Efferent projections of the periaqueductal gray in the rabbit

NeuroscienceVol. 40, No. 1, pp. 191-216,1991 Printed in Great Britain

0306-4522/91$3.00+ 0.00 Pergamon Press plc 0 1991IBRO

EFFERENT PROJECTIONS OF THE PERIAQUEDUCTAL GRAY IN THE RABBIT S. T. MELLER* and B. J. DENNtst Department of Physiology, The University of Adelaide, SA 5000, Australia Abstract-The efferent projections of the periaqueductal gray in the rabbit have been described by anterograde tract-tracing techniques following deposits of tritiated leucine, or horseradish peroxidase, into circumscribed sites within dorsal, lateral or ventral periaqueductal gray. No attempts were made to place labels in the fourth, extremely narrow (medial), region immediately surrounding the aqueduct whose size and disposition did not lend itself to confined placements of label within it. These anatomically distinct regions, defined in Nissl-stained sections, corresponded to the same regions into which deposits of horseradish peroxidase were made in order for us to describe afferent projections to the periaqueductal gray. In this present study distinct ascending and descending fibre projections were found throughout the brain. Terminal labelling was detected in more than 80 sites, depending somewhat upon which of the three regions of the periaqueductal gray received the deposit. Therefore, differential projections with respect to both afferent and efferent connections of these three regions of the periaqueductal gray have now been established. Ventral deposits disclosed a more impressive system of ramifying, efferent fibres than did dorsal or lateral placements of labels. With ventral deposits, ascending fibres were found to follow two major pathways from periaqueductal gray. The periventricular bundle bifurcates at the level of the posterior commissure to form hypothalamic and thalamic components which distribute to the anterior pretectal region, lateral habenulae, and nuclei of the posterior commissure, the majority of the intralaminar and midline thalamic nuclei, and to almost all of the hypothalamus. The other major ascending pathway from the periaqueductal gray takes a ventrolateral course from the deposit site through the reticular formation or, alternatively, through the deep and middle layers of the superior colliculus, to accumulate just medial to the medial geniculate body. This contingent of fibres travels more rostrally above the cerebral peduncle, distributing terminals to the substantia nigra, ventral tegmental area and parabigeminal nucleus before fanning out and turning rostrally to contribute terminals to ventral thalamus, subthalamus and zona incerta, then continuing on to supply amygdala, substantia innominata, lateral preoptic nucleus, the diagonal band of Broca and the lateral septal nucleus. Caudally directed fibres were also observed to follow two major routes. They either leave the periaqueductal gray dorsally and pass through the gray matter in the floor of the fourth ventricle towards the abducens nucleus and ventral medulla, or are directed ventrally after passing through either the inferior colliculus or parabrachial nucleus. These ventrally directed fibres merge just dorsal to the pons on the ventral surface of the brain. Terminal labelling was found over a large number of midbrain and hindbrain structures including midbrain and pontine reticular formation, locus coeruleus, cuneiform nucleus and caudal raphe nuclei. An extension of this ventrally directed bundle of fibres continues more caudally passing in a sheet over the pyramids, trapezoid body and inferior olive. From here a diffuse system of fibres traverses the meclullary reticular formation to form a compact bundle that distributes to nucleus ambiguus and to the ventral reticular formation of the medulla. Less impressive and less extensive labelling was found with deposits in dorsal and lateral periaqueductal gray. The large number of sites receiving projections from the periaqueductal gray, the diversity of routes by which these projections course through the brain, and a degree of differentiation of projections with respect to site of origin is indicative of the degree of morphological complexity of this area. This is to be expected, for functional findings have indicated that the periaqueductal gray is implicated in the modification, integration and expression of motor, sensory, autonomic, limbic and endocrine systems. Certain functional implications are discussed in the light of the projections disclosed.

