R

Tp

SMa

b

c

d

e

a

ARRAA

KPHIMMA

1

entaPtb

i

S0

0d

Behavioural Brain Research 226 (2012) 32– 40

Contents lists available at SciVerse ScienceDirect

Behavioural Brain Research

j ourna l ho me pa ge: www.elsev ier .com/ locate /bbr

esearch report

he periaqueductal gray as a critical site to mediate reward seeking duringredatory hunting

andra Regina Mota-Ortiza,b, Marcia Harumi Sukikaraa,c, Jackson Cioni Bittencourtb,arcus Vinícius Baldod, Carol Fuzeti Eliasb,e, Luciano Freitas Felicioc, Newton Sabino Canterasb,∗

Laboratory of Neural Basis of Behavior, City University of Sao Paulo, UNICID, Sao Paulo, SP 03071-000, BrazilDepartment of Anatomy, Institute of Biomedical Sciences, University of Sao Paulo, Sao Paulo, SP 05508-900, BrazilDepartment of Pathology, School of Veterinary Medicine, University of Sao Paulo, Sao Paulo, SP 05508-900, BrazilDepartment of Physiology and Biophysics, Institute of Biomedical Sciences University of Sao Paulo, Sao Paulo, SP 05508-900, BrazilDepartment of Internal Medicine, Division of Hypothalamic Research, University of Texas Southwestern Medical Center, Dallas, TX 75390-9077, USA

r t i c l e i n f o

rticle history:eceived 29 July 2011eceived in revised form 20 August 2011ccepted 23 August 2011vailable online 30 August 2011

eywords:AGypothalamic orexinergic cells

nsect huntingorphineotivational drive

a b s t r a c t

Previous studies using morphine-treated dams reported a role for the rostral lateral periaqueductal gray(rlPAG) in the behavioral switching between nursing and insect hunting, likely to depend on an enhancedseeking response to the presence of an appetitive rewarding cue (i.e., the roach). To elucidate the neuralmechanisms mediating such responses, in the present study, we first observed how the rlPAG influencespredatory hunting in male rats. Our behavioral observations indicated that bilateral rlPAG NMDA lesionsdramatically interfere with prey hunting, leaving the animal without chasing or attacking the prey, butdo not seem to affect the general levels of arousal, locomotor activity and regular feeding. Next, usingPhaseolus vulgaris-leucoagglutinin (PHA-L), we have reviewed the rlPAG connection pattern, and pointedout a particularly dense projection to the hypothalamic orexinergic cell group. Double labeled PHA-L andorexin sections showed an extensive overlap between PHA-L labeled fibers and orexin cells, revealingthat both the medial/perifornical and lateral hypothalamic orexinergic cell groups receive a substan-

ppetitive reinforcers tial innervation from the rlPAG. We have further observed that both the medial/perifornical and lateralhypothalamic orexinergic cell groups up-regulate Fos expression during prey hunting, and that rlPAGlesions blunted this Fos increase only in the lateral hypothalamic, but not in the medial/perifornical,orexinergic group, a finding supposedly associated with the lack of motivational drive to actively pursuethe prey. Overall, the present results suggest that the rlPAG should exert a critical influence on rewardseeking by activating the lateral hypothalamic orexinergic cell group.

. Introduction

The PAG has been commonly recognized as an exit relay for thexpression of a variety of behaviors, including sexual [1], mater-al [2,3] and a vast array of defensive responses [4–6], as well ashe accompanying modulation of nociceptive transmission [5,7],utonomic changes [5,8], and vocalization [9,10]. By and large, theAG-related responses have been regarded as being rather stereo-yped and are thought to depend on descending projections to the

rainstem and spinal cord.

On the other hand, there is a growing body of evidence suggest-ng that the PAG should also influence the motivational drive to

∗ Corresponding author at: Department of Anatomy, Institute of Biomedicalciences, University of Sao Paulo, Avenida Professor Lineu Prestes, 2465, CEP: 05508-00, São Paulo, SP, Brazil. Tel.: +55 01130917628; fax: +55 01130918449.

E-mail address: newton@icb.usp.br (N.S. Canteras).

166-4328/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.bbr.2011.08.034

© 2011 Elsevier B.V. All rights reserved.

perform certain adaptive responses. In line with this view, studiesusing intracranial self-administration and intracranial place con-ditioning have shown that the PAG is one of the brain sites thatcan support the rewarding effects of morphine [11–13]. Moreover,studies from our laboratory brought up the suggestion of an inte-grative role for the PAG in influencing the selection of adaptivebehavioral responses [14]. It is well known that administration ofmorphine disrupts maternal behavior during lactation. We haveobserved that a certain level of morphine-induced activation inthe rostral lateral PAG (rlPAG), located in the outer half of thelateral column at the levels of the oculomotor nucleus, seems tobe required to inhibit maternal behavior, and that this activationdepends on a direct effect of the morphine on the rlPAG [15].Curiously, the rlPAG activation pattern, seen in dams in response

to morphine treatment, is closely similar to the one occurringduring predatory hunting behavior [16], suggesting the idea thatpredatory or foraging activity would be favored during morphine-induced inhibition of maternal responses. We were able to prove

ral Br

trbiTbtam

eiepvrcm

2

2

etaaNPaoIt

2

hbsItrpa

2

wNma

2

wac

baa[swoc−a

aiat

S.R. Mota-Ortiz et al. / Behaviou

his hypothesis by testing morphine-treated dams in an envi-onment containing pups and roaches, and further showed thatilateral N-methyl-d-aspartate (NMDA) lesions placed in the rlPAG

mpaired predatory hunting and restored maternal behavior [14].hese results suggested an important role for the rlPAG in theehavioral switching between nursing and insect hunting, likelyo depend on an enhanced seeking response to the presence ofn appetitive rewarding cue (i.e., the roach). However, the neuralechanisms mediating such responses are still poorly understood.To this end, in the present study, we first observed how bilat-

ral NMDA lesions placed in the rlPAG influence predatory huntingn male rats. Next, to understand the possible paths mediating theffects on predatory hunting, we reviewed the rlPAG projectionattern and explored the impact of the rlPAG lesions on the acti-ation of the hypothalamic orexinergic cell group, one of the mainlPAG targets. Overall, the results suggest a key role for the rlPAG inontrolling reward seeking behavior, and unveil the putative pathsediating this effect.

