After thefinal test day (day 11), percentage of time spent in the conditioningcompartment post-training (i.e. on the final test day) was comparedwith percentage of time spent in the same compartment on the pre-conditioning day using a pairedt-test. For TF data collected on thetraining days, TF latencies in the conditioned compartment werecompared with TF latencies in the unconditioned compartment usinga pairedt-test. The accepted level of significance for all tests wasP, 0.05. The data are presented as mean^S.E.M.

RESULTS

Microinjection of methoxamine (50mg) into the rostralventromedial medulla

The purpose of this experiment was to activate RVM ONcells pharmacologically and determine the effect of thismanipulation on the affective state of the animal. Wehypothesized that, in the absence of significant noxiousinput from the periphery, activation of RVM ON cellswould be sufficient to activate nociceptive neurons in thespinal and trigeminal dorsal horns, including nociceptiveneurons projecting to supraspinal targets. The production ofthis nociceptive signal was to be indexed in two ways: first, byconfirming the ability of ON cell activation to produce anincrease in spinal nociceptive responsiveness (i.e. hyper-algesia on the TF assay) and, second, by testing the abilityof this treatment to produce CPA (see Experimental Proce-dures). We felt it reasonable to assume that an ON cell-produced nociceptive signal would result in a negative affec-tive state such that repeated pairings of this treatment with adistinct environment would result in a CPA (i.e. the animalwould learn to avoid the environment in which it had experi-enced ON cell-induced “pain”).

Activation ofa1-adrenergic receptors in the RVM results inactivation of ON cells22 and increases spinal nociceptiveresponsiveness (i.e. produces hyperalgesia40). In pilot experi-ments, we determined that intra-RVM microinjection of thea1-receptor agonist MTX (50mg in 0.5ml) results in statisti-cally significant hyperalgesia on the noxious heat-evoked TFassay. Since ON cell activation produces hyperalgesia on theTF assay,5,26 we assumed that this dose of MTX activatespredominantly ON cells. We used this dose of MTX, there-fore, to determine whether RVM-induced activation of spinaldorsal horn neurons is sufficient to generate a nociceptivesignal that is aversive to the animal.

Histology.Histological examination showed that cannulaswere correctly implanted into the RVM in seven rats treatedwith intra-RVM MTX (50 mg; Fig. 1). Data from these ratswere included for statistical analysis.

Nociceptive reactivity.Intra-RVM microinjection of MTX(50mg) produced significantly lower TF latencies comparedwith intra-RVM saline, indicating the production of hyper-algesia (Fig. 2B; pairedt-test,P, 0.05).

Affective changes: place conditioning.On the pre-condi-tioning day (i.e. before drug treatments), the mean percentageof time spent by rats in the conditioned compartment

RVM, pain and affect 863

Bregma -11.60 mm

1

IntA

IntDM

IntPPC

icp

sp5

sol

RPa

B4

C3

mlf

8n

IS

ECu

DC

LatPC

rs

LPGi

4V

SolM

SolIM

RPa

Bregma -11.30 mm

IntDM

4V

4V

A4

8n

sp5

LR4V

rs

7

PPy

ml

RMg

ROb

GiA

7VM

7DI

7VI

7DL

7L

Sp5O

DMSp5V

VCP

X

SpVe

LVe

Y

Inf

LatPC

P7

SolRL

prf

4&5

pcuf

Sim

IntA

IntDM

jx das

4V

4V

A4

LR4V

rs

vsc

PPy

bas

RPa

mlf

CI

asc7

DC

X

SpVe

LVe

VeCb

MVePC

MVeMC

Pr

Y

Inf

IntA

Lat

LatPC

2&3

1

SMV

IntDL

sol

psf

Bregma -11.00 mm

1

psf

Lat

das

Y

scp

g7

rs

ts

pd

mlf

CI

icp

Lat

IntA

A4

Y

Pr

SMV

PPy

ias

Bregma -10.04 mm

4V

LR4V

icp

scp

jx

rs

LPGi

DPGi

IS

Lat

LVe

Pr

g7

6

mlf

CI

8vn

bas

SMV

A5

MVeMC

4V

8cn

tz

ml

bas

SPO

Tz

MSO

IS

LPBV

Methoxamine (50.0 µg)

Fig. 1. Histological results corresponding to rats that received microinjec-tion of MTX (50mg in 0.5ml) into the RVM. Representations of fourcoronal sections through the rat RVM are shown in sequence from anteriorto posterior. The numbers in the right margin indicate millimeters posteriorto bregma. The filled circles show the approximate positions of the cannulatips corresponding to rats in the MTX treatment group (n�7). Adapted

from Paxinos and Watson.38

Fig. 2. Nociceptive and place-conditioning data corresponding to rats thatreceived microinjection of MTX (50mg in 0.5ml) into the RVM. Verticalbars indicate S.E.M. Percentage of time spent in each compartment on thepre- vs post-conditioning test days is shown in A, and TF latenciesproduced by each microinjection treatment during the first conditioningtrial are shown in B. (B) *P, 0.05, pairedt-test, compared with saline

microinjection.

(compartment paired with intra-RVM MTX), unconditionedcompartment (i.e. compartment paired with intra-RVMsaline) and neutral compartment was 28.8^3.7%,34.2^3.9% and 37.0̂ 5.2%, respectively (Fig. 2A). Therewere no statistically significant differences in percentage oftime spent by rats in each of these compartments (one-factorrepeated measures ANOVA,F2,12�0.639,P. 0.05), indicat-ing that rats did not prefer any one compartment over theothers on the pre-conditioning day. Similarly, on the post-conditioning test day (after the nine-day conditioning phasein which intra-RVM MTX was paired with the conditionedcompartment), there were no statistically significant differ-ences in percentage of time spent by rat in each of compart-ments (one-factor repeated measures ANOVA,F2,12�1.366,P. 0.05). Furthermore, the percentage of time spent in theMTX-conditioned compartment on the post-conditioning daywas not significantly different from the time spent in thesame compartment on the pre-conditioning day (Fig. 2A;35.7^5.7% post-conditioning vs 28.8̂ 3.7% pre-condition-ing; paired t-test, P. 0.05). In other words, intra-RVMmicroinjection of MTX (50mg) did not appear to changethe affective state of the animal.