In recent times, reference to the functions of periaqueductal gray (PAG) has predominantly heen in regard to a proposed role for it in stimulation-produced or opiate-induced analgesia, or in emotional states, yet its involvement, additionally, in a wide range of other functions has also been proposed (for *Current address: College of Medicine, Department of Pharmacology, The University of Iowa, Iowa City, IA 52242, U.S.A. tTo whom correspondence should be addressed. Abbrezktiom: DLF, dorsal longitudinal fasciculus of Schiitz; HRP, horseradish peroxidase; PAG, periaqueductal gray; SPA, stimulation-produced analgesia.

reviews see Refs 41,98). Despite its apparent importance in these various functions, there is still little appreciation of the overall role that the PAG may play in integrating such expressions, and little understanding of the neural basis by which each may he mediated. In order to gain greater insight into the functions of this (or any other) region, it remains necessary to establish, conclusively, both the basic structure and the particulars of its alferent and efferent connections with other regions of the central nervous system. A number of descriptions of efferent connections of the PAG have heen established in lesion28~37~6L~62~79~““‘~‘W 191

192

S. T. MELLER

and B. J. DENNIS

Abbreviations used in figures aa acb ad aha ah1 am amb arc apt Ec bnac bnst BP CA M cc cd cl cm cnf

:bb dh dhy dig dll dmh dr dtn ep ff FX G hl, h2 hndb iam ic IC III imd io la Id IC ICZ

Ih Iha IP Im ma md LZF mm mrf Z? P PAG pah pb pbl pbm pbr PC PC PED

anterior amygdaloid nucleus nucleus accumbens anterior dorsal nucleus of thalamus anterior hypothalamic area lateral hypothalamic area anterior medial nucleus of thalamus nucleus ambiguus arcuate area anterior pretectal area anterior ventral nucleus of thalamus nucleus of brachium of inferior colliculus bed nucleus of anterior commissure bed nucleus of stria terminalis brachium pontis anterior commissure central amygdaloid nucleus corpus callosum caudate nucleus centralis lateralis nucleus of thalamus centre median nucleus of thalamus cuneiform nucleus central superior nucleus diagonal band of Broca dorsal horn dorsal hypothalamic area dorsal lateral geniculate nucleus dorsal nucleus of lateral lemniscus dorsomedial hypothalamic nucleus dorsal raphe nucleus dorsal tegmental nucleus entopeduncular nucleus fields of Fore1 fornix genu of the facial nerve fields of Fore1 horizontal nucleus of diagonal band interanteromedial nucleus of thalamus inferior colliculus internal capsule occulomotor nucleus interomediodorsal nucleus of thalamus inferior olive lateral amygdaloid nucleus lateral dorsal nucleus of thalamus locus coeruleus locus coeruleus pars alpha lateral habenula nucleus lateral hypothalamic area lateral posterior nucleus of thakmus lateral reticular nucleus medial amygdaloid nucleus medial dorsal nucleus of thalamus medial geniculate nucleus medial longitudinal fasciculus mammillary body midbrain reticular formation nucleus of Posterior commissure optic tract pons periaqueductal gray paraventricular hypothalamic nucleus parabigeminal nucleus lateral parabrachial nucleus medial parabrachial nucleus parabrachial nucleus paracentral thalamic nucleus posterior commissure cerebral peduncle

Pf Pgl ph phg pmn pnc pnca pno PO PO1 pom poma Pope PP Prm PrV $t pva P”P rd re ret rgc rh ml rn ro rPa rv sc sch sg sgn si Sl

smm snc snl snr so sol spvc spVi spvo sth tbc tgr TZ V vh VI VII Vlln VbC VI vlg VII VP1 vpm vm vmh vndb vrf vta X X11 7.i

