University of Groningen

Spinal efferents and afferents of the periaqueductal grayMouton, Leonora

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.

Document VersionPublisher's PDF, also known as Version of record

Publication date:1999

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):Mouton, L. J. (1999). Spinal efferents and afferents of the periaqueductal gray: Possible role in pain, sexand micturition [S.l.]: [S.n.]

CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.

Download date: 19-02-2018

21

PAG projections to lamina VIII and medial VII

Chapter 1

The Periaqueductal Gray in the Cat projects to Lamina VIIIand the Medial Part of Lamina VII

throughout the Length of the Spinal Cord

Leonora J. Mouton and Gert Holstege

Exp. Brain Res. 101:253-264 (1994)

AbstractThe periaqueductal gray (PAG) plays an important role in analgesia as well as in motor activi-ties, such as vocalization, cardiovascular changes and movements of the neck, back and hindlimbs. Although the anatomical pathways for vocalization and cardiovascular control are ratherwell understood, this is not the case for the pathways controlling the neck, back and hind limbmovements. This led us to study the direct projections from the PAG to the spinal cord in thecat. In a retrograde tracing study horseradish peroxidase (HRP) was injected into differentspinal levels, which resulted in large HRP-labeled neurons in the lateral and ventrolateralPAG and the adjacent mesencephalic tegmentum. Even after an injection in the S2 spinalsegment a few of these large neurons were found in the PAG. Wheat germ agglutinin-conju-gated HRP injections in the ventrolateral and lateral PAG resulted in anterogradely labeledfibers descending through the ventromedial, ventral and lateral funiculi. These fibers termi-nated in lamina VIII and the medial part of lamina VII of the caudal cervical, thoracic, lumbarand sacral spinal cord. Interneurons in these laminae have been demonstrated to project toaxial and proximal muscle motoneurons. The strongest PAG-spinal projections were to theupper cervical cord, where the fibers terminated in the lateral parts of the intermediate zone(laminae V, VII and the dorsal part of lamina VIII). These laminae contain the premotorinterneurons of the neck muscles.This distribution pattern suggests that the PAG-spinal pathway is involved in the control ofneck and back movements. Comparing the location of the PAG-spinal neurons with the re-sults of these stimulation experiments leads to the supposition that the PAG spinal neuronsplay a role in the control of the axial musculature during threat display.

IntroductionThe mesencephalic periaqueductal gray(PAG) can be considered as a part of the lim-bic system (Holstege, 1990). The PAG is bestknown for its relation to nociception control(Mayer et al., 1971; Liebeskind et al., 1973;Fardin et al., 1984; Oliveras and Besson,1988; Levine et al., 1991), but physiologicalexperiments in rat and cat have shown thatstimulation in the PAG also produces motoractivities, such as vocalization (Kanai andWang, 1962; Jürgens and Pratt, 1979; Larson,1985; Bandler et al., 1991; Jürgens andChang-Lin, 1993), cardiovascular changes(Lindgren, 1955; Abrahams et al., 1960;Lovick, 1985a; Carrive and Bandler, 1991;

Bandler et al., 1991), and movements of neck,back and limbs (Liebeskind et al., 1973;Fardin et al., 1984; Bandler and Carrive, 1988;Bandler et al., 1991).The question arises through which pathwaysthe PAG controls these motor output systems.The most simple explanation would be that itprojects directly to the motoneurons innervat-ing the muscles involved. Another possibil-ity is that the PAG projects to premotorinterneurons in the brainstem or in the spinalcord. It has been demonstrated that the PAGdoes not influence motoneurons directly, butindirectly via interneurons in caudalbrainstem and cervical cord. In respect to