Microinjection of methoxamine (50mg) plus U69,593(0.178mg) into the rostral ventromedial medulla

The results suggest that intra-RVM microinjection of MTX(50mg in 0.5ml) produces an increase in the excitability ofspinal dorsal horn neurons (i.e. hyperalgesia; Fig. 2B) withoutproducing CPA (Fig. 2A). The apparent failure of thismanipulation to produce CPA suggests that, in the absenceof significant noxious peripheral input, this particular intra-RVM manipulation does not produce a nociceptive signal thatis aversive (i.e. “painful”) to the animal in and of itself. Thereare several possible explanations for the lack of consistencybetween the TF and place-conditioning data. For example,intra-RVM microinjection of MTX may sensitize theTF response without producing painper se, or our place-conditioning paradigm was not sensitive enough to detectthe aversive (i.e. “painful”) effects of this manipulation.

Subsequent experiments were designed to address theseissues further (see below).

With regard to the first explanation (above), it is possiblethat the excitability of dorsal horn nociceptive neurons wasincreased by intra-RVM MTX without affecting the “base-line” firing rate of these neurons (i.e. in the absence ofnoxious peripheral input). This would explain the inabilityof this manipulation to produce a CPA. In the presence of anoxious peripheral stimulus (e.g. stimulation of the tail withnoxious heat), however, the increase in excitability of spinaldorsal horn neurons would result in hyperalgesia on the TFassay.

The question remains as to whether the RVM is capable ofgenerating a “virtual” nociceptive signal that is aversive to theanimal. Clearly, in order to achieve this sort of effect, it wouldbe necessary for the RVM to increase the activity of nocicep-tive projection neurons in the spinal cord in the absence of anoxious stimulus. Although activation ofa1-adrenergic recep-tors in the RVM is clearly sufficient to activate ON cells22 andto produce hyperalgesia on the TF assay40 (and presentresults), it is possible that RVM OFF cells remain activeunder these circumstances, since (i) OFF cells are tonicallyactive in the conscious animal30 and (ii) activation ofa1-adrenergic receptors in the RVM does not inhibit OFF cellsdirectly.22 It is possible, then, that OFF cell activity counter-acts the effects of ON cell activation such that only evokedactivity of spinal nociceptive neurons (e.g., by noxious tailstimulation) is increased by intra-RVM MTX.

We attempted to circumvent the possible problem of tonicOFF cell activity by inhibiting the activity of this cell type inaddition to activating ON cells with MTX. Intra-RVM micro-injection of k-opioid receptor agonists selectively inhibitsOFF cells31,37 and prevents these cells from eliciting anti-nociception.37 We therefore used dual intra-RVM micro-injections of thek-opioid receptor agonist U69,593 andMTX in an attempt to inhibit OFF cells and activate ONcells, respectively. In this way, we hoped to maximize thechance of increasing the activity of nociceptive neurons inthe spinal dorsal horn.

After confirming that intra-RVM microinjection of U69,593alone has no effect on nociceptive reactivity nor a place-conditioning effect, this manipulation was combined withintra-RVM microinjection of MTX (see below). The effectsof this combined manipulation on nociceptive reactivity andin the place-conditioning paradigm were then examined.

Microinjection of U69,593(0.178mg)

Histology. The histological examination showed that theinjection cannulas successfully targeted the RVM in six ratstreated with 0.178mg of U69,593 (Fig. 3). Data from theserats were included for statistical analysis.

Nociceptive reactivity. Intra-RVM microinjection ofU69,593 (0.178mg) did not produce significant hyperalgesiaas compared with intra-RVM microinjection of vehicle (Fig.4B; P� 0.37).

Affective changes: place conditioning.On the pre-conditioning day (i.e. before drug treatments), the meanpercentage of time spent by rats in the conditioned com-partment (compartment paired with intra-RVM U69,593),unconditioned compartment (i.e. compartment paired with

N. Hirakawaet al.864

Bregma -11.60 mm

1

IntA

IntDM

IntPPC

icp

sp5

sol

RPa

B4

C3

mlf

8n

IS

ECu

DC

LatPC

rs

LPGi

4V

SolM

SolIM

Bregma -11.30 mm

IntDM

4V

4V

A4

8n

sp5

LR4V

rs

7

PPy

ml

RMg

ROb

GiA

7VM

7DI

7VI

7DL

7L

Sp5O

DMSp5V

VCP

X

SpVe

LVe

Y

Inf

LatPC

P7

SolRL

pcuf

IntA

IntDM

jx das

4V

4V

A4

LR4V

rs

vsc

PPy

bas

RPa

mlf

CI

asc7

DC

X

SpVe

LVe

VeCb

MVePC

MVeMC

Pr

Y

Inf

IntA

Lat

PFl

LatPC

2&3

1

SMV

IntDL

sol

Bregma -11.00 mm

1

das

Y

g7

rs

ts

pd

mlf

CI

icp

A4

Y

4V

SMV

PPy

ias

rs

LPGi

bas

A5

IS

U69,593 (0.178 µg)

Fig. 3. Histological results corresponding to rats that received microinjec-tion of U69,593 (0.178mg in 0.5ml) into the RVM. Representations ofthree coronal sections through the rat RVM are shown in sequence fromanterior to posterior. The filled triangles show the approximate positions ofthe cannula tips corresponding to rats in the U69,593 treatment group

(n�6). Adapted from Paxinos and Watson.38

intra-RVM vehicle) and neutral compartment was 36.2^4.5%, 30.8̂ 4.8% and 34.0̂ 6.7%, respectively (Fig. 4A).There were no statistically significant differences in percent-age of time spent by rats in each of these compartments (one-factor repeated measures ANOVA,F2,10� 0.167,P. 0.05),indicating that rats did not prefer any one compartment to theothers on the pre-conditioning day. Similarly, on the post-conditioning test day (after the nine-day conditioning phasein which intra-RVM U69,593 was paired with the conditionedcompartment), there were no statistically significant differ-ences in percentage of time spent by rats in each of the compart-ments (Fig. 4A; one-factor repeated measures ANOVA,F2,10�1.22, P. 0.05). Furthermore, the percentage of timespent in the U69,593-conditioned compartment on the post-conditioning day was not significantly different from the timespent in the same compartment on the pre-conditioning day(Fig. 4A; 25.4̂ 5.1% post-conditioning vs 36.2̂4.5% pre-conditioning; pairedt-test, P. 0.05). In other words, intra-RVM microinjection of U69,593 (0.178mg) did not appear tochange the affective state of the animal.

Microinjection of methoxamine (50mg) plus U69,593(0.178mg)

Histology. The histological examination showed that the

cannulas were correctly implanted into the RVM in 13 rats(Fig. 5) and data from these rats were included for statisticalanalysis.