nucleus parafascicularis nucleus paragigantocellularis lateralis posterior hypothalamus nucleus prepositus hypoglossi paramedian reticular formation nucleus pontis caudalis nucleus pontis caudalis pars a nucleus pontis oralis posterior thalamic nucleus lateral preoptic nucleus medial preoptic nucleus magnocellular preoptic nucleus periventricular preoptic nucleus peripeduncular nucleus dorsal premammillary nucleus principal trigeminal nucleus parataenialis nucleus pyramids anterior paraventticular nucleus posterior paraventricular nucleus nucleus reticularis dorsalis of medulla nucleus reuniens reticular nucleus of thalamus nucleus reticularis gigantocellulatis rhomboid nucleus nucleus raphe magnus red nucleus nucleus raphe obscurus nucleus rdphe pallidus nucleus reticularis ventralis of medulla superior colliculus suprachiasmatic nucleus suprageniculate nucleus nucleus supragenualis substantia innominata lateral septal nucleus supramammillary nucleus substantia nigra pars compacta substantia nigra pars lateralis substantia nigra pars reticularis supraoptic nucleus lateral superior olivary nucleus spinal trigeminal nucleus pars caudalis spinal trigeminal nucleus pars intermediatre spinal trigeminal nucleus Pars orahs subthalamic nucleus area of the tuber cinerium tegmental reticular nucleus trapezoid body spinal trigeminal nerve ventral horn abducens nucleus facial nucleus facial nerve ventrobasal nucleus of thalamus ventrolateral nucleus of thakmus ventral lateral geniculate nucleus ventral nucleus of lateral kmnkcus ventral posteriolateral m&us of thalamus ventral posteriomedial nuckus of thakmu.9 ventromedial nucleus of tha1amu.s ventromedial hypothalamic nuckus vertical nucleus of the diagonal band ventral reticular formation ventral tegmental area vagus nucleus hypoglossal nucleus zona incerta

Efferents to periaqueductal gray in the rabbit

193

molars and orbital bones.42 Since. the suggested modification tract-tracing studies.47,89,W,96,97,L17 of the rabbit head-holder resulted in instability, larger The majority of these reports have been limited to rabbits were avoided. providing details of connections with only restricted Either 30-9OpCi of [‘Hlleucine (specific activity regions of the brain, thus only partially describing 48 Ci/mmol), (32 rabbits) or lo-100 nl of 30% HRP (Sigma type VI or Boehringer grade I) (35 rabbits) was administered systems within the total complement of the ascending by slow pressure injection from a Drummond microtrol and descending projections that we have come to microsyringe (tip diameter 2064Opm), using a horizontal recognize. Therefore, the principal aim of this study approach to PAG9’ in order to minimize the number of has been to determine the entire complement of PAG brain structures to be penetrated. With this approach label efferent projections in the rabbit, by using both the could reasonably be placed along almost the entire length of the PAG. Localized deposits were placed within dorsal, autoradiographic method and the anterograde translateral and ventral regions of the PAG as shown in Figs 1 port of horseradish peroxidase (HRP). and 2. Deposits were made over a period of 20-40 min with There is still debate as to whether the PAG can the micropipette left in place for a further 30-45 min before justifiably be divided into cytologically distinct removal, to reduce spread of the label. regions having functional significance,22~3@‘~W*“o~“7~ Autoradiography ‘34~14’~‘42 or whether it should rather be regarded as a After three to seven days, the animals were deeply heteromorphic structure.89*90*‘4’*142 Recent attempts anaesthetized by an intravenous injection of Nembutal have been made to detail projections with respect (Ceva, Australia). Following injection of 2500 I.U. of to specific zones within the PAG,22*47*96-98 or with heparin, the descending aorta was clamped and transcardiac perfusion was carried out with a prefixation rinse of 800 ml respect to specific functional systems within particuof 0.9% saline in phosphate buffer 0.067 M, pH 7.4, lar zones.‘“-L2,32 followed by 1 1 of 10% phosphate-buffered formalin, at Preliminary studies have indicated that the rabbit room temperature and subsequently by 5OOml of 10% PAG can reasonably be divided into cytologically sucrose in phosphate buffer. The brain, and spinal cord to the level of the third cervical segment, was removed and different distinct regionsW with topographically efferent96.97and afferent9* projections. Therefore, as stored in 30% sucrose in phosphate buffer, at 4°C for one to three days. part of an overall study on the neuroanatomy of Brains were sectioned coronally on a freezing microtome the PAG in the rabbit,9699 the second aim of this at 50 pm and serial sections were stored in distilled water. Immediately after sectioning, each third and fourth section study was to characterize its differential efferent was washed in two changes of distilled water and mounted projections. or by anterograde