. Materials and methods

.1. Animals and housing

Adult male Wistar rats (n = 24), weighing 250–320 g at the beginning of thexperiments, obtained from the local São Paulo breeding facilities, were used inhe present study. The animals were kept under controlled temperature (23 ◦C)nd illumination (12 h cycle) in the animal quarters, and had free access to waternd standard laboratory diet. Experiments were carried out in accordance with theational Institutes of Health Guide for the Care and Use of Laboratory Animals (NIHublications No. 80-23, 1996). All experimental procedures had been previouslypproved by the Committee on Care and Use of Laboratory Animals of the Institutef Biomedical Sciences —University of São Paulo, Brazil (Protocol number 084/2005).n the present study, we attempted to minimize the number of animals used andheir suffering.

.2. Experimental apparatus and procedure

One week before the experimental procedures, animals were individuallyoused into a plexiglas cage (50 cm × 35 cm × 16 cm), and were handled repeatedlyy the same investigator who had conducted the behavioral tests. The hunting ses-ions were carried out between 9:00 and 12:00 h, during the light phase of the cycle.n the hunting session, animals were induced to hunt by a simultaneous introduc-ion, into the hunting cage, of five mature intact cockroaches (Periplaneta americana),aised for this purpose in our laboratory. The behavior was observed during a 15-mineriod after the roaches had been placed into the cage, and videotaped for furthernalysis.

.3. Behavior analysis

Behaviors were scored by a trained observer using the ethological analysis soft-are “The Observer” (version 5.0; Noldus Information Technology, Wageningen, Theetherlands). For the behavior analysis of predatory hunting, we have first deter-ined the latency to start hunting (i.e., the actual time taken to grab the first prey),

nd carefully examined the motor pattern to capture, hold and kill the prey.

.3.1. Experiment 1In experiment 1, we examined the changes in the hunting behavior of animals

ith bilateral lesions in a PAG region located in the outer half of the lateral columnt the levels of the oculomotor nucleus, i.e., the rostral lateral PAG (rlPAG), andompared the results with those from sham-lesioned animals.

For the lesion procedure, rats were deeply anesthetized with sodium pentobar-ital (Cristália, Itapira, SP, Brazil; 40 mg/kg i.p.) and were placed in a stereotaxicpparatus. Bilateral iontophoretic deposits of a 0.15 M solution of N-methyl-d-spartate (NMDA, Sigma, St. Louis, MO, USA) were placed in the rlPAG of 10 animalscoordinates: 2.9 mm rostral to the interaural line; 0.65 mm from the midline of theagital sinus; 4.5 mm ventral to the surface of the brain]. In addition, saline injectionsere placed in other 6 rats (sham-lesioned group). NMDA deposits were produced

ver 15 min through a glass micropipette (30 �m tip diameter), using a constant-urrent device (model CS3, Midgard Electronics, Canton, MA, USA) set to deliver10 �A, with 7-s pulse and interpulse durations. Animals recovered for 2 weeksfter surgery, before the predatory hunting session.

One day before the predatory hunting session both sham- and rlPAG-lesionednimals were submitted to a locomotor activity analysis. These animals were placedn the center of the open field arena and the number of line crossing, in the centernd peripheral areas, was recorded during 5 min. The open field apparatus used inhe present study was a round metal arena (95 cm in diameter × 30 cm high) divided

ain Research 226 (2012) 32– 40 33

into 25 squares to enable measurement of activity (line crossing). Moreover, duringthe 15-day period preceding the behavioral testing, both sham- and rlPAG-lesionedanimals had their weight gain measured.

Ninety minutes after the hunting session, animals were deeply anesthetizedwith sodium pentobarbital (40 mg/kg i.p.) and perfused transcardially with a solu-tion of 4.0% paraformaldehyde in 0.1 M phosphate buffer at pH 7.4; the brains wereremoved and left overnight in a solution of 20% sucrose in 0.1 M phosphate bufferat 4 ◦C. The brains were then frozen and five series of 40 �m-thick sections were cutwith a sliding microtome in the frontal/transverse plane. To delineate the extent ofthe NMDA lesions, one series was processed for immunohistochemistry with NeuNantibody (Neuronal nuclei, clone A60, MAB377, Chemicon, Millipore, USA; dilution1:1000), at the entire rostro-caudal extent of the PAG. The primary antiserum waslocalized using a variation of the avidin–biotin complex system. In brief, sectionswere incubated for 90 min at room temperature in a solution of biotinylated goatanti-mouse IgG (Vector Laboratories, Burlingame, CA, USA; dilution 1:200), and thenplaced in the mixed avidin–biotin horseradish peroxidase (HRP) complex solution(ABC Elite Kit; Vector Laboratories) for the same period of time. The peroxidase com-plex was visualized by a 10-min exposure to a chromogen solution containing 0.02%3,3-diaminobenzidine tetrahydrochloride (DAB, Sigma, St Louis, MO, USA) with 0.3%nickel–ammonium sulfate in 0.05 M Tris–buffer (pH 7.6), followed by incubationfor 10 min in chromogen solution with hydrogen peroxide (1:3000), to produce ablue-black product. The reaction was stopped by extensive washing in potassiumphosphate-buffered saline (KPBS; pH 7.4). Sections were mounted on gelatin-coatedslides, and then dehydrated and coverslipped with DPX (Sigma). An adjacent serieswas always stained with Thionin to serve as a reference series for cytoarchitectonicpurposes.