Nociceptive reactivity.When the rats were treated withintra-RVM U69,593/MTX, TF latencies decreased signifi-cantly as compared with treatment with intra-RVMU69,593 alone (Fig. 6B; pairedt-test, P, 0.05), indicatingthe production of hyperalgesia.

Affective changes: place conditioning.On the pre-conditioning day (i.e. before drug treatments), the meanpercentage of time spent by rats in the conditioned compart-ment (compartment paired with intra-RVM U69,593/MTX),unconditioned compartment (i.e. compartment paired withintra-RVM U69,593/saline) and neutral compartment was41.1^ 4.5%, 27.8̂ 3.2% and 31.0̂ 2.7%, respectively(Fig. 6A). There were no statistically significant differencesin percentage of time spent by rats in each of these compart-ments (one-factor repeated measures ANOVA,F2,24�2.56,P. 0.05), indicating that rats did not prefer any one compart-ment to the others on the pre-conditioning day. On the post-conditioning test day, however (after the nine-day condition-ing phase in which intra-RVM U69,593/MTX was pairedwith the conditioned compartment), there was a statisticallysignificant decrease in the percentage of time spent in theconditioned compartment compared with time spent inthe same compartment on the pre-conditioning day (Fig.6A; 25.0^2.6% post-conditioning vs 41.4̂ 4.5% pre-conditioning; pairedt-test,P, 0.05). In other words, intra-RVM microinjection of U69,593/MTX produced CPA.

Off-site controls.In the above experiment, several injectioncannulas did not target the RVM successfully. Thus, severalanimals (n�8) received microinjections of MTX plus

RVM, pain and affect 865

Bregma -11.60 mm

1

IntA

IntDM

IntPPC

icp

sp5

sol

RPa

B4

C3

mlf

8n

IS

ECu

DC

LatPC

rs

LPGi

4V

SolM

SolIM

RPa

Bregma -11.30 mm

IntDM

4V

4V

A4

8n

sp5

LR4V

rs

7

PPy

ml

RMg

ROb

GiA

7VM

7DI

7VI

7DL

7L

Sp5O

DMSp5V

VCP

X

SpVe

LVe

Y

Inf

LatPC

P7

SolRL

prf

4&5

pcuf

Sim

IntA

IntDM

jx das

4V

4V

A4

LR4V

rs

vsc

PPy

bas

RPa

mlf

CI

asc7

DC

X

SpVe

LVe

VeCb

MVePC

MVeMC

Pr

Y

Inf

IntA

Lat

LatPC

2&3

1

SMV

IntDL

sol

psf

Bregma -11.00 mm

1

psf

Lat

das

Y

scp

g7

rs

ts

pd

mlf

CI

icp

Lat

IntA

A4

Y

Inf

Pr

4V

SMV

PPy

ias

Bregma -10.80 mm

4V

LR4V

icp

scp

jx

rs

LPGi

DPGi

IS

Lat

LVe

Pr

g7

6

mlf

CI

8vn

bas

SMV

A5

MVeMC

Bregma -10.04 mm

4V

8cn

tz

ml

bas

SPO

Tz

MSO

IS

LPBV

Methoxamine (50 µg)plus U69,593 (0.178 µg)

Fig. 5. Histological results corresponding to rats that received microinjec-tion of MTX (50mg in 0.5ml) plus U69,593 (0.178mg in 0.5ml) into theRVM. Representations of five coronal sections through the rat RVM areshown in sequence from anterior to posterior. The filled squares show theapproximate positions of the cannula tips corresponding to rats in the MTX/U69,593 treatment group (n�13). Adapted from Paxinos and Watson.38

Fig. 4. Nociceptive and place-conditioning data corresponding to rats thatreceived microinjection of U69,593 (0.178mg in 0.5ml) into the RVM.Vertical bars indicate S.E.M. Percentage of time spent in each compartmenton the pre- vs post-conditioning test days is shown in A, and TF latenciesproduced by each microinjection treatment during the first conditioning

trial are shown in B.

U69,593 into the pyramidal tracts ventral to the RVM (Fig. 7).When data from these sites were pooled and analysed in theirown right, the analysis revealed that the “off-site” microinjec-tions did not produce either a significant change in spinalnociceptive reactivity (Fig. 8B; pairedt-test,P, 0.05) or asignificant place-conditioning effect (Fig. 8A; pairedt-test,P. 0.05).

Microinjection of different doses of methoxamine or U69,593into the rostral ventromedial medulla

The data in Fig. 6 suggest that pharmacological activationof ON cells combined with pharmacological inhibition ofOFF cells results in activation of spinal pain transmissionneurons and is aversive to the animal. Thus, the productionof a negative affective state can be brought about by pharma-cological manipulation of RVM neurons. Having determined,then, that a negative affective state can be elicited fromthe RVM, we sought to examine the effects of additionalintra-RVM doses of U69,593 alone and MTX alone, withthe view of obtaining preliminary dose–response analysesof these drugs on spinal nociceptive reactivity and affectivestate.

Histology.Histological examination showed that cannulaswere correctly implanted into the RVM in eight rats treatedwith intra-RVM U69,593 (0.356mg; Fig. 9) and five ratstreated with intra-RVM MTX (12.5mg; Fig. 11). Data fromthese rats were included for statistical analysis.

Nociceptive reactivity.Intra-RVM U69,593 (0.356mg)did not produce a statistically significant change in TFlatencies compared with intra-RVM vehicle (Fig. 10B; pairedt-test, P. 0.05). While intra-RVM microinjection of MTX(12.5mg) produced a trend towards a reduction in TFlatencies, the reduction did not reach statistical significance(Fig. 12B; pairedt-test,P. 0.05).

Affective changes: place conditioning.U69,593 (0.356mg).On the pre-conditioning day (i.e. before drug treatments), themean percentage of time spent by rats in the conditionedcompartment (compartment paired with intra-RVM U69,593),unconditioned compartment (i.e. compartment paired withintra-RVM vehicle) and neutral compartment was 36.8^2.0%, 32.8̂ 3.0% and 30.5̂ 3.2%, respectively (Fig. 10A).There were no statistically significant differences in percentageof time spent by rats in each of these compartments (one-factorrepeated measures ANOVA,F2,14�0.863,P. 0.05), indicat-ing that rats did not prefer any one compartment to the others onthe pre-conditioning day. On the post-conditioning test day,however (after the nine-day conditioning phase in whichintra-RVM U69,593 was paired with the conditioned compart-ment), there was a statistically significant decrease in thepercentage of time spent in the conditioned compartmentcompared with time spent in the same compartment on thepre-conditioning day (Fig. 10A; 19.1̂ 5.3% post-conditioningvs 36.8̂ 2.0% pre-conditioning; pairedt-test, P, 0.05). Inother words, intra-RVM microinjection of U69,593 (0.356mg)produced CPA.