EXPERIMENTAL PROCEDURES A total of 67 lop-eared rabbits (2-3 kg) were used in this study. After induction and maintenance of anaesthesia by Fluothane (ICI, Australia), the animal’s head was supported by a clamping device, attached to a stereotaxic frame which holds the head stable by counter pressure between the upper

on gelatinized or egg albumin-coated slides before being air-dried in a dust-free environment. Next, sections were dehydrated in a series of graded alcohols, defatted in xylene and then rehydrated. Slides were dipped in Ilford K2 nuclear emulsion (diluted 1:1 with distilled water) at 41”C, (under an Ilford S910 safelight). Emulsion-coated slides were placed on a cold plate at 5°C to set for 10min and then allowed to air-dry for a further 2 h in a dust-free area. Slides, stored in the dark in open slide boxes containing silica gel,

Lateral

Fig. 1. A series of drawings of coronal sections of the midbrain indicating the degree of spread of the label in the coronal plane in each experiment. Subsequent figures use the experiments depicted with a star as representrative examples of labelling resulting from deposits in ventral (Fig. 4), lateral (Fig. 7) or dorsal (Fig. 8) PAG. Deposits which spread beyond the defined subdivisions of PAG, and beyond the boundaries of the complex, are represented in the “oversized” series. The control series covers territory that could have been the source of spurious labelling had label leaked from the penetrating pipette. Within the box is a representative outline of PAG, at an intercollicular level, indicating the four subdivisions that we have recognized on cytological grounds.99

194

S. T. MELLEK and B. J. DENNIS

Fig. 2. Photomicrographs of a deposit of tritiated leucine, in ventral PAG, in the particular experiment for which a comprehensive description of efferent projections is given in the text. (A) The extent of spread of label (B,C,D) Photomicrographs of the representative regions signified in A. The deposit sites show a decrease in the density of silver grains from the centre to the periphery of the injection site. Labelling in D would be due partly to background labeiling (20-30 grains per 1000 pm’); partly to spread of label; and partly to labelling of connections intrinsic to the PAG. Scale bars = 100 pm. were placed in a desiccator overnight. Subsequently the slide boxes, with fresh silica gel bags enclosed, were sealed inside black piastic envelopes, and stored at 4°C for six to 12 weeks. The slides were then developed in Kodak D-19 (I 5%) for 2-3 min under an S910 Ilford safelight, fixed for 8 min in 30% sodium thiosulphate and washed in running water for up to 6Omin, rinsed several times in distilled water, dehydrated and coverslipped. Adjacent sections were stained lightly with neutral red to aid delineating nuclear boundaries. sections were viewed with an Olympus Vanox microscope using both bright- and dark-field illumination to

determine the distribution of fine silver grains throughout the brain. Nuclear boundaries were determined according to an atlas prepared in this laboratory, and in reference to the atlases of Winkler and Potter““~‘” and Meessen and 01szweski.93 Horseradish petoxidase histochemistry

Details of the method employed have been described in a previous paper in which afferent projections to the PAG were established.% Those same experimental cases have been used here to describe. the course of efferent fibres from the

Efferents to periaqueductal gray in the rabbit PAG, as assessed by the distribution of anterogradely transported HRP using tetramethyl benzidene as the chromogen. ItlmJLTs