2.3.2. Experiment 2In experiment 2, we have investigated the projections of the rlPAG. Five animals

received a single injection of Phaseolus vulgaris-leucoagglutinin (PHA-L, Vector Lab-oratories) into the rlPAG. First, they were anesthetized with a mixture of ketamineand xylazine (v/v; 1 ml/kg body weight), and then the iontophoretic injection ofa 2.5% solution of PHA-L in 0.1 M sodium phosphate-buffered saline (pH 7.4) wasmade over a 10-min period through a stereotaxically positioned glass micropipette(10 �m tip diameter) by applying a +5 �A current, pulsed at 7-s intervals, witha constant-current source (Midgard Electronics). After a survival time of 14–16days, the animals were perfused, and the brains processed as described for theNeuN immunohistochemistry. One series of sections was processed for immuno-histochemistry with an antiserum directed against PHA-L (Dako, Carpinteria, CA,USA) at a dilution of 1:5000, and the antigen–antibody complex was localized byusing a biotinylated goat anti-rabbit IgG (Vector Laboratories; dilution 1:200) fol-lowing the procedure described for the NeuN immunohistochemistry. The sectionswere mounted on gelatin-coated slides and then treated with osmium tetroxide toenhance visibility of the reaction product. Slides were then dehydrated and cover-slipped with DPX. An adjacent series was always stained with Thionin to serve as areference for cytoarchitecture.

A third series was processed for double PHA-L and orexin immunohistochem-istry. First, the sections were incubated with a rabbit antibody directed againstPHA-L (Dako; dilution 1:5000), and the antigen–antibody complex was localizedby employing DAB and nickel–ammonium sulfate as chromogens, to produce ablue-black reaction product. Next, the sections were incubated with the orexin anti-sera (Phoenix Pharmaceuticals, Inc., Belmont, CA; dilution 1:10,000) and processedusing only DAB as chromogen, to result in a brown staining of the immunoreactivecells.

2.3.3. Experiment 3In experiment 3, we examined the Fos expression in orexin cells and compared

control animals, which are not allowed to hunt with the ones that had performedpredatory hunting, both from the sham- and rlPAG-lesioned groups described inSection 2.3.1. Control animals (n = 06) were handled in exactly the same way as theanimals that had performed predatory hunting, but on the test day, they were notallowed to hunt (i.e., the animals were not exposed to the roaches). Their brainswere processed as described for the sham- and rlPAG-lesioned animals, in Section2.3.1.

The sections were processed for double Fos and orexin immunohistochem-istry. First, they were incubated with anti-Fos antiserum raised in rabbit (Ab-5,Calbiochem, San Diego, CA, USA; lot # D09803; dilution 1:20,000), and theantigen–antibody complex was localized by employing DAB and nickel–ammoniumsulfate as chromogens, to produce a blue-black reaction product. Next, the sectionswere incubated with the orexin antisera (Phoenix Pharmaceuticals, Inc.; dilution1:10,000) and processed using only DAB as chromogen, to result in a brown stain-ing of the immunoreactive cells. An adjacent series was always stained with thioninto serve as a reference for cytoarchitecture.

Counts of the number of orexin and double Fos + orexin immunoreactive cellswere evaluated by an observer without knowledge of the animal’s experimental sta-

tus, and were generated by using the 10× objective of a Nikon Eclipse 80i microscopeequipped with a camera lucida. The quantification of orexin and double Fos + orexinlabeled cells was performed in a series of five 200 �m-apart sections, correspondingto the levels 27–31 of The Brain Maps [17]. The orexin cells have been divided into twogroups, namely, the medial/perifornical cell group and the lateral hypothalamic cell

3 ral Br

glcslo

sAct

2

hty6tpTadboS

cotm

3

3

aiawpoctu

tpirtfbp((

shstpmpptftn

4 S.R. Mota-Ortiz et al. / Behaviou

roup, and, to trace the borders between these cell groups, we followed the parcel-ation described by Swanson [17]. Accordingly, the medial/perifornical orexinergicell group comprises the anterior part of the dorsomedial nucleus and the juxtador-omedial and suprafornical regions of the lateral hypothalamic area, whereas theateral hypothalamic orexinergic cell group includes the dorsal and ventral regionsf the lateral hypothalamic area.

The figures were prepared for publication by using the Adobe Photoshop (ver-ion 4.0; Adobe Systems, Mountain View, CA, USA) for photomicrographs and thedobe Illustrator (version 10.0; Adobe Systems) for line drawings. Only sharpness,ontrast, and brightness were adjusted. Unless otherwise indicated, parcellation ofhe brain regions follows Swanson [17].

.4. Statistical analysis

In Section 2.3.1, because some of the tested animals did not express predatoryunting during the 15-min observation period, instead of using the actual time takeno start capturing the roaches, we chose to score these latencies for statistical anal-sis as follows: 1, 0–150 s; 2, 151–300 s; 3, 301–450 s; 4, 451–600 s; 5, 601–750 s;, 751–900 s; 7, > 900 s. This transformation allowed us to include, in our analysis,hose animals that did not present predatory hunting within the 15-min observationeriod, thus lacking a definite measure (the seventh rank in the present analysis).he scores given to the time taken to start capturing the roaches for the sham-nd rlPAG-lesioned groups were entered into a nonparametric test for indepen-ent groups (Mann–Whitney U-test). In addition, in Section 2.3.1, the comparisonetween sham- and rlPAG-lesioned animals for the number of line crossings, in thepen field test, and for the percentage of weight gain was analyzed using a two-tailedtudent’s t-test for independent groups.

In Section 2.3.3, the data on the number of orexin and double Fos + orexin labeledells were analyzed using a one-way ANOVA. As a significant univariate test had beenbtained, we performed a post hoc analysis (Tukey’s HSD test) to isolate the respec-ive effect. The significance level was set at = 5%. Average results are expressed as

ean ± SEM throughout the text.

. Results

.1. Experiment 1

In experiment 1, we analyzed the predatory hunting of sham-nd rlPAG-lesioned animals. The parameters described for NMDAontophoretic injections yielded relatively small PAG lesions, char-cterized by neuronal loss (Fig. 1). In seven animals, the lesionsere centered bilaterally in the lateral part of the PAG, encom-assing the outer half of the lateral column at the levels of theculomotor nucleus, and extending, to a smaller degree, to adja-ent parts of the mesencephalic reticular nucleus (Fig. 1). Notably,hese lesions were centered in the PAG region previously shown top-regulate Fos expression during insect hunting [16].