22

Chapter 1

nociception control the PAG projects to therostral ventromedial medulla, specifically themidline nucleus raphe magnus and adjacentreticular formation, which maintains directconnections with laminae I and V of the spi-nal cord (Abols and Basbaum, 1981;Basbaum and Fields, 1984 ; Mason et al.,1985; Holstege, 1988a). In respect to vocali-zation the PAG uses the nucleusretroambiguus as a relay to motoneurons offor example the larynx, pharynx and abdomi-nal muscles (Holstege, 1989; Zhang, 1992).Considering the influence of the PAG on car-diovascular control it appears that neurons inthe rostral ventrolateral medulla play the roleof relay between the PAG and the sympatheticmotoneurons in the intermediolateral cell col-umn of the spinal cord (Lovick 1985b; Carriveet al., 1989; Lovick, 1991).Regarding the neck movements elicited bystimulation in the PAG, the exact pathwaysare not yet elucidated. Retrograde tracingstudies in the cat and monkey have shownthat the PAG projects to the cervical and up-per thoracic spinal cord (Castiglioni et al.,1978; Huerta and Harting, 1982; Mantyh,1983; Holstege, 1988b). Anterogradeautoradiographical tracing studies in the opos-sum and in the cat have shown that these PAG-spinal neurons terminate on interneurons inlaminae V, VII and the dorsal part of laminaVIII in the upper cervical cord and in lami-nae VII and VIII in the more caudal cervicaland upper thoracic cord (Martin et al., 1979;Holstege, 1988a). It is suggested that theseinterneurons are the premotor interneurons ofthe neck muscle motoneurons which are re-sponsible for neck movements after PAGstimulation (Holstege, 1988a). Direct projec-tions from the PAG to neck musclemotoneurons have never been demonstrated.Stimulation in the PAG can elicit movementsof the back and hind limbs also. Themotoneurons involved in these movementsare located in the cervical, thoracic, lumbarand sacral segments of the spinal cord. Thequestion arises whether the PAG projects to

these motoneurons and/or to their premotorinterneurons. Retrograde studies in opossumand monkey revealed a few labeled PAG neu-rons projecting to the caudal thoracic and tothe lumbar spinal cord (Castiglioni et al.,1978; Martin et al., 1979), but their termina-tion pattern is still unknown. In the cat only avery light direct PAG projection to the upperlumbar levels has been reported (Holstege,1988b), but projections to more caudal spi-nal segments have never been demonstrated.The present retrograde and anterograde trac-ing study in the cat tries to precisely deter-mine the location of the PAG-spinal neuronsand their distribution pattern.

Materials and methodsSurgical proceduresA total of 12 adult male cats was used and thesurgery procedures, pre- and postoperativecare, handling and housing of the animalsfollowed protocols approved by the Facultyof Medicine of the University of Groningen.For surgery animals were initiallyanesthetized with intramuscular ketamin(Nimatek, 0.1 ml/kg) and xylazine (Sedamun,0.1 ml/kg), after which they were keptanesthetized by ventilation with a mixture ofO2, N2O and halothane. During surgeryelectrocardiographic activity (ECG) and bodytemperature were monitored. Following asurvival time of 3 days the animals were ini-tially anesthetized with ketamin (0.1 ml/kg)and xylazine (0.1 ml/kg) i.m., followed by 6ml 6% pentobarbital sodium i.p. The cats wereperfused transcardially with two liters of 0.9%saline at 37o C, directly followed by two litersof 0.1M phosphate buffer, containing 4% su-crose, 1% paraformaldehyde and 2% glutar-aldehyde.

Retrograde tracing studyIn 9 anesthetized cats after laminectomy ap-proximately 80 µl 10% horseradish peroxi-dase (HRP) in saline was injected into vari-ous levels of the spinal cord using a Hamil-ton microsyringe. In some cases HRP was

23

PAG projections to lamina VIII and medial VII

Table 1. Schematic drawings of the location ofthe injection sites and hemisections at different spi-nal cord levels of the cat.