Injection sites

The coronal outlines in Fig. 1 (and those in Fig. 2 in a previous paper describing afferent connections9*), represent the sites of greatest spread of the tracers in the coronal plane in each experiment. The patterns of labelling (with deposits in ventral, lateral or dorsal PAG) were consistent, irrespective of whether assessment was made on autoradiographic or anterogradely labelled HRP material. Tracers deposited at more rostra1 sites along the rostral-caudal axis of PAG tended to show more marked labelling of rostra1 structures whereas more caudally placed deposits resulted in more marked labelling of caudal structures. However, irrespective of the level of placement, the same structures were consistently labelled each time. Since a major objective of these experiments was to discover any evidence of differential projections between three cytologically distinct regions of PAG, experiments in which there was a significant degree of spread between two of these zones were not used. Any deposit within ventral PAG was found to effectively label the total complement of structures. Tracing techniques Autoradiography. The size of a deposit site is dependent on a number of variable factors. We attempted to minimize these by using pipettes with small diameter tips (20-40 pm), injecting relatively small volumes (50-500 nl) of high concentrations of radioactive leucine (30-9OpCi), and injecting at a slow rate (about 20 nl/min). The extent of each deposit site was assessed as the maximal area where the relative density of silver grains over cell bodies exceeded that over the surrounding neuropil, as indicated in Figs 1 and 2. It is not possible to determine the exact zone of uptake for there is no way of knowing just how heavily labelled cells at the periphery of the deposit site should appear to justify the conclusion that they would have been involved in the effective uptake and transport of detectable label. Using the above criterion, 17 of the 32 tritiated leucine injection sites were found to be restricted entirely to the PAG (Fig. 1). Of these, 11 were restricted entirely to one or other of three of the four subdivisions of the PAG (four to ventral, five to lateral and two to the dorsal subdivision). No deposits were confined to the narrow (medial) zone surrounding the aqueduct. Anterograde

transport

of horseradish

peroxidase.

The rationale for assessing the degree of spread of HRP from the deposit site has been discussed in a previous paper where the reasons for confidence in supposing that HRP had been localized within

195

discrete regions of PAG were detailed.98 In that report, regions sending projections to, or through, the PAG were assessed by labelling cells in the relevant areas by retrograde transport of HRP. Here, the same material has been used to establish the extent of anterograde transport of HRP in an effort to determine, by a second approach, just which structures received efferent projections from PAG. Control experiments

In processing the autoradiographic slides, appropriate controls were conducted to verify that neither positive nor negative chemographic effects had influenced the patterns of labelling described for each deposit site. Four of the control experiments were aimed at discovering what neural structures might label if the marker leaked into the aqueduct. Thirteen control experiments established what would have labelled if the deposit had extended beyond PAG; five experiments with leucine (Fig. l), and eight with HRP. Deposits ranged from 30 @i in 50 nl to 90 pCi in SOOn of tritiated leucine, and lo-100 nl 30% HRP. Deposits within the aqueduct resulted in no observable labelling of neural structures apart from a light halo of silver grains present in the ependymal lining of the aqueduct and third and fourth ventricles when autoradiographs were assessed using dark-field illumination. No concentration of silver grains, above background level was detected over the ependymal lining, or over neurons, using bright-field illumiResults from those experiments which nation. demonstrated a greater degree of halo development around the aqueduct than was seen with the ventricular control experiments were eliminated from consideration. At least one experiment with deposits in each of the specified regions demonstrated no observable halo in the ependymal lining of the ventricular system. Other control experiments were directed at establishing areas having efferent connections with cerebellar cortex, inferior colliculus, superior colliculus and the midbrain reticular formation. In each case, the pattern and strength of labelling in these control experiments differed appreciably with regard to those connections established with deposits in PAG. In contrast to the control experiments, the major areas that were established as receiving projections from PAG were: the bed nucleus of the stria terminalis, medial amygdaloid nucleus, lateral preoptic nucleus, lateral, dorsal and posterior hypothalamus, the center median nucleus and paraventricular nucleus of the thalamus, cuneiform nucleus, locus coeruleus, lateral parabrachial nucleus, midbrain pontine and medullary reticular formation and raphe magnus, and as well, fibres and terminals were found overlying the pyramids, trapezoid body and inferior olive. The chief difference in the pattern of labelling observed with PAG and inferior colliculus deposits