Before considering the results on predatory hunting, it is impor-ant to note that the percentage of weight gain, during the 15-dayeriod elapsed from the surgical procedure to the behavioral test-

ng, did not differ between the sham- (n = 06; 4.98% ± 0.26) andlPAG-lesioned animals (n = 07; 4.38% ± 0.18) (p = 0.151, Student’s-test), suggesting that PAG lesions did not affect the regular choweeding. Likewise, the locomotor activity, tested as the total num-er of line crossings recorded in an open field apparatus one dayrevious to the behavioral testing, did not differ between the sham-n = 06; 83.8 ± 7.3) and rlPAG-lesioned groups (n = 07; 97.4 ± 11.2)p = 0.335, Student’s t-test).

During predatory hunting, behavioral analysis revealed thatham-lesioned animals started chasing the prey shortly after theyad been delivered into the testing box, taking less than 18 s totart chasing the prey, orienting themselves very efficiently towardhe moving prey, while trying to capture them. The capture waserformed with the mouth, assisted by the forepaws. These ani-als caught the prey very efficiently, and afterwards, held the

rey firmly with the forepaws and delivered the killing bite, rip-ing off the roaches’ head. The rats usually took the killed roaches

o a corner of the cage and tried to conceal the captured preyrom other potential predators (dodging behavior) while eatinghem voraciously. In sharp contrast, rlPAG-lesioned animals didot attempt to chase and hunt the roaches during the 15-min

ain Research 226 (2012) 32– 40

observation period. Thus, compared to the sham-lesioned group,rlPAG lesioned animals presented a dramatic increase in the latencyto start chasing the roaches (Mann–Whitney U-test, z = −3.61;p < 0.001). Occasionally, some of the rlPAG lesioned animals couldgrab prey moving nearby, without attempting to chase them. Dur-ing the 15-min observation period, rlPAG-lesioned animals weremostly engaged in behaviors other than hunting, including groom-ing, general exploratory activity and resting.

3.2. Experiment 2

The results of Section 3.1 support the idea that the rlPAG isa critical element to influence predatory hunting. To understandthe potential paths involved in this control, in experiment 2, wehave reviewed the efferent connections of rlPAG. Material was ana-lyzed from three experiments in which the PHA-L injection-labeledneurons were mostly confined to the rlPAG, coinciding with thePAG region particularly activated during insect hunting and whereNMDA lesions resulted in a clear deficit in the predatory behavior(see Section 3.1). In all of these experiments, a very similar pat-tern of anterogradely labeled fibers were observed, and of these,we chose experiment rlPAGPHAL#3 as a prototype to illustrate ourresults, because the injection in that experiment labeled the mostextensive population of cells in the rlPAG (Fig. 2A). Overall, thepattern of projection presently found for the rlPAG resembles inmany ways the one previously described by Cameron et al. [18,19],reporting the projection of a PHA-L injection centered in the lat-eral PAG, at the level of the oculomotor nucleus, which had beentaken as a representative case centered in the rostral ventrolateralPAG. Therefore, in the present report, we will provide only a briefdescription of the general rlPAG projection pattern, and will focusour analysis on the targets potentially related to predatory hunt-ing. Labeled axons from the rlPAG may be divided into an ascendinggroup that innervates targets rostral to the nucleus in the rostralmesencephalon, diencephalon and telencephalon, and a descend-ing group to the brainstem. As could be gleaned from the work ofCameron et al. [18,19] with PHA-L injection centered in the lateralPAG, we have presently found that descending fibers from the rlPAGproject to brainstem regions involved in autonomic control (i.e.,ventrolateral PAG, A5 noradrenergic cell group, nucleus ambiguus,and medullary rostroventrolateral reticular region), as well as inregulating the sleep/wake cycle and behavioral arousal, namely,the pedunculopontine nucleus, the rostral part of reticular pon-tine nucleus, the dorsal raphe nucleus, the laterodorsal tegmentalnucleus, the locus coeruleus, and the subcoeruleal region. More-over, in line with what had been previously described for thelateral PAG [19], descending projections from the rlPAG also pro-vide sparse to moderate inputs to the mesencephalic reticularnucleus, the retrorubral area, the ventral part of the caudal pon-tine nucleus, the magnocellular reticular nucleus, and the nucleusraphe magnus, in addition to strong projections to the lateral partof the parabrachial nucleus.

In general agreement with previous findings by Cameron et al.[18], ascending fibers from the rlPAG course were initially throughthe midbrain periventricular system. At rostral mesencephalic lev-els, a small contingent of these fibers take a ventral course andprovide a rather sparse projection to the ventral tegmental area,particularly aimed at the parabrachial pigmented nucleus [20](Fig. 2B). Part of these fibers takes a lateral course to project tothe compact part of substantia nigra and adjacent parvicellularpart of the subparafascicular nucleus. At the transition betweenmesencephalon and diencephalon, fibers ascending through the

midbrain periventricular system may be divided into two path-ways, i.e., a dorsal path projecting to thalamic targets and a ventralone coursing through the subthalamic and hypothalamic regions.Labeled fibers coursing through the dorsal thalamus provide sparse

S.R. Mota-Ortiz et al. / Behavioural Brain Research 226 (2012) 32– 40 35

Fig. 1. Experiment 1: NMDA lesion appearance. Photomicrograph of NeuN immunostained section of the PAG, illustrating the extent and appearance of a bilateral lesion inthe rlPAG from a representative case in experiment 1. Scale bars: 200 �m. AQ: cerebral aqueduct; III: oculomotor nucleus; MRN: midbrain reticular nucleus; PAG dm, dl, l:periaqueductal gray, dorsomedial, dorsolateral, lateral parts; RN: red nucleus.