2199

2173

2214

2188

2192

2152

2221

2141

2139

case hemisection

C2 ---

C3 ---

C5T2

T4 T1

T12 T10

L6 L3

S1 L4

S2 L5

S3 L5

injection

injected unilaterally (left side), in others bi-laterally. Since HRP is transported from bothterminals and damaged axons, multiple nee-dle penetrations were made in the spinal grayand white matter. In all cases, except for theupper cervical ones, prior to the injection aright-sided hemisection was made some seg-ments rostral to the injection site. This wasdone in order to verify that retrogradelylabeled neurons in the mesencephalon distrib-uted their axons only through the funiculi ofthe left half of the spinal cord.After perfusion the brains and spinal cordswere removed, post-fixed for two hours andstored overnight in 20% sucrose in phosphatebuffer at 4o C. Subsequently the brainstemwas cut in 40 µm frozen sections of whichevery fourth section was incubated accord-ing to the tetramethylbenzidine method, de-hydrated and coverslipped. The spinal seg-ments with the injection sites were cut in 40µm sections and every fourth section wasprocessed with diamino-benzidine (DAB), tobe able to determine the exact area of injec-tion. The precise extent of the hemisectionswas also determined. Sections were studiedwith a Zeiss darkfield stereomicroscope anda Zeiss Axioskop light microscope. In eachcase labeled neurons were plotted in draw-ings of different levels of the mesencephalonwith the aid of a computer.

Anterograde tracing studyIn order to determine where the PAG-spinalfibers terminate, in 3 cats injections of 20 nl5% wheat germ agglutinin-conjugated horse-radish peroxidase (WGA-HRP) in saline wasinjected in the left PAG. These injections wereplaced stereotaxically in the ventrolateral andlateral PAG, which according to the retrogradestudy contained the PAG-spinal neurons.The WGA-HRP was injected through glassmicropipettes using a pneumatic picopump(World Precision Instruments PV830). ThePAG was approached dorsally in case 2239and dorsolaterally in cases 2248 and 2250.The latter approach was chosen to prevent

neurons in the superior colliculus being in-volved in the injection site. All surgical andhistological procedures were similar to theretrograde study and every fourth section ofthe caudal brainstem and of 10 to 20 segmentsthroughout the length of the spinal cord werestudied. The mesencephalon was cut in 40 µmsections and every fourth section was proc-essed with DAB to determine the exact areaof injection.

24

Chapter 1

Table 2. Numbers of HRP labeled neurons observed in the ipsilateral and contralateral PAG and adja-cent tegmentum.

with an injection in the cervical and upperthoracic spinal cord a few labeled neuronswere found at the border of the dorsal PAGas well. Relatively few labeled neurons werefound on the right (contralateral) side of thePAG, indicating that the PAG-spinal projec-tion is predominantly ipsilateral. After cervi-cal and thoracic HRP injections the labeledneurons in the ventrolateral PAG were locatedin a rostrocaudally oriented column extend-ing caudally from the level of the caudal poleof the decussation of the brachiumconjunctivum to rostrally the level of the cau-dal pole of the oculomotor nucleus. The high-est number of labeled neurons per section wasfound just rostral to the level of the trochlearnucleus. In the cases of the lumbar and sacralinjections the labeled neurons in the PAGwere found exclusively around the level ofthe trochlear nucleus.Per 40 µm section the number of labeled neu-rons in the left ventrolateral and lateral PAGand adjacent tegmentum varied between 20neurons, after an HRP injection in the uppercervical cord, and one single neuron after aninjection in the sacral cord. As mentioned inthe Materials and methods one out of four 40µm sections was incubated. In these sections

Spinal cord sections were studied with a ZeissAxioplan microscope using darkfield polar-ized illumination. The labeled fibers and neu-rons were plotted using a drawing tube, pro-jecting directly on a digitizer which was con-nected to a computer.