196

S. ‘I’. MLI.L~K

was that the latter site produced significant levels of terminal labelling within the medial geniculate body (which did not label with PAG deposits). In addition, midline thalamic nuclei and the bed nucleus of the stria terminalis did not label with deposits in the inferior colliculus, yet labelled strongly with PAG deposits. Major efferent connections of the superior colliculus were with the suprageniculate nucleus and pontine and medullary reticular fields. Although pontine and medullary reticular formation also receive major PAG projections their connections are predominantly with the ventral PAG region; not with dorsal or lateral PAG (which would be the regions to which spread of the label would most likely occur with deposits in superior colliculus). Further, deposits confined to the superior colliculus produced very little thalamic or hypothalamic labelling whereas those particular regions were found to receive major projections from PAG. The major projections of the midbrain reticular formation were with superior colliculus, PAG, paraventricular and parafascicular nuclei of the thalamus, facial nucleus, lateral reticular nucleus and other reticular areas in the hindbrain. Although this pattern is somewhat similar to that seen with deposits confined to the PAG, several significant differences were noted in both the relative strengths of the projections and the precise locations of labelled fields. With deposits within midbrain reticular formation (including cuneiform nucleus), a very light degree of labelling was observed in structures labelled rostra1 to the posterior commissure, and labelling of midbrain structures appeared denser than with PAG deposits. Additionally, labelling was observed in globus pallidus, a structure which did not label with PAG deposits. Although providing a minor projection to the lateral vestibular nucleus, cerebellar cortical output is almost exclusively to intracerebellar nuclei (chiefly to the fastigial and interpositus nuclei). In contrast, PAG deposits did not label the lateral vestibular nucleus, or the intracerebellar nuclei. The need for certain of these controls was that the horizontal approach to different parts of PAG involved penetration of the cerebellum and colliculi. Contrasting the different patterns of labelling seen with deposits placed at control sites led to confidence that the projections described below resulted from labelling confined to PAG. Areas labeled

with deposits in periaqueductal gray

In order to gain a complete picture of the neural connections of an area it requires not only a description of the course of the axons, but also the identification of their respective terminal fields. After anterograde axoplasmic transport of leucine, silver grains are found in streams corresponding to labelled nerve fibres, or as clouds of fine silver granules representing terminal sites. Terminal fields for

and B. J.

Lhw

HRP were also recognized as tine grains of HRP generally spread diffusely throughout a region. The results of labellinp by the two methods were csscntially similar. Placement of [‘Hlleucine within the three speciticd divisions of the PAG resulted in consistently distinguishable patterns of labelling when assessed by either bright- or dark-field illumination. Few individual structures labelled to anywhere near the same degree when deposits were placed selectively in either ventral. lateral. or dorsal PAG. All labelled regions were represented bilaterally with the ipsilateral component being found to predominate in each case.

Autoradiographic observations Results are presented first in detail for a particular experiment with placement of tritiated leucine at a midcollicular level of ventral PAG. Deposits in ventral PAG disclosed the most extensive projections of any subdivision. A summary of the course of the terminal regions and the fibre projections from ventral PAG is provided by Fig. 3. Contrast is then made with the differences taken in the course of projections, also from a midcollicular level, for lateral and dorsal PAG deposits, and distinctions in the pattern of terminal labelling shown in respect to these. In each case a midcollicular deposit was used as this showed the greatest strength of projection for each complement of fibres. there being only variations in the intensity of labelling with more rostra1 or caudal deposits in each case. Projections determined periaqueductal gray

with deposits

in

ventral

The deposit in experiment I (Figs I, 2, 4) was the most extensive of any within ventral PAG that did not spread into lateral PAG. The injection spread rostrally to the caudal-most level of the oculomotor nucleus and caudally through the PAG to the level of the inferior colliculus, thereby occupying almost the entire length of ventral PAG. Comparison of labelling found with placement of the tracer into ventral PAG in other experiments were equally extensive, whether or not the label appeared to have spread into medial PAG. With smaller deposits the amount of labelling was less, though the actual structures labelled were the same. A detailed description of connections is therefore given with respect to the experiment in which the most clear-cut statement can be made. From these deposits, two ascending (a periventricular and a ventrorostral projection) and two descending bundles (a projection in the dorsal longitudinal fasciculus of Schiitz (DLF), and a ventrocaudal projection) were found. Ascending projections From this ventral PAG deposit, a large proportion of labelled efferent fibres was found to pass rostrally. either through the periventricular fibre system or

Efferents to periaqueductal gray in the rabbit

197

Tholomic component Periventriilor

Hypothalamic

canpoiw

projectlon

cord

Fig. 3. A diagrammatic summary of the major efferent fibre pathways from PAG, and the terminal fields they supply.