Methoxamine (12.5mg). On the pre-conditioning day (i.e.

N. Hirakawaet al.866

Bregma -11.60 mm

1

IntA

IntDM

IntPPC

icp

sp5

sol

RPa

B4

C3

mlf

8n

IS

ECu

DC

LatPC

rs

LPGi

4V

SolM

SolIM

RPa

Bregma -11.30 mm

IntDM

4V

4V

A4

8n

sp5

LR4V

rs

7

PPy

ml

RMg

ROb

GiA

7VM

7DI

7VI

7DL

7L

Sp5O

DMSp5V

VCP

X

SpVe

LVe

Y

Inf

LatPC

P7

SolRL

prf

4&5

pcuf

Sim

IntA

IntDM

jx das

4V

4V

A4

LR4V

rs

vsc

PPy

bas

RPa

mlf

CI

asc7

DC

X

SpVe

LVe

VeCb

MVePC

MVeMC

Pr

Y

Inf

IntA

Lat

LatPC

2&3

1

SMV

IntDL

sol

psf

Bregma -11.00 mm

1

Lat

das

Y

scp

g7

rs

ts

pd

mlf

CI

icp

Lat

IntA

A4

Y

Inf

Pr

4V

SMV

PPy

ias

Bregma -10.80 mm

4V

LR4V

icp

scp

jx

rs

LPGi

DPGi

IS

LVe

Pr

g7

6

mlf

CI

8vn

bas

SMV

A5

MVeMC

8cn

tz

ml

bas

SPO

Tz

MSO

IS

Methoxamine (50 µg)plus U69,593 (0.178 µg)

Fig. 7. Histological results corresponding to rats that received “off-site”microinjections of MTX (50mg in 0.5ml) plus U69,593 (0.178mg in0.5ml). Representations of four coronal sections through the rat RVM areshown in sequence from anterior to posterior. The filled squares show theapproximate positions of the cannula tips corresponding to rats in the “off-site” MTX/U69,593 treatment group (n�8). Adapted from Paxinos and

Watson.38

Fig. 6. Nociceptive and place-conditioning data corresponding to rats thatreceived microinjection of MTX (50mg in 0.5ml) plus U69,593 (0.178mgin 0.5ml) into the RVM. Vertical bars indicate S.E.M. Percentage of timespent in each compartment on the pre- vs post-conditioning test days isshown in A, and TF latencies produced by each microinjection treatmentduring the first conditioning trial are shown in B. (A) *P, 0.05, pairedt-test, compared with percentage of time spent in the conditioning compart-ment on the pre-conditioning test day. (B) *P, 0.05, paired t-test,

compared with U69,593 microinjection alone.

before drug treatments), the mean percentage of time spent byrats in the conditioned compartment (compartment pairedwith intra-RVM MTX), unconditioned compartment (i.e.compartment paired with intra-RVM saline) and neutralcompartment was 24.8̂ 6.1%, 24.2̂ 3.9% and 51.0̂6.5%, respectively (Fig. 12A). Clearly, rats spent a signifi-cantly higher percentage of time in the neutral compartmentvs the other two compartments (one-factor repeated measuresANOVA, F2,8�4.95,P, 0.05), indicating that they preferredthe neutral compartment to the other two compartments on thepre-conditioning day. On the post-conditioning test day,however (after the nine-day conditioning phase in whichintra-RVM MTX was paired with the conditioned compart-ment), there was a statistically significant increase in thepercentage of time spent in the conditioned compartmentcompared with time spent in the same compartment on thepre-conditioning day (Fig. 12A; 53.7̂ 7.0% post-condition-ing vs 24.8̂ 6.1% pre-conditioning; pairedt-test,P, 0.05).In other words, intra-RVM microinjection of MTX (12.5mg)produced CPP.

DISCUSSION

Summary of results

Co-administration of thea1-adrenergic receptor agonistMTX (50 mg) and thek-opioid receptor agonist U69,593(0.178mg) into the RVM resulted in both hyperalgesia onthe rat TF assay and CPA (Fig. 6). These data support ourhypothesis (see Introduction) that pharmacological manipula-tion of RVM neurons can elicit a “virtual” nociceptive (pain?)signal by activating nociceptive neurons in the spinal dorsalhorn. However, results from additional experiments usingintra-RVM microinjections of U69,593 alone or MTX alonehighlighted (i) the bi-directionality of the RVM’s influence onaffective state and (ii) the fact that RVM-induced affectivechanges can be dissociated from RVM-induced changes inspinal nociceptive transmission. Intra-RVM microinjectionof MTX (50 mg) alone resulted in hyperalgesia on the TFassay, but produced no place-conditioning effect (Fig. 2).By contrast, intra-RVM microinjection of U69,593 (0.356mg)produced no change in spinal nociceptive reactivity butproduced CPA (Fig. 10). Finally, intra-RVM microinjectionof MTX (12.5mg) produced no change in spinal nociceptivereactivity but produced CPP (Fig. 12), a finding that suggeststhat the RVM is capable of eliciting a positive affective statethat is rewarding to the animal (see below for further discus-sion). Thus, using the place-conditioning paradigm we haveprovided the first evidence that pharmacological manipulationof RVM neurons can elicit bi-directional changes in the affec-tive state of the conscious, freely moving animal.