ResultsRetrograde tracing studyLocation of injection sites and hemisectionsTable 1 gives a schematic representation ofthe injections and hemisections. The injec-tions involved at least the entire left half ofthe spinal cord, except for cases 2199 and2173, in which part of the ventromedial anddorsomedial funiculi were not involved in theinjection site. Hemisections involved the rightside of the spinal cord and sometimes ex-tended into the left dorsal funiculus. Only incase 2152 did the hemisection extend into theleft ventromedial funiculus.

Labeled neurons in the PAGIn all 9 cases densely labeled neurons werefound ipsilaterally in the ventrolateral andlateral PAG and, except for the sacral cord-injected cases, in the laterally adjacent mes-encephalic tegmentum (Fig. 1). In the cases

Case Injection Number of labeled neurons site

Ipsilateral Contralateral

PAG Tegmentum Total PAG Tegmentum Total

2199 C2 320 136 456 93 15 108 2173 C3 181 73 254 24 5 29 2214 T2 126 51 177 13 0 13 2188 T4 62 11 73 7 0 7 2221 T12 37 7 44 1 0 1 2192 L6 22 1 23 0 0 0 2141 S1 5 0 5 0 0 0 2152 S2 2 0 2 0 0 0 2139 S3 1 0 1 0 0 0

25

PAG projections to lamina VIII and medial VII

SC

IC

IC

IC

III

Aq

PAG

SC

SC

PAG

IV

Aq

Injection T4

Hemisection T1

SC

ICAq

ICPAG

Aq

SC

PAG

SC

IC

IV

SC

Hemisection L3

Injection L6

SC

case 2188 case 2192

NRNR

Fig. 1. Schematic drawings of various levels of the cat mesencephalon with labeled periaqueductal(PAG) neurons after an HRP injection in the left spinal cord and a contralateral hemisection. Eachdrawing represents one 40 µm thick section. (Aq: aqueduct of Silvii, IC: inferior colliculus, PAG:periaqueductal gray, NR: nucleus ruber, SC: superior colliculus, III: oculomotor nucleus, IV: trochlearnucleus)

26

Chapter 1

the number of labeled neurons in the PAG andthe laterally adjacent tegmentum was counted(Table II). Ipsilaterally, the total number oflabeled neurons varied from over 500 neu-rons, after an injection in C2, to one neuron,after an injection in S3. In the cervical andupper thoracic cases about one third of theselabeled neurons were located in the laterallyadjacent tegmentum. In the lower thoraciccases the fraction of labeled neurons in thetegmentum was much smaller. In the lum-bosacral cases almost no labeled neurons werepresent in the adjacent tegmental field.Contralaterally, the total number of labeledneurons was relatively small. Only in the C2injected case were more than 100 labeled neu-rons observed, of which almost 90% werelocated within the borders of the PAG. In noneof the cases with an injection caudal to C3labeled neurons were observed in the contral-ateral adjacent tegmentum. In the lumbosac-ral cases no labeled neurons were found inthe contralateral PAG or in the adjacent teg-mentum.The diameter of the ventrolateral PAG-spi-nal neurons was relatively large and differedfrom 15 to 40 µm, with a mean of approxi-mately 33 µm (Fig. 2).In the cases with an injection in the cervicaland upper thoracic spinal cord some faintlylabeled neurons were observed at the borderof the dorsal PAG and in the dorsally adjoin-

ing superior colliculus. These labeled dorsalneurons were relatively small with a diam-eter of approximately 15 µm, and were lo-cated more caudally than the labeled neuronsin the ventrolateral and lateral PAG.

Anterograde tracing studyInjection sitesIn 3 cases WGA-HRP injections were placedin the ventrolateral and lateral PAG and adja-cent tegmentum (Fig. 3). In case 2239 theinjection site involved the ventrolateral andlateral PAG at the level between the trochlearnucleus and the oculomotor nucleus. In case2248 the injection site was located morerostrally and involved the ventrolateral partof the PAG and the ventrolaterally adjacenttegmentum. In the last case (2250) the injec-tion site was located slightly more caudallythan in case 2239. It included the lateral partof the ventrolateral PAG, the laterally adjoin-ing tegmentum and part of the inferiorcolliculus.