Fig. 2. Experiment 2. (A) Brightfield photomicrograph, to illustrate the appearance of a PHA-L injection site for a representative injection localized in the rlPAG (experimentrlPAGPHAL#3). (B–D) Representative darkfield photomicrographs showing the distribution pattern of PHA-L immunoreactive axons in the ventral tegmental area (B), theintralaminar thalamic central medial nucleus (C), and the lateral hypothalamic area (D). Scale bars: (A) and (D), 200 �m; (B) and (C), 100 �m. AQ: cerebral aqueduct; CM:central medial nucleus thalamus; DMH: dorsomedial hypothalamic nucleus; fx: fornix; III: oculomotor nucleus; IMD: intermediodorsal nucleus thalamus; LHA: lateralhypothalamic area; ml: medial lemniscus; mp: mammillary peduncle; PAG dm, dl, l: periaqueductal gray, dorsomedial, dorsolateral, lateral parts; RH: rhomboid nucleus;SMT: submedial nucleus thalamus; VMH: ventromedial hypothalamic nucleus; VTA: ventral tegmental area; ZI: zona incerta.

36 S.R. Mota-Ortiz et al. / Behavioural Brain Research 226 (2012) 32– 40

Fig. 3. Experiment 2. (A–C) A series of camera lucida drawings of double PHA-L and orexin immunostained sections, arranged from rostral to caudal (A–C), illustrating thedistribution of PHA-L labeled fibers in the hypothalamic region containing orexinergic cells (represented as red dots). For interpretation of the references to color in text, thereader is referred to the web version of the article. The numbers on the top right refers to Brain Maps [17] plate numbers, to indicate the approximate rostro-caudal levels. (D)Photomicrograph, to illustrate the overlap between PHA-L labeled fibers and orexin cells. Note the presence of PHA-L labeled terminal boutons in straight juxtapposition toorexinergic cells. Scale bars: (A–C) 200 �m; (D) 25 �m. cpd: cerebral pedeuncle; DMHa, p, v: dorsomedial hypothalamic nucleus, anterior, posterior, ventral parts; fx: fornix;LHAd: lateral hypothalamic area, dorsal region; LHAjd: lateral hypothalamic area, juxtadorsomedial region; LHAjv: lateral hypothalamic area, juxtaventromedial region;LHAs: lateral hypothalamic area, suprafornical region; LHAsfp: lateral hypothalamic area, subfornical region, posterior zone; LHAvl: lateral hypothalamic area, ventral region,l mmilh

tAinpatncn

sa

ateral zone; LHAvm: lateral hypothalamic area, ventral region, medial zone; mtt: maypothalamic nucleus; vtt: ventrolateral hypothalamic tract; ZI: zona incerta.

o moderate inputs to a number of midline and intralaminar nuclei.t caudal thalamic levels, the rlPAG projects moderately to the

ntermediodorsal nucleus and more sparsely to the parafascicularucleus. Proceeding rostrally, fibers coursing through the dorsalath project preferentially to the central medial nucleus, where

moderate plexus of labeled axons were found (Fig. 2C), in addi-ion to providing relatively sparse projections to other intralaminaruclei, including the paracentral and central lateral nuclei, and toertain midline nuclei, such as the intermediodorsal and reuniens

uclei.

Ascending fibers coursing through the ventral path form a sub-tantial projection field in the ventral part of the zona incertand adjacent parts of the lateral hypothalamic area (Fig. 2D). At

lothalamic tract; opt: optic tract; sup: supraoptic commissures; VMH: ventromedial

tuberal levels, this projection extends to several fields of the lateralhypothalamic area, including the ventromedial, dorsal, suprafor-nical and juxtadorsomedial regions (Fig. 3A–C), which coincidewith the general region containing orexinergic cells. Given thathypothalamic orexinergic neurons have been implicated with thepreferences for cues associated with food reward (such as the prey,in the present case) and food seeking behavior [21,22], we haveexamined the relationship between the distribution of the lateralhypothalamic orexin cells and the projecting fibers from the rlPAG.

Examining the experiments with PHA-L deposits confined to therlPAG, double labeled PHA-L and orexin sections, at tuberal lev-els of the lateral hypothalamic area, revealed an extensive overlapbetween PHA-L labeled fibers and orexin cells (Fig. 3A–C), and a

S.R. Mota-Ortiz et al. / Behavioural Brain Research 226 (2012) 32– 40 37

F unostam ic oreS

cP

vllzttttIlao

e[toat

3

cafgutmdpow[ppd

ig. 4. Experiment 3: A series of photomicrographs of double Fos and orexin immedial/perifornical orexinergic group (upper row) and from the lateral hypothalam

cale bars: 50 �m.

loser observation showed that most of the orexin cells containedHA-L labeled bouton appositions (Fig. 3D).

At anterior hypothalamic levels, fibers coursing through theentral path provide a rather sparse projection chiefly aimed at theateral aspects of the anterior hypothalamic nucleus and adjacentateral hypothalamic fields, including the intermediate and ventralones of the anterior lateral hypothalamic region. Proceeding ros-rally, at preoptic levels, ascending rlPAG fibers project sparsely tohe lateral preoptic area, and a few of these fibers proceed dorsallyo project to the posterior division of the bed nuclei of the striaerminalis, particularly to the interfascicular and transverse nuclei.n addition, a small number of fibers coursing along the hypotha-amus may enter the ansa peduncularis and extend laterally to themygdala, where a few fibers appear to project to the medial partf the central amygdalar nucleus.

Therefore, apart from confirming previous findings on the gen-ral rlPAG projection pattern, as described by Cameron et al.18,19], we have presently expanded the analysis on the projec-ion to LHA, particularly to the region containing the hypothalamicrexin group, which represents one of the main targets of the rlPAG,nd where a substantial number of rlPAG fibers establish close bou-on appositions with the orexin labeled cells.