Tail flick testing during conditioning as a potential confoundto the present results

As stated in the Introduction, the purpose of the presentexperiments was to test the ability of the RVM to modifythe affective state of the animal in the absence of a peripheralnoxious stimulus. Our place-conditioning apparatus wasdesigned, however, to allow brief TF testing on each ratduring the first conditioning trial in each of the conditioningcompartments (A and B; see Experimental Procedures). Thiswas done in order to monitor the effect of each intra-RVMdrug treatment on spinal nociceptive reactivity, so that this

RVM, pain and affect 867

1

IntA

IntDM

IntPPC

icp

sol

B4

C3

mlf

8n

IS

ECu

DC

LatPC

4V

SolM

SolIM

RPa

Bregma -11.30 mm

IntDM

4V

4V

A4

8n

sp5

LR4V

rs

7

PPy

ml

RMg

ROb

GiA

7VM

7DI

7VI

7DL

7L

Sp5O

DMSp5V

VCP

X

SpVe

LVe

Y

Inf

LatPC

P7

SolRL

prf

4&5

pcuf

Sim

IntA

IntDM

jx das

4V

4V

A4

LR4V

rs

vsc

PPy

bas

RPa

mlf

CI

asc7

DC

X

SpVe

LVe

VeCb

MVePC

MVeMC

Pr

Y

Inf

IntA

Lat

LatPC

2&3

1

SMV

IntDL

sol

psf

Bregma -11.00 mm

1

psf

Lat

das

Y

scp

g7

rs

ts

pd

mlf

CI

icp

Lat

IntA

A4

Y

Inf

Pr

4V

SMV

PPy

ias

Bregma -10.80 mm

4V

LR4V

icp

scp

jx

rs

LPGi

DPGi

IS

Lat

LVe

Pr

g7

6

mlf

CI

8vn

bas

SMV

A5

MVeMC

Bregma -10.04 mm

4V

8cn

tz

ml

bas

SPO

Tz

MSO

IS

LPBV

U69,593 (0.356 µg)

Fig. 9. Histological results corresponding to rats that received microinjec-tion of U69,593 (0.356mg in 0.5ml) into the RVM. Representations of fourcoronal sections through the rat RVM are shown in sequence from anteriorto posterior. The filled triangles show the approximate positions of thecannula tips corresponding to rats in the U69,593 treatment group

(n�8). Adapted from Paxinos and Watson.38

Fig. 8. Nociceptive and place-conditioning data corresponding to rats thatreceived “off-site” microinjections of MTX (50mg in 0.5ml) plus U69,593(0.178mg in 0.5ml). Vertical bars indicate S.E.M. Percentage of time spentin each compartment on the pre- vs post-conditioning test days is shown inA, and TF latencies produced by each microinjection treatment during the

first conditioning trial are shown in B.

could be correlated with any place-conditioning effectobserved on the final test day.

Given this experimental design, however, it could beargued that the application of noxious heat to the tail duringthe first conditioning trial, not the intra-RVM drug treatment,produced the CPA observed in Figs 6 and 10. An obviouscounterargument to this position would be that TF testingoccurred in both conditioning compartments (i.e. thecompartment paired with the intra-RVM drug treatment andthe compartment paired with the intra-RVM vehicle treat-ment), thereby making it unlikely that one conditioningcompartment was made more “aversive” than the other bythe presence of tail stimulation. One might argue further,however, that a particular intra-RVM drug treatment mighthave resulted in the tail stimulation being either more or lesspainful, causing the animal to feel either more or less“distressed” in the compartment paired with intra-RVMdrug than in the compartment paired with intra-RVM vehicle.Even if this were the case, we would argue the following. (1)The TF assay is a threshold measurement—both the intensityand duration of the noxious stimulation are very low. In thepresent experiments, TF testing occurred for a total of 15–20 s (including all three TF trials) during the entire 30 minconditioning session. (2) TF testing occurred during only the

N. Hirakawaet al.868

Fig. 10. Nociceptive and place-conditioning data corresponding to rats thatreceived microinjection of U69,593 (0.356mg in 0.5ml) into the RVM.Vertical bars indicate S.E.M. Percentage of time spent in each compartmenton the pre- vs post-conditioning test days is shown in A, and TF latenciesproduced by each microinjection treatment during the first conditioningtrial are shown in B. (A) *P, 0.05, pairedt-test, compared with percentageof time spent in the conditioning compartment on the pre-conditioning test

day.

Bregma -11.60 mm

1

IntA

IntDM

IntPPC

icp

sp5

sol

RPa

B4

C3

mlf

8n

IS

ECu

DC

LatPC

rs

LPGi

4V

SolM

SolIM

RPa

IntDM

4V

4V

A4

8n

sp5

LR4V

rs

7

PPy

ml

RMg

ROb

GiA

7VM

7DI

7VI

7DL

7L

Sp5O

DMSp5V

VCP

X

SpVe

LVe

Y

Inf

LatPC

P7

SolRL

prf

4&5

pcuf

Sim

IntA

IntDM

jx das

4V

4V

A4

LR4V

rs

vsc

bas

RPa

mlf

CI

asc7

DC

X

SpVe

LVe

VeCb

MVePC

MVeMC

Pr

Y

Inf

IntA

Lat

LatPC

2&3

1

SMV

IntDL

sol

psf

Bregma -11.00 mm

1

Lat

das

Y

scp

g7

rs

ts

pd

mlf

CI

icp

Lat

IntA

A4

Y

Inf

Pr

4V

SMV

PPy

ias

Bregma -10.80 mm

4V

LR4V

icp

scp

jx

rs

LPGi

DPGi

IS

LVe

Pr

g7

6

mlf

CI

8vn

bas

SMV

A5

MVeMC

4V

8cn

tz

bas

SPO

Tz

IS

Methoxamine (12.5 µg)

Fig. 11. Histological results corresponding to rats that received microinjec-tion of MTX (12.5mg in 0.5ml) into the RVM. Representations of threecoronal sections through the rat RVM are shown in sequence from anteriorto posterior. The filled circles show the approximate positions of thecannula tips corresponding to rats in the MTX treatment group (n�5).

Adapted from Paxinos and Watson.38

Fig. 12. Nociceptive and place-conditioning data corresponding to rats thatreceived microinjection of MTX (12.5mg in 0.5ml) into the RVM. Verticalbars indicate S.E.M. Percentage of time spent in each compartment on thepre- vs post-conditioning test days is shown in A, and TF latenciesproduced by each microinjection treatment during the first conditioningtrial are shown in B. (A) *P, 0.05, pairedt-test, compared with percentageof time spent in the conditioning compartment on the pre-conditioning test

day.

first of three conditioning sessions in each conditioningcompartment, making it unlikely that rats formed a significantassociation between tail pain and a particular compartment.(3) In several experiments, RVM-elicited place-conditioningeffects were observed without concomitant changes in spinalnociceptive reactivity (Figs 10, 12).