Descending pathwayFrom the injection site many labeled fiberspassed ventrally to descend through the lat-eral tegmental field of caudal mesencephalonand rostral pons. They gradually shifted ven-tromedially, and at upper medullary levelsthey were found just medial to the superiorolivary complex and facial nucleus and more

Fig. 2. Brightfield photomicrographs of one labeled PAG-spinal neuron after an HRP injection in L6(case 2192). A: Bar represents 300 µm; B: Bar represents 50 µm.

27

PAG projections to lamina VIII and medial VII

caudally just lateral to the inferior olive. Manyfibers terminated in the ventromedial tegmen-tal field and nucleus raphe magnus. Somefibers continued caudally and came to lie inthe ventral funiculus of the caudal medulla.Further caudally half of these labeled fiberscontinued into the ventral and lateral funiculiof the upper cervical cord, the other halfgradually shifted medially to descend in theventromedial funiculus (Fig. 4). At lower cer-vical levels descending labeled fibers werepresent in the lateral, ventrolateral and ven-tromedial funiculi. At thoracic and lumbarlevels the number of descending fibers dimin-ished gradually. Most of these fibers werefound in the ventromedial funiculus and onlya very few in the lateral funiculus. At sacrallevels no more than two labeled descendingfibers per section were observed.In contrast to other spinal pathways, such asthe coeruleo-, raphe-, reticulo- and vestibu-lospinal pathways (Holstege, 1988a;Holstege, 1988b), the PAG-spinal fibers did

not descend in the most peripheral portion ofthe white matter of the spinal cord.

Distribution pattern in the spinal graymatterAt the upper cervical level many labeled fibersterminated ipsilaterally in the lateral portionsof laminae V, VII and VIII of Rexed (1954;Fig. 5). Contralaterally, some fibers termi-nated in these same laminae. Caudal to C1the termination of labeled fibers shifted tomore central and medial parts of the ventralhorn (medial part of lamina VII and laminaVIII). Ipsilaterally, at levels caudal to T2 de-scending labeled fibers were distributedmainly to medial lamina VII with some fibersterminating in the dorsal part of lamina VIII.Contralaterally no labeled fibers were foundbeyond the level of T2. No labeled fibers werefound to terminate in lamina IX. The labeledfibers also did not terminate in the intermedi-olateral cell column, except for some thinfibers at upper thoracic levels. Such fibers

2248

SC

NR

PAG

2239

III

SC

22482239

22502248

2250

2239

IV

2248

IC 2239PAG

2250

2250

ICAq

Fig. 3. Schematic drawings of the WGA-HRP injection sites in the ventrolateral PAG and the adjacenttegmentum in cases 2239, 2248 and 2250. Abbreviations: Aq: aqueduct of Silvii, IC: inferior colliculus,PAG: periaqueductal gray, NR: nucleus ruber, SC: superior colliculus, III: oculomotor nucleus, IV:trochlear nucleus.

28

Chapter 1

case 2239

C5

C8

T2

T8L1

L6

L7

S1S2

C2

C3

C1

Fig. 4. Schematic drawing of the labeled fibers at various levels of the spinal cord after a WGA-HRPinjection in the ventrolateral PAG of the cat (case 2239). It must be emphasized that each drawing of aspinal cord segment represents six 40 µm thick sections.

29

PAG projections to lamina VIII and medial VII

Fig. 5. Darkfield polarized photomicrographs of various levels of the spinal cord of the cat after aWGA-HRP injection in the ventrolateral PAG (case 2239). Bar represents 500 µm.

C1

T2

L5

L7 S2

L6

T8

C5

30

Chapter 1

Fig. 6. Schematic illustration of the projectionsfrom interneurons in the intermediate zone (laminaV - VIII) via the propriospinal path to the moto-neurons (from Holstege, 1991).

have been reported by Holstege (1988a, b)also, using autoradiographic tracing tech-niques.