.3. Experiment 3

In this experiment, we examined the Fos expression in orexinells and compared control animals not allowed to hunt (i.e.,nimals not exposed to the roaches) with the ones that per-ormed predatory hunting, both from the sham- and rlPAG-lesionedroups, as previously described (see Section 2). As far as the pop-lation of orexin cells is concerned, one-way ANOVA revealedhat the number of orexin immunoreactive cells both in the

edial/perifornical and the lateral hypothalamic cell groups did notiffer significantly among the experimental groups [F(2,16) = 2.91;

> 0.08] (Figs. 4 and 5A, B). Regarding the number of doublerexin + Fos labeled cells, in the medial/perifornical cell group, one-ay ANOVA showed differences among the experimental groups

F(2,16) = 8.5; p = 0.003] (Figs. 4 and 5C). Post hoc pairwise com-arison revealed that the groups of animals that had performedredatory hunting (i.e., sham-lesioned and rlPAG-lesioned groups)id not differ in the number of double labeled cells (Tukey post hoc

ined sections from the control, sham- and rlPAG-lesioned groups, taken from thexinergic group (lower row). Arrow heads indicate double Fos + Orexin labeled cells.

test, p = 0.96). However, when compared to the control group notallowed to hunt, both sham- and rlPAG-lesioned animals presenteda significant increase in the number of double orexin + Fos labeledcells (Tukey post hoc test, p < 0.01). In the lateral hypothalamicorexinergic cell group, the experimental groups also differed in thenumber of double labeled cells [one-way ANOVA, F(2,16) = 69.56;p < 0.0001] (Figs. 4 and 5D). In this case, however, post hoc pair-wise comparison revealed that only the sham-lesioned animalspresented a significant increase in the number of double labledcells (Tukey post hoc test, p = 0.00016), and, remarkably, the rlPAG-lesioned animals failed to increase the number of orexin + Foslabeled cells, and did not differ from the control animals (Tukeypost hoc test, p = 0.986).

4. Discussion

In the present work, we report a series of studies combin-ing behavioral observation, NMDA lesions, and neuroanatomicalexperiments supporting a role for the PAG in influencing themotivational drive to hunt, an effect seemingly mediated by itsascending projections to the lateral hypothalamic orexinergic cellgroup.

In male rats, relatively circumscribed rlPAG NMDA lesions wereable to produce a dramatic change in prey hunting, namely, the ani-mal did not chase or attack the prey. As previously reported, rlPAGNMDA lesions were also able to blunt prey hunting in morphine-treated dams, which have been shown to present a clear preferenceof forage or hunting over maternal care. In this previous study, wehave further investigated the PAG site specificity for this effect,and observed that only lateral, but not dorsal or ventrolateral, PAGlesions were able to influence predatory hunting [14].

Similarly to what had been observed in sham-lesioned animals,rlPAG lesioned animals did not seem to present any deficit in han-dling the prey, as could be observed in rlPAG lesioned animals thatoccasionally grabbed and held prey moving close nearby. Likewise,sham- and rlPAG-lesioned animals performed equally well in theopen field test, suggesting no interference in the general levels of

arousal and locomotor activity. In addition, since the percentage ofweight gain did not differ between sham- and NMDA lesioned ani-mals, regular chow feeding does not seem to be affected by rlPAGlesions. In fact, predatory hunting and regular feeding appear to

38 S.R. Mota-Ortiz et al. / Behavioural Brain Research 226 (2012) 32– 40

Fig. 5. Experiment 3: Frequency histograms showing the number of orexin labeled cells (A) and (B) and the number of double Fos + orexin labeled cells (C) and (D) in them ic cellc

bmaab

edial/perifornical (A) and (C) and the lateral hypothalamic (B) and (D) orexinergontrol group, p < 0.01.

e mediated by different circuits. Regular food intake is largely

ediated by a circuit centered around the arcuate nucleus and the

utonomic part of the paraventricular nucleus [22], both of whichre known to present clear Fos increase during regular feeding,ut not during prey hunting [23]. Altogether, our observations in

groups. Data are expressed as mean ± SEM. *Differs significantly compared to the

the open field test and weight gain clearly suggested that rlPAG

lesions do not seem to interfere with the general levels of arousal,locomotor activity and regular feeding.

To understand how the rlPAG lesions would be able to causethe predatory hunting changes observed, we examined the rlPAG

ral Br

cttLwipab

octlapliamawaoimt[

soslImiasnmsdpfpdrte

ttgvfndpolafaaif

S.R. Mota-Ortiz et al. / Behaviou

onnection pattern. In the present investigation, the general pat-ern of projections found for the rlPAG resembles in many wayshose previously described by Cameron et al. [18,19] with PHA-

deposits centered in the rostral part of the lateral PAG. Here,e were able to refine this analysis for the targets with potential

nterest for the predatory behavior. Of particular interest, the rlPAGrojects to a number of brain sites related to arousal control, as wells to brain regions underlying the motivational drive to seekingehavior.

Regarding the regions involved in behavioral arousal, we havebserved, in the brainstem, that the rlPAG projects to the pedun-ulopontine nucleus, the rostral part of reticular pontine nucleus,he dorsal raphe nucleus, the laterodorsal tegmental nucleus, theocus coeruleus, and the subcoeruleal region [24–27]. In addition,t tuberal levels of the hypothalamus, the rlPAG provides a massiverojection to the suprafornical and juxtadorsomedial regions of the

ateral hypothalamic area containing the medial/perifornical orex-nergic cell group, which shows diurnal changes consistent with

role in the production or maintenance of arousal [28]. In theedial/perifornical orexinergic cell group, we were able to identify

n extensive overlap between PHA-L labeled fibers and orexin cells,hich presented an impressive number of PHA-L labeled bouton

ppositions, likely to form synapses on the orexinergic cells. More-ver, in the thalamus, the rlPAG has a projection to midline andntralaminar nuclei, particularly aimed to the intralaminar central

edial nucleus, which has been thought as participating in the con-rol of cortical arousal supporting multimodal sensory processing29].

From what we have just discussed, the rlPAG connection patternupports a potential role in modulating arousal levels. However,ur behavioral observations indicate that rlPAG lesions do noteem to affect general arousal levels, since sham- and rlPAG NMDAesioned animals performed equally well in the open field test.n line with this idea, the increased Fos expression seen in the

edial/perifornical orexinergic cell group, during predatory hunt-ng, does not seem to be affected by rlPAG lesions. This way,lternative paths to the medial/perifornical orexinergic cell group,uch as those previously described from the prefrontal cortex, beduclei of the stria terminalis, lateral septum, posterior hypothala-ic nucleus and supramammillary nucleus [30], may potentially

upport the activation of the arousal-related orexinergic cell groupuring predatory hunting. Hence, although the anatomical findingsoint to a potential rlPAG role in the arousal control, the presentunctional evidence suggests that rlPAG lesions do not seem toroduce any important arousal deficit. This fact is opposed to theramatic behavioral change seen on the predatory hunting afterlPAG lesions, leaving the motivational drive to hunt as, perhaps,he key element to explain the predatory behavior deficits in ourxperiments.