through the ventral fibre projection. These fibres diverge to take a ventral and lateral course from the deposit site before proceeding rostrally (Fig. 3). Photomicrographs of examples of labelling of forebrain structures, with a ventral deposit (except where indicated), are to be seen in Fig. 5. Periventricular

jibres

The periventricular fibres provide interconnections between the brainstem and hypothalamus9 The fibres running in the periventricular bundle from the deposit site were found to travel through the PAG and to bifurcate into midline and hypothalamic com-

ponents just anterior to the level of the posterior commissure. The thalamic component was found to ascend further rostrally, branching as it did, to distribute fibres and terminals to the anterior pretectal nucleus, the nucleus of the posterior commissure and to the lateral habenula nucleus. No labelling was observed over the medial habenular nucleus. This bundle continued its course still more rostrally and spread to distribute labelled fibres relatively evenly to, or through, most of the intralaminar and midline thalamic nuclei (Fig. SC). Labelled fibres were seen passing into the posterior and anterior para-

Fig. 4(A-E)-caption

overleaf.

S. T. MELLER and B. J. DENNIS

198

S

Fig. 4(FS) Fig. 4. Schematic outline of the brain, from tekncephafon to spinal cord show, in three plates, the distribution of labelkd fibms and tmminak following a deposit of Mated k&e into ventral PAG. The relative intensity of labelling is indicated on the figuras by the density of dots. ventricular thalamic nuclei where terminal labelling was moderate. Labelled fibrcs and terminal labclling were also seen in nucleus parafaacicularis. Labelling seen over the posterior thalamic nucleus and supra-.

..--

_---_

..~

geniculate nucleus was possibly due, in part, to this thalamic projection. Thalamic fibres appeared to join tire internal and external medullary lamina to label tire paracentral, centre median, central lateral, _...---

. .

.-._.-

- ..-

..--_

Fig. 5. Photomicrograpbs of reprcacntative ascendiug projections and terminal tklds. (A) LabaUing observed in the bed nuekua of the stria tmminahs. (Part of the lateral ventrick is to be sun below the letter A.) (B) Terminal hbalbng in the baaolateral amyg&ioid group. (C) L&a&g throu@ut the dorsal thalamus and madial hypothalamus, resultant on the deposit in vantrd PAG. ?hic oval would eorrwpond to a section midway between kvels D and E in Fig. 4. (D) LabaIling throughout dorsal and medial thalamus, at a kvel just rostra1 to outline E in Fig. 7, resultant on a deposit in lateral PAG. Q Labahing in the lateral hypothalamus, extending into zona incerta (at the top of the llgure). (F) VentraUy directed, labelled fibrcs leaving the site of a lateral deposit; and neighbouring terminal fields. Scale bars = 100 pm.

--

Fig. 5.

700

S. T. MELLER and

intermediodorsal and reticular thalamic nuclei. Although fibres were seen to pass through the ventrobasal complex it was difficult to determine if this group of thalamic nuclei was in fact really labelled. since the density of silver grains over them was barely above background level (in the order of 20-30 grains per 1000 pm2). The mediodorsal nucleus showed evidence of only very light, patchy labelling. Labelled fibres and terminals were also observed over the interanteriomedial. rhomboid and reuniens nuclei although, again, terminal labelling in these thalamic nuclei was relatively light. No label was detectable over anterior thalamic nuclei. As the hypothalamic bundle continued its course rostrally from the PAG through the periventricular region of the hypothalamus (Fig. SC), it spread out laterally from the posterior and medial hypothalamic areas. Fibres left the posterior hypothalamus in a ventrolateral direction contributing to heavy terminal labelling seen over lateral hypothalamus. zona incerta (Fig. SE) and fields of Fore]. This same projection was responsible for light terminal labelling observed lateral to the arcuate nucleus, and possible labelling over the supramammillary and premammillary nuclei. Further rostrally, fibres spread dorsally and laterally to give rise to dense terminal labelling over the dorsal hypothalamic area and the medial part of the lateral hypothalamus. This bundle also contributed to light labelling of fibres and terminals seen over dorsomedial and, to a lesser extent. ventromedial hypothalamic nucleus. Part of the hypothalamic bundle continued rostrally through anterior levels of the hypothalamus where it contributed fibres to the supraoptic decussation. Other fibres in this bundle continued still more rostrally through the medial and lateral preoptic areas, where terminal labelling over these structures was observed. More anteriorly, fibres passed through the fomix and coursed dorsally to contribute to the labelling observed over the lateral septal nucleus. Ventrorostral projection