DiscussionThe present study demonstrates that neuronsin the PAG project throughout the length ofthe spinal cord. In earlier anterograde tracingstudies in the cat PAG projections to cervicaland upper thoracic segments were found andonly a sparse distribution to lower thoracicand upper lumbar segments was reported(Holstege, 1988a). PAG projections to thecaudal lumbar and sacral spinal cord havenever been described before. The numbers ofretrogradely labeled neurons, as presented intable 2, should be interpreted with caution.PAG neurons projecting to, for example, thelumbosacral cord may have been labeled af-ter an HRP injection in the thoracic or cervi-cal cord, since plain HRP is taken up not onlyby terminating fibers but also by damagedfibers of passage (Kristensson and Olsson,1974). This implies that the number of PAGneurons labeled after an HRP injection in acertain segment can be larger than the numberof PAG neurons whose fibers actually termi-nate there. Irrespective of this, the resultsclearly show that there are far more PAG neu-rons projecting to the cervical cord than tothe thoracic or lumbar cord and that only avery few PAG neurons project as far as thesacral cord.From several brainstem-spinal pathways, forinstance the interstitiospinal pathway(Fukushima et al., 1978), the vestibulospinalpathway (Abzug et al., 1974), the rubrospi-nal pathway (Shinoda et al., 1977), and thepontine reticulospinal pathway (Matsuyamaet al., 1993), it has been shown that onebrainstem neuron sends collateral fibers tomore than one segment of the spinal cord. Thisraises the question how collateralized thePAG-spinal system is. In other words, doPAG-spinal fibers terminating in the caudalspinal cord, for instance at the lumbar level,have collaterals to more rostral segments, for

instance the thoracic and cervical levels, ordo all PAG-spinal fibers project to one or afew adjacent segments only. Another possi-bility is that part of the PAG-spinal fibers arehighly collateralized and part of them have aspecific projection. Whichever, it is clear thatthere exist PAG neurons projecting exclu-sively to the cervical cord. Whether there ex-ist PAG neurons projecting exclusively to thethoracic, lumbar or sacral cord, respectively,cannot be determined. Double-labeling trac-ing experiments might solve this problem.

Function of the PAG-spinal pathwayOur study shows that the PAG-spinal fibersdo not terminate directly on motoneurons, buton interneurons in the ventral horn. Earlierfindings (Rustioni et al., 1971; Sterling andKuypers, 1968; Molenaar et al., 1974 ;Molenaar, 1978 ) have demonstrated that thedorsolateral part of the intermediate zone (lat-eral part of lamina V to VII) of the brachialand lumbosacral cord contains interneuronsprojecting via propriospinal pathways to the

31

PAG projections to lamina VIII and medial VII

Fig. 7. Schematic drawing of the spinal pathways from reticular formation adjacent to the interstitialnucleus of Cajal (INC-RF), the reticular formation adjoining the rostral interstitial nucleus of the mediallongitudinal fasciculus (riMLF-RF), the lateral vestibular nucleus (LVN), the medial pontine tegmen-tum and the rostral medulla (PMTm) and the periaqueductal gray (PAG). At each level of the spinal cordboth the areas of descending fibers in the white matter and the area of terminating fibers in the graymatter are indicated. Note that this scheme does not gives an indication about the number of fibersbelonging to the different pathways. (Spinal pathways from INC-RF, riMLF-RF, LVN and PMTm areearlier described by Holstege, 1988b)

LVN

riMLF-RF

PMTm

PAG

C3 C8 T7 L7

INC-RF

SPINAL PROJECTION FROM:

32

Chapter 1

motoneurons of distal limb muscles, whileinterneurons in the medial intermediate zoneproject bilaterally to axial musclemotoneurons. Interneurons located in be-tween these two areas project to proximallimb muscle motoneurons (Fig. 6). Thepresent results show that the PAG-spinalfibers projecting to the caudal cervical, tho-racic and lumbosacral cord terminate in themedial portion of the intermediate zone(lamina VIII and medial part of lamina VII),which would imply that the PAG is involvedin the control of axial and proximal muscula-ture.Projections similar to the PAG-spinal path-way are derived from other brainstem areas,such as the reticular formation adjacent to theinterstitial nucleus of Cajal (INC-RF; Nyberg-Hansen, 1966; Holstege and Cowie, 1989),the reticular formation adjacent to the rostralinterstitial nucleus of the medial longitudinalfasciculus (riMLF-RF; Holstege, 1988b;Holstege and Cowie, 1989), the lateral ves-tibular nucleus (LVN; Nyberg-Hansen andMascitti, 1964; Petras, 1967; Holstege andKuypers, 1982) and the medial part of thecaudal pontine and upper medullary tegmen-tum (PMTm; Nyberg-Hansen, 1965; Petras,1967; Holstege and Kuypers 1982; Holstege1988b). All these four brainstem nuclei areinvolved in the control of axial and proximalmuscles, by way of terminating in the medialpart of the intermediate zone (Fig. 7). Con-sidering the function of these different de-scending systems, physiological and lesionstudies make clear that they all have a spe-cialized function within the vestibular-ocu-lomotor framework. The INC-RF is thoughtto be responsible for the slow, small move-ments of neck and back muscles controllingthe position of the head necessary to integratevertical gaze shifts and head movements(Hyde and Toczek, 1962, Fukushima et al,1985; Fukushima, 1987). The riMLF-RFmight control the fast axial movements dur-ing rapid vertical eye movements. The LVNis responsible for movements of neck, back

and hind limbs necessary for maintaining thebody equilibrium, the position of the head andthe direction of gaze (Wilson and Peterson,1981). Finally, the PMTm takes part in thecontrol of the postural musculature connectedwith fast horizontal eye movements (Hassler,1972; Büttner-Ennever and Büttner, 1988).Although the PAG-spinal pathway seems tobe involved in axial musculature control, it isnot clear in what framework this system func-tions. The ventrolateral and lateral PAG, inwhich the PAG-spinal neurons are located,have never been shown to receive afferentsfrom vestibular or oculomotor structures likeeach of the other four brainstem areas(Büttner-Ennever and Büttner, 1988). Theventrolateral and lateral PAG, however, re-ceives many afferents from limbic structuressuch as the lateral hypothalamic area (Berkand Finkelstein, 1982; Holstege, 1987), thecentral nucleus of the amygdala (Hopkins andHolstege, 1978; Price and Amaral, 1981;Rizvi et al., 1991) and the bed nucleus of thestria terminalis (Holstege et al., 1985). Theselimbic structures send fibers to neither theriMLF-RF, the INC-RF, the LVN nor to thePMTm, which suggests that the PAG-spinalpathway does not function in a vestibulo-ocu-lomotor framework.A more obvious possibility would be that thePAG-spinal pathway controls the axial mus-cles within the framework of the emotionalmotor system, as defined by Holstege (1992).The emotional motor system contains thedescending pathways from limbic structuresto caudal brainstem and spinal cord and canbe divided functionally in a medial and a lat-eral part. The medial part, containing for in-stance the diffuse coeruleo- and raphe-spinalpathways, has a global effect on the level ofactivity of the somatosensory- and motoneu-rons in general. The lateral part of the emotio-nal motor system contains some specific partsof the brain, such as the central nucleus ofthe amygdala, the bed nucleus of the stria ter-minalis, the lateral hypothalamus, and thosecells of the PAG, that are thought to be in-

33

PAG projections to lamina VIII and medial VII

Fig. 8. Schematic overview of behavioral patterns after stimulation at different rostro-caudal levels ofthe PAG in the freely moving cat, related to the location of the PAG-spinal neurons. Darkest areasrepresent the levels with the strongest behavioral reactions (stimulation experiments), with the highestnumber of labeled neurons per section (retrograde study), and with the centres of the WGA-HRP injec-tions (anterograde study). Abbreviations: lat: lateral, ventrolat: ventrolateral, III: oculomotor nucleus,IV: trochlear nucleus, i.a.p.: interaural plane according to Berman, 1968.