As pointed above, the rlPAG projects to brain sites underlyinghe motivational drive to seeking behavior, including the ven-ral tegmental area and the lateral hypothalamic orexinergic cellroup. According to the present results, the rlPAG appears to pro-ide only a sparse input to the ventral tegmental area, where aew labeled fibers could be found in the parabrachial pigmenteducleus. Similar results could also be confirmed with fluoro-goldeposits in the ventral tegmental [31]. In sharp contrast, the rlPAGrovides rather massive inputs to the dorsal and ventral regionsf the lateral hypothalamic area containing the lateral hypotha-amic orexinergic cell group, a finding also suggested in previousnatomical studies [30]. In the present study, as already discussedor the medial/perifornical orexinergic cell group, we were also

ble to identify an extensive overlap between PHA-L labeled fibersnd the lateral hypothalamic orexinergic cells, which presented anmpressive number of PHA-L labeled bouton appositions, likely toorm synaptic contacts on the orexinergic cells.

ain Research 226 (2012) 32– 40 39

We have further seen that predatory hunting up-regulatesFos expression in the lateral hypothalamic orexinergic cell group,an effect that is seemingly blunted by rlPAG lesions. The lat-eral hypothalamic orexergic neurons have been implicated inreward seeking and are likely to be activated by stimuli associ-ated with positive reinforcements [32,33]. Enhanced activation ofthe lateral hypothalamus orexinergic neurons has been reported inresponse to highly salient appetitive reinforcers, which may under-lie the motivational drive to selectively seek drugs or highly salientfoods [34], such as the roaches in the present experiments. Takentogether, the present findings support the view that, during insecthunting, the rlPAG would act as a key element to convey to the lat-eral hypothalamic orexinergic cell group the appetitive reinforcingproperties of the prey, likely to influence the motivational drive tohunt.

A study on the pattern of afferent inputs to the rlPAG revealsa broader role for the rlPAG in controlling the decision-makingprocess between hunting, foraging, or even approaching a givenrewarding cue, and other behavioral responses [35]. The rlPAG isseemingly importantly driven by medial prefrontal cortical areasinvolved in controlling decision-making processes [35]. In addition,the rlPAG receives a wealth of information from different neu-ral sites related to feeding (via the visceral and gustatory corticalareas, and the central amygdalar nucleus, likely to convey infor-mation related to food hedonic values), drinking (via the medianpreoptic nucleus, conveying information related to plasma osmo-larity) and hunting (via the lateral part of the superior colliculus,conveying information related to prey displacement) [35]. Of par-ticular relevance in the present context, part of the afferent inputsto the rlPAG utilizes opioidergic neurotransmission, such as thosefrom the anterior part of the anterior hypothalamic nucleus and themedial part of the central amygdalar nucleus; and we have previ-ously shown, in lactating rats, that systemically injected morphinehas a direct effect on the rlPAG, providing a rlPAG activation seem-ingly critical for the switching from maternal to hunting behavior[15]. Overall, the rlPAG is in a position to convey appetitive reinforc-ing properties from salient cues, as well as serve as a nodal point totransmit decision-making information to the lateral hypothalamicorexinergic cell group.

In the present study, we have shown that the rlPAG appears toprovide equally dense inputs to both medial/perifornical and lat-eral hypothalamic orexinergic cell groups, however, only the latterfailed to show enhanced Fos expression during predatory huntingin animals bearing rlPAG lesions. Importantly, this finding suggeststhat the rlPAG, compared to the other sources of inputs to theorexin cells, should exert a prominent role in activating the lateralhypothalamic orexinergic cell group, and hence critically influencereward seeking.

Overall, the present results support the view of the rlPAGacts as a key element in controlling reward seeking behav-ior via the lateral hypothalamic orexinergic cell group. Furtherstudies are obviously needed to explore whether this rlPAGrole in reward seeking, presently found for prey hunting, couldalso be extended to social-related rewarding cues or even drugseeking.

Acknowledgements

This research was supported by grants from Fundac ão deAmparo à Pesquisa do Estado de São Paulo (FAPESP, no. 05/59286-4;

04/13849-5) awarded to N.S.C and J.C.B., and Conselho Nacional deDesenvolvimento Científico e Tecnológico awarded to N.S.C, L.F.F.,J.C.B. and M.V.B. S.R.M.O. and M.H.S. and were supported by FAPESPfellowships (FAPESP, no. 01/14053-1; 04/14312-5; 06/57647-2).

4 ral Br

R

[

[

[

[

[

[

[

[[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

0 S.R. Mota-Ortiz et al. / Behaviou

eferences

[1] Sakuma Y, Paff DW. Facilitation of female reproductive behavior from mesen-cephalic central gray in the rat. Am J Physiol 1979;237:78–284.

[2] Lonstein JS, Stern JM. Role of the midbrain periaqueductal gray in maternalnurturance and aggression: c-fos and electrolytic lesion studies in lactatingrats. J Neurosci 1997;17:3364–78.

[3] Lonstein JS, Stern JM. Site and behavioral specificity of periaqueductal graylesions on postpartum sexual, maternal, and aggressive behaviors in rats. BrainRes 1998;804:21–35.

[4] Fanselow MS. The midbrain periaqueductal gray as a coordinator of actionin response to fear and anxiety. In: Depaulis A, Bandler R, editors. The mid-brain periaqueductal gray matter: functional, anatomical and neurochemicalorganization. New York: Plenum Press; 1991. p. 151–73.

[5] Bandler R, Shipley MT. Columnar organization of the midbrain periaque-ductal gray: modules for emotional expression. Trends Neurosci 1994;17:379–89.