Contributions of the various cell classes in the rostralventromedial medulla to the present effects

Our choices of drugs for the present studies were based onsome assumptions regarding the specificity of their actions.Since we had evidence that activation ofa1-adrenergicreceptors in the RVM selectively activates ON cells,22 weused thea1 receptor agonist methoxamine in an attempt toactivate ON cells and produce “virtual” nociception. Similarly,our use of thek-opioid receptor agonist U69,593 was basedon previous evidence indicating that OFF cells are selectivelyinhibited byk agonists.31,37 It could be argued, however, thatwe cannot be absolutely certain that, at the concentrationsused, each of these drugs had a selective action on a particularRVM cell class. Nevertheless, we feel that this uncertaintydoes not detract from our principal finding that bi-directionalchanges in affective state can be elicited from the RVM.Future experiments will be designed to determine accuratelythe relative contributions of ON, OFF and neutral cells toRVM-elicited affective changes.

Dose-dependent effects of methoxamine in the rostralventromedial medulla

Intra-RVM microinjection of MTX (50mg) produced anincrease in spinal nociceptive reactivity (i.e. hyperalgesiaon the TF assay), but no place-conditioning effect (Fig. 2).By contrast, intra-RVM microinjection of MTX (12.5mg)produced no effect on spinal nociceptive reactivity whileproducing CPP (Fig. 12). Although the mechanisms relatingto the disparate, dose-related effects of MTX are not clear, it ispossible that the selectivity of MTX is relevant to this issue.While MTX is a highly selective agonist fora1- vs a2-adrenoceptors, it is necessary to point out that thea1 receptorcomprises three subtypes (a1A, a1B, a1D). MTX binds to eachof these subtypes with varying affinities (see Ref. 45). Impor-tantly, in situhybridization studies indicate that mRNA corre-sponding to each of the three subtypes is expressed in rat brain(see Ref. 45). At present, however, it is not possible to knowthe precise neuroanatomical distribution of the receptorproteins due to the lack of specific antibodies for all thesubtypes. Nevertheless, it is likely that more than onea1

receptor subtype exists in the RVM. Varying the dose ofintra-RVM MTX may result in differential activation patternsof a1 receptor subtypes in this region. This, in turn, could leadto differential activation of rostrally vs caudally projectingfibers (see below), thereby accounting for the disparateplace-conditioning and pain-modulating effects observed inthe present experiments.

Global importance of neuronal activity in the rostralventromedial medulla to animal behavior

The place-conditioning paradigm permits assessment of theprimary motivational properties of a manipulation such as adrug treatment.43 Generally, it is accepted that if an animalavoids or prefers an environment with which the manipulation

is paired, this directly reflects the acute aversive or rewardingproperties of the manipulation, respectively. The rewarding oraversive properties of the manipulation are, in turn, a directresult of the influence of the manipulation on the affectivestate of the animal. If the manipulation generates a “positive”affective state (i.e. a state that is pleasurable), the manipula-tion is rewarding to the animal. If the manipulation produces a“negative” affective state (e.g., as a result of nociception), themanipulation is aversive to the animal. In this context, thefinding that pharmacological manipulation of RVM neuronscan result in either CPA or CPP constitutes the first report ofbi-directional changes in affective state associated with abrain area traditionally associated with bi-directional painmodulation. Another pain-modulating region in the brain-stem—the PAG (see Ref. 15)—has been shown at varioustimes to be capable of eliciting either a rewarding (e.g., Ref.9) or an aversive (e.g., Ref. 12) signal. However, thesedisparate effects have not been observed by a single groupof experimenters using a single experimental paradigm (aswas the case in the present experiments) and have not beencorrelated with changes in nociception.

Although relatively caudal brainstem regions such as thepedunculopontine tegmental nucleus4,36and locus coeruleus39

have been implicated in reward processes, to date the RVM isthe most caudal brain area in the neuraxis shown to be capableof eliciting affective changes. Importantly, the present datareveal that RVM-induced affective changes can occur in theabsence of ongoing nociceptive input from the periphery.While it is not difficult to envision the RVM influencing ananimal’s affective state by modifying the salience orperceived intensity of a noxious stimulus (see Introduction),the ability of the RVM to have such bi-directional effects inthe absence of significant noxious peripheral input is intri-guing and has important implications for the influence ofthis region on the behavior of the normal, uninjured animal.

Dissociation of pain-modulating and place-conditioningeffects

As indicated in the Introduction, we hypothesized that, inthe absence of significant nociceptive input from the periph-ery, RVM ON cells are capable of producing “virtual noci-ception” by activating, via their descending projections,nociceptive projection neurons in the spinal cord. While notaddressing the issue of ON cells specifically, the data shownin Fig. 6 are consistent with the idea that an aversive noci-ceptive signal (pain?) can be elicited from the RVM—intra-RVM microinjection of MTX combined with intra-RVMmicroinjection of U69,593 produced both an increase inspinal nociceptive responsiveness and CPA (Fig. 6).

The data in Fig. 6 notwithstanding, it is clear that RVM-induced place-conditioning effects and RVM-induced pain-modulating effects can be dissociated. Place-conditioningeffects were observed without concomitant changes in spinalnociceptive reactivity (Figs 10, 12) and vice versa (Fig. 1).Clearly, the RVM is capable of modifying the affective stateof the animal without affecting spinal nocifensor reflexesand vice versa. Taken together, the data from the presentexperiments suggest that the RVM may be capable ofmodifying an animal’s affective state through either itsdescending projections to the spinal dorsal horn1,3,17 orits extensive ascending connections with numerous higherbrain centers.7,8,10,41,42,44This, in turn, suggests that possibility

RVM, pain and affect 869

that, under some circumstances, RVM-elicited CPA reflectsthe production of a “virtual pain” signal, and under othercircumstances it does not. The fact that RVM-elicited CPAcan be achieved without a concomitant increase in spinalnociceptive reactivity indicates that RVM-elicited CPA canreflect a negative affective state (e.g., resulting from “discom-fort”) that is not dependent on activation of nociceptiveprojection neurons in the spinal cord. Indeed, the situationis analogous to the effects of electrical stimulation of thePAG in humans:25 depending on the stimulation parameters,stimulation of the PAG can evoke either (i) pain relief,(ii) pain relief accompanied by novel feelings of vagueunpleasantness, (iii) feelings of fear and terror accompaniedby novel, diffuse pain, or (iv) feelings of fear and terror with-out novel pain. Nevertheless, the unresolved issue of whetherRVM-elicited CPA ever reflects a “virtual pain” signal doesnot detract from our observation that the RVM canprofoundly influence animal behavior by eliciting changesin the animal’s affective state.