AP coordinates Bandler et al., 1991 strong immobility moderate immobility strong flight moderate flight strong threat moderate threat Mouton and Holstege, 1994 retrograde tracing study: labeled PAG-spinal neurons anterograde tracing study: injection site 2250 injection site 2248 injection site 2239

2.0

1.5

1.0

0.5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

posterior anteriori.a.p

.

IV III

subtentorial pretentorial

ventrolateralventrolateral

lateral

lat./dorsolateral

lat./ventrolateral

lateral PAG

periaqueductal gray

ventrolateral

ventrolateral

lat/ventrolat.

lateral/ventrolateral

volved in specific functions, such as vocali-zation and blood pressure control during emo-tional behavior. If one considers the PAG-spi-nal pathway as part of the lateral componentof the emotional motor system, it might con-trol axial movements in specific emotional ac-tivities.Experiments in the freely moving cat (Bandlerand Carrive, 1988; Zhang et al., 1990; Bandleret al., 1991) have shown that both electrical

stimulation and stimulation with excitatoryamino acids in the PAG can elicit three basicpatterns of behavior, threat display, flight orimmobility. In this respect threat display con-sists of moderate pupil dilation and pilo-erec-tion, vocalization (howling usually mixedwith hissing, or hissing alone), retraction ofthe ears and/or arching of the back. Flight ischaracterized by moderate pupil dilation andpiloerection, vocalization (mewing), rapid

34

Chapter 1

running and multiple jumps. Immobility is thesituation in which the cat shows a period ofprofound inactivity. Each of these responsescan be elicited in specific parts of the PAG(Fig. 8). Zhang et al. (1990) showed thatstrong or moderate immobility is found afterstimulating the ventrolateral part of the cau-dal one third of the PAG (subtentorial PAG),while a strong flight response is observed af-ter stimulating more dorsally, i.e., in the lat-eral subtentorial PAG around the P0.5 levelof Berman (1968). The area in which stimu-lation produced moderate flight responsesoccupied a slightly larger portion of the sub-tentorial PAG. Threat display was found morerostrally in the PAG (pretentorial or middlethird of the PAG). Strong threat display canbe elicited by injecting excitatory amino acidin the lateral and ventrolateral parts of thecaudal pretentorial PAG. Moderate threat dis-play was evoked from more rostral parts ofthe lateral pretentorial PAG (Bandler andCarrive, 1988).As described in the results section, the PAG-spinal neurons are located in the lateral andventrolateral PAG. They form a rostrocaudal-ly oriented column extending caudally from

the level of the decussation of the brachiumconjunctivum to rostrally around the level ofthe caudal pole of the oculomotor nucleus.The bulk of labeled neurons was found justrostral to the level of the trochlear nucleus,i.e., in the caudal pretentorial PAG.Comparing the location of the PAG-spinalneurons with the results of the stimulationexperiments of Bandler and co-workers (Fig.8) leads to the supposition that the PAG-spi-nal neurons play a role in threat display. Pos-sibly, arching of the back as a component ofthreat display is the result of the activity ofPAG-spinal neurons controlling axial muscu-lature.In conclusion, the present results show thatthe PAG projects to the lateral parts of theintermediate zone of the upper cervical cordand to the medial parts of the intermediatezone of the caudal cervical, thoracic, lumbarand sacral spinal cord. In these areas thepremotor interneurons of the axial musclesare located. It is hypothesized that the PAG-spinal pathway forms the anatomical frame-work for arching of the back as part of threatdisplay.