[6] Cezario AF, Ribeiro-Barbosa ER, Baldo MV, Canteras NS. Hypothalamic sitesresponding to predator threats the – role of the dorsal premammillary nucleusin unconditioned and conditioned antipredatory defensive behavior. Eur J Neu-rosci 2008;28:1003–15.

[7] Besson JM, Fardin V, Oliveras JL. Analgesia produced by stimulation of the peri-aqueductal gray matter: true antinociceptive effects versus stress effects. In:Depaulis A, Bandler R, editors. The midbrain periaqueductal gray matter: func-tional, anatomical and neurochemical organization. New York: Plenum Press;1991. p. 121–38.

[8] Lovick TA. Integrated activity of cardiovascular and pain regulatory systems:role in adaptive behavioral responses. Prog Neurobiol 1993;40:631–44.

[9] Jurgens U. The role of the periaqueductal grey in vocal behavior. Behav BrainRes 1994;62:107–17.

10] Gruber-Dujardin E. Role of the periaqueductal gray in expressing vocal-ization. In: Brudzynski SM, editor. The handbook of mammalian vocal-ization: an integrative neuroscience approach. Amsterdam: Elsevier; 2010.p. 313–27.

11] David V, Cazala P. A comparative study of self-administration of morphineinto the amygdala and the ventral tegmental area in mice. Behav Brain Res1994;65:205–11.

12] Olmstead MC, Franklin KBJ. The development of a conditioned place preferenceto morphine: effects of microinjections into various CNS sites. Behav Neurosci1997;111:1324–34.

13] Van der Kooy D, Mucha RF, O’Shaughnessy M, Bucenieks P. Reinforcing effectsof brain microinjections of morphine revealed by conditioned place preference.Brain Res 1982;243:107–17.

14] Sukikara MH, Mota-Ortiz SR, Baldo MV, Felicio LF, Canteras NS. A role for the

periaqueductal gray in switching adaptive behavioral responses. J Neurosci2006;26:2583–9.

15] Miranda-Paiva CM, Ribeiro-Barbosa ER, Canteras NS, Felicio LF. A role for theperiaqueductal grey in opioidergic inhibition of maternal behaviour. Eur J Neu-rosci 2003;18:667–74.

[

ain Research 226 (2012) 32– 40

16] Comoli E, Ribeiro-Barbosa ER, Canteras NS. Predatory hunting and exposure toa live predator induce opposite patterns of Fos immunoreactivity in the PAG.Behav Brain Res 2003;138:17–28.

17] Swanson LW, Brain Maps. Structure of the rat brain. Amsterdam: Elsevier; 2004.18] Cameron AA, Khan IA, Westlund KN, Cliffer KD, Willis WD. The efferent pro-

jections of the Periaqueductal Gray in rat: a Phaseolus vulgaris-leucoagglutininstudy. I. Ascending projections. J Comp Neurol 1995;351:568–84.

19] Cameron AA, Khan IA, Westlund KN, Cliffer KD, Willis WD. The efferent pro-jections of the Periaqueductal Gray in rat: a Phaseolus vulgaris-leucoagglutininstudy. II. Descending projections. J Comp Neurol 1995;351:568–84.

20] Paxinos G, Watson C. The rat brain: in stereotaxic coordinates. Amsterdam:Elsevier; 2009.

21] Sakurai T. Roles of orexins in the regulation of feeding and arousal. Sleep Med2002;3:S3–9.

22] Saper CB, Chou TC, Elmquist JK. The need to feed: homeostatic and hedoniccontrol of eating. Neuron 2002;36:199–211.

23] Comoli E, Ribeiro-Barbosa ER, Negrão N, Goto M, Canteras NS. Functional map-ping of the prosencephalic systems involved in organizing predatory behaviorin rats. Neuroscience 2005;130:1055–67.

24] Vertes RP, Martin GF. Autoradiographic analysis of ascending projections fromthe pontine and mesencephalic reticular formation and the median raphenucleus in the rat. J Comp Neurol 1988;275:511–41.

25] Vertes RPA. PHA-L analysis of ascending projections of the dorsal raphe nucleusin the rat. J Comp Neurol 1991;313:643–68.

26] Aston-Jones G, Rajkowski J, Cohen J. Role of locus coeruleus in attention andbehavioral flexibility. Biol Psychiatry 1999;46:1309–20.

27] Wang HL, Morales M. Pedunculopontine and laterodorsal tegmental nuclei con-tain distinct populations of cholinergic, glutamatergic and GABAergic neuronsin the rat. Eur J Neurosci 2009;29:340–58.

28] Harris GC, Wimmer M, Aston-Jones G. A role for lateral hypothalamic orexinneurons in reward seeking. Nature 2005;437:556–9.

29] Van der Werf YD, Witter MP, Groenewegen HJ. The intralaminar and midlinenuclei of the thalamus. Anatomical and functional evidence for participation inprocess of arousal and awareness. Brain Res Rev 2002;39:107–40.

30] Yoshida K, McCormack S, Espana RA, Croocker A, Scammell TE. Afferents to theorexin neurons of the rat brain. J Comp Neurol 2006;495:845–61.

31] Geisler S, Zahm D. Afferents of the ventral tegmental area in the rat-anatomicalsubstratum for integrative functions. J Comp Neurol 2005;490:270–94.

32] Harris GC, Aston-Jones G. Arousal and reward: a dichotomy in orexin function.Trends Neurosci 2006;29:571–7.

33] Aston-Jones G, Smith RJ, Moorman DE, Richardson KA. Role of lateral hypotha-lamic orexin neurons in reward processing and addiction. Neuropharmacology2009;56:112–21.

34] Borgland SL, Chang S, Bowers S, Thompson JL, Vittoz N, Floresco SB, et al. Orexin

A/Hypocretin-1 selectively promotes motivation for positive reinforces. J Neu-rosci 2009;29:11215–25.

35] Mota-Ortiz SR, Sukikara MH, Felicio LF, Canteras NS. Afferent connections tothe rostrolateral part of the periaqueductal gray: a critical region influencingthe motivation drive to hunt and forage. Neural Plast 2009. ID 612698.