Reward, analgesia and the rostral ventromedial medulla

Figure 12 shows an unexpected result indicating that theRVM is capable of producing CPP. From this, it can beinferred that, in addition to being capable of eliciting a nega-tive affective state, the RVM is capable of eliciting a “posi-tive” affective state in the absence of significant noxiousperipheral input. The simplest explanation for this result isthat the RVM is capable of activating a CNS “reward”system. One well-described system comprises mesolimbicdopaminergic projections from the ventral tegmental area tothe nucleus accumbens in addition to circuit components inthe ventral pallidum and peduncupontine tegmental nucleus,2

and is intimately linked with learning mechanisms that resultin reinforcement of particular behaviors (e.g., as in theproduction of CPP). As is probably the case for some formsof RVM-induced aversion, the RVM may achieve activationof reward circuitry by way of its ascending connections withvarious forebrain targets.7,8,10,41,42,44

It is well established that drugs of abuse such as opioids,cannabinoids and psychomotor stimulants activate rewardcircuitry, lending credence to the notion that this sort of actionis a causal factor in the development of drug addiction.27

Interestingly, it has not escaped notice that the most effectiveanalgesics also possess high abuse potential,19,20 promptingsome investigations into the potential overlap betweenmechanisms of analgesia and mechanisms of reward.Although these investigations have not provided strong

evidence of a mechanistic link between these two phenom-ena,11,35 it is intriguing that the RVM, a brain area tradition-ally associated with pain modulation, is capable of eliciting apositive affective state (Fig. 12). Although RVM-inducedCPP was produced in the present experiments without con-comitant antinociception at the spinal level (Fig. 12), itremains possible that the brain’s reward and analgesiasystems are linked at the level of the RVM in a manner thatwas not captured fully by the present behavioral paradigms.

Clinical implications of “virtual pain”

One of our stated goals was to gain insight into the bio-logical significance of RVM ON cell activation (see Introduc-tion). We hypothesized that, in the absence of significantnoxious input from the periphery, a pharmacological manipu-lation of the RVM that normally results in increased spinalnociceptive reactivity is sufficient to generate a “virtual pain”signal. Although not addressing the role of ON cell activationdirectly, the data in Fig. 6 suggest that it is possible to manip-ulate RVM neuronal activity and produce both an increase inspinal nociceptive responsiveness and a negative affectivestate (reflected by CPA) that may be indicative of the genera-tion of a pain signal.

For the normal animal, modulation of RVM neuronal activ-ity may represent one means by which higher brain centersgenerate changes in an animal’s affective state in response toappropriate environmental stimuli. Since, in this context, anegative affective state may be generated by the productionof “virtual pain”, the possibility should be considered thatabnormal or prolonged changes in RVM neuronal activitycontribute to the pathogeneses of chronic pain states. Thefact that RVM neuronal activity can be influenced bydescending projections from a variety of forebrain, includingcortical, regions23,24 suggests a potential physiological basisfor certain kinds of human pain syndromes, including thosethat are thought of as “psychogenic”. Such a circuit provides aphysiological basis for the influence of psychological factors(e.g., depression, anxiety) on the development of chronic painconditions.13,14 We suggest that future conceptualizations ofchronic pain states should include a potential contributionfrom the RVM.

Acknowledgements—This work was supported by grants from the USPublic Health Service (NS 21445, NS 07265), the Wheeler Center forthe Neurobiology of Addiction and the Medical Research Council ofCanada. N.H. was supported by a postdoctoral fellowship from theMinistry of Education, Science and Culture, Japan.

REFERENCES

1. Abols I. A. and Basbaum A. I. (1981) Afferent connections of the rostral medulla of the cat: a neural substrate for midbrain–medullary interactionsin themodulation of pain.J. comp. Neurol.201,285–297.

2. Bardo M. T. (1998) Neuropharmacological mechanisms of drug reward: beyond dopamine in the nucleus accumbens.Crit. Rev. Neurobiol.12,37–67.3. Basbaum A. I. and Fields H. L. (1979) The origin of descending pathways in the dorsolateral funiculus of the spinal cord of the cat and rat: further studies

on the anatomy of pain modulation.J. comp. Neurol.187,513–531.4. Bechara A. and van der Kooy D. (1989) The tegmental pedunculopontine nucleus: a brain-stem output of the limbic system critical for the conditioned

place preferences produced by morphine and amphetamine.J. Neurosci.9, 3400–3409.5. Bederson J. B., Fields H. L. and Barbaro N. M. (1990) Hyperalgesia during naloxone-precipitated withdrawal from morphine is associated with increased

on-cell activity in the rostral ventromedial medulla.Somatosensory Motor Res.7, 185–203.6. Bitner R. S., Nikkel A. L., Curzon P., Arneric S. P., Bannon A. W. and Decker M. W. (1998) Role of the nucleus raphe magnus in antinociception

produced by ABT-594: immediate early gene responses possibly linked to neuronal nicotinic acetylcholine receptors on serotonergic neurons.J. Neurosci.18, 5426–5432.

7. Bobillier P., Seguin S., Petitjean F., Salvert D., Touret M. and Jouvet M. (1976) The raphe nuclei of the cat brain stem: a topographical atlas of theirefferent projections as revealed by autoradiography.Brain Res.113,449–486.

N. Hirakawaet al.870

8. Bowker R. M. (1986) The relationship between descending serotonin projections and ascending projections in the nucleus raphe magnus: a doublelabeling study.Neurosci. Lett.70, 348–353.

9. Cazala P. (1990) Dose-dependent effects of morphine differentiate self-administration elicited from lateral hypothalamus and mesencephalic central grayarea in mice.Brain Res.527,280–285.

10. Chiba T. and Murata Y. (1985) Afferent and efferent connections of the medial preoptic area in the rat: a WGA–HRP study.Brain Res. Bull.14,261–272.11. Clarke P. B. S. and Franklin K. B. J. (1992) Infusions of 6-hydroxydopamine into the nucleus accumbens abolish the analgesic effect of amphetaminebut

not of morphine in the formalin test.Brain Res.580,106–110.12. Di Scala G. and Sandner G. (1989) Conditioned place aversion produced by microinjections of semicarbazide into the periaqueductal gray of the rat.

Brain Res.483,91–97.13. Fields H. (1991) Depression and pain: a neurobiological model.Neuropsychiat. Neuropsychol. behav. Neurol.4, 88–92.14. Fields H. L. (1997) Pain modulation and headache. InHeadache(eds Goadsby P. J. and Silberstein S. D.). Butterworth-Heinemann, Boston.15. Fields H. L. and Basbaum A. I. (1999) Central nervous system mechanisms of pain modulation. InTextbook of Pain(eds Wall P. D. and Melzack R.).

Churchill Livingstone, Edinburgh.16. Fields H. L., Heinricher M. M. and Mason P. (1991) Neurotransmitters in nociceptive modulatory circuits.A. Rev. Neurosci.14, 219–245.17. Fields H. L., Malick A. and Burstein R. (1995) Dorsal horn projection targets of ON and OFF cells in the rostral ventromedial medulla.J. Neurophysiol.

74, 1742–1759.18. Fields H. L., Vanegas H., Hentall I. D. and Zorman G. (1983) Evidence that disinhibition of brain stem neurones contributes to morphine analgesia.

Nature306,684–686.19. Franklin K. B. (1998) Analgesia and abuse potential: an accidental association or a common substrate?Pharmac. Biochem. Behav.59, 993–1002.20. Franklin K. B. J. (1989) Analgesia and the neural substrate of reward.Neurosci. Biobehav. Rev.35, 157–163.21. Hargreaves K., Dubner R., Brown F., Flores C. and Joris J. (1988) A new and sensitive method for measuring thermal nociception in cutaneous

hyperalgesia.Pain 32, 77–88.22. Heinricher M. M., Haws C. M. and Fields H. L. (1988) Opposing actions of norepinephrine and clonidine on single pain-modulating neurons in the rostral

ventromedial medulla. InProceedings of the Vth World Congress on Pain (Pain Research and Clinical Management)(eds Dubner R., Gebhart G. F. andBond M. R.). Elsevier, Amsterdam.

23. Hermann D. M., Luppi P. H., Peyron C., Hinckel P. and Jouvet M. (1996) Forebrain projections of the rostral nucleus raphe magnus shown byiontophoretic application of cholera toxin b in rats.Neurosci. Lett.216,151–154.

24. Hermann D. M., Luppi P. H., Peyron C., Hinckel P. and Jouvet M. (1997) Afferent projections to the rat nuclei raphe magnus, raphe pallidus andreticularis gigantocellularis pars alpha demonstrated by iontophoretic application of cholera toxin (subunit b).J. chem. Neuroanat.13, 1–21.

25. Iacono R. P. and Nashold B. S., Jr (1982) Mental and behavioral effects of brain stem and hypothalamic stimulation in man.Human Neurobiol.1,273–279.

26. Kaplan H. and Fields H. L. (1991) Hyperalgesia during acute opioid abstinence: evidence for a nociceptive facilitating function of the rostral ventro-medial medulla.J. Neurosci.11, 1433–1439.

27. Koob G. F. and Le Moal M. (1997) Drug abuse: hedonic homeostatic dysregulation.Science278,52–58.28. Manning B. H. (1998) A lateralized deficit in morphine antinociception after unilateral inactivation of the central amygdala.J. Neurosci.18,9453–9470.29. Mayer D. J. and Manning B. H. (1995) The role of opioid peptides in environmentally-induced analgesia. InThe Pharmacology of Opioid Peptides(ed.

Tseng L. F.). Harwood Academic, Chur, Switzerland.30. McGaraughty S., Reinis S. and Tsoukatos J. (1993) Two distinct unit activity responses to morphine in the rostral ventromedial medulla of awake rats.

Brain Res.604,331–333.31. Meng I. D. and Fields H. L. (1998) Inhibition of OFF-cell activity in the rostral ventromedial medulla after microiontophoresis of the kappa agonist

U50,488.Soc. Neurosci. Abstr.24, 889.32. Meng I. D., Manning B. H., Martin W. J. and Fields H. L. (1998) An analgesia circuit activated by cannabinoids.Nature395,381–383.33. Mitchell J. M., Lowe D. and Fields H. L. (1998) The contribution of the rostral ventromedial medulla to the antinociceptive effects of systemic morphine

in restrained and unrestrained rats.Neuroscience87, 123–133.34. Morgan M. M., Heinricher M. M. and Fields H. L. (1992) Circuitry linking opioid-sensitive nociceptive modulatory systems in periaqueductal grayand

spinal cord with rostral ventromedial medulla.Neuroscience47, 863–871.35. Olmstead M. C. and Franklin K. B. J. (1993) Effects of pedunculopontine tegmental nucleus lesions on morphine-induced conditioned place preference

and analgesia in the formalin test.Neuroscience57, 411–418.36. Olmstead M. C., Munn E. M., Franklin K. B. and Wise R. A. (1998) Effects of pedunculopontine tegmental nucleus lesions on responding for intravenous

heroin under different schedules of reinforcement.J. Neurosci.18, 5035–5044.37. Pan Z. Z., Tershner S. A. and Fields H. L. (1997) Cellular mechanism for anti-analgesic action of agonists of thek-opioid receptor.Nature389,382–385.38. Paxinos G. and Watson C. (1997)The Rat Brain in Stereotaxic Coordinates. Academic, New York.39. Ritter S. and Stein L. (1973) Self-stimulation of noradrenergic cell group (A6) in locus coeruleus of rats.J. comp. physiol. Psychol.85, 443–452.40. Sagen J. and Proudfit H. K. (1985) Evidence for pain modulation by pre- and postsynaptic noradrenergic receptors in the medulla oblongata.Brain Res.

331,285–293.41. Takagi H., Senba E., Shiosaka S., Sakanaka M., Inagaki S., Takatsuki K. and Tohyama M. (1981) Ascending and cerebellar non-serotonergic projections

from the nucleus raphe magnus of the rat.Brain Res.206,161–165.42. Takagi H., Shiosaka S., Tohyama M., Senba E. and Sakanaka M. (1980) Ascending components of the medial forebrain bundle from the lower brain stem

in the rat, with special reference to raphe and catecholamine cell groups. A study by the HRP method.Brain Res.193,315–337.43. Tzschentke T. M. (1998) Measuring reward with the conditioned place preference paradigm: a comprehensive review of drug effects, recent progress and

new issues.Prog. Neurobiol.56, 613–672.44. Vertes R. P. (1984) A lectin horseradish peroxidase study of the origin of ascending fibers in the medial forebrain bundle of the rat. The lower brainstem.

Neuroscience11, 651–668.45. Zhong H. and Minneman K. P. (1999)a1-Adrenoceptor subtypes.Eur. J. Pharmac.375,261–276.

(Accepted28 June2000)

RVM, pain and affect 871