Other labelled fibres which also followed an ascending course were found to leave the deposit site in a ventrolateral direction through the reticular formation, or to arch dorsally and laterally through the deep and middle layers of the superior colliculus and then travel ventrally, accumulating just medial to the media1 geniculate body. These fibres then passed just lateral to the red nucleus and came to lie above the cerebral peduncle at the level of the substantia nigra and ventral tegmental area. Terminal labelling was observed over the substantia nigra: pars reticulata, pars lateralis, and pars compacta, with the largest accumulation of silver grains being found evident over the latter. Terminals were also detectable over the ventral tegmental areas, midbrain reticular formation and parabigeminal nucleus.

B. J. DESNIS On the ventral surface of the midbrain. labelled fibres from this projection were observed to fan out and turn rostrally to pass over the dorsolaterdl and lateral edge of the mammillary body and cerebral peduncle as a sheet of fibres just ventral to the media1 lemniscus. This bundle distributed branches that passed laterally to terminate in the peripeduncular nucleus, ventral lateral geniculate nucleus, and possibly also contributed to the labelling seen within zona incerta. Fibres which travelled through the subthalamic nucleus probably also gave rise to some of the terminal labelling seen over this area. The mam bundle of fibres continued rostrally to the ventral and lateral region of the lateral hypothalamus where it appeared to join the median forebrain bundle. There it distributed fibres laterally through the substantia innominata to label the entopeduncular nucleus and medial. basolateral (Fig. SB) and central amygdala. Heavy labelling seen in the lateral hypothalamus resulted from fibres entering by this route, as well as through periventricular fibres. Sparse terminal labelling observed over the supramammillary nucleus and the dorsal premammillary nucleus might have been due to fibres coursing in the periventricular system ol fibres, rather than through the ventral projection. No labelled fibres or terminals were seen in the arcuate nucleus. The fibres which passed through the lateral hypothalamus continued rostrally though the anterior hypothalamus to a ventral position within the lateral preoptic nucleus where labelled terminals were observed. Continuing on their course, labelled axons branched to run either rostrally in the diagonal band of Broca (fibres from which then appeared to terminate in the lateral septum), or travelled laterally and dorsally to enter the bed nucleus of the stria terminalis where terminal labelling was heavy (Fig. SA). No labelling was seen over nucleus accumbens. medial septum, ventral pallidum. caudate nucleus. putamen, olfactory tubercle, or in any region of neocortex. Descending projections

Caudally directed fibres either passed within the DLF or swept laterally, then ventrally towards the pons. where they were found to lie in a position just dorsal to the pontine nuclei. Photomicrographic evidence of a number of regions labelled through the caudal outflow is provided in Fig. 6. Dorsal fibres within the dorsal longitudinal fasciculus of Schiirz

Fibres passed from the PAG through the gray matter in the floor of the fourth ventricle distributing terminals to the dorsal tegmental nucleus; then more caudally through the abducens nucleus where an extension of this group of fibres descended in the reticular formation towards the ventral medulla. Terminal labelling was also seen over the abducens nucleus, nucleus supragenualis and nucleus reticularis gigantocellularis.

Efferents to periaqueductal gray in the rabbit

Labelled axons were observed to leave the PAG via a number of alternative, ventrally directed routes. A group of heavily labelled axons was seen to exit from the PAG through the cuneiform nucleus, where heavy terminal labelling was observed (Fig. 6A). These fibres then curved ventrally through the lateral and medial parabrachial nuclei to enter the dorsolateral region of nucleus reticularis pontis oralis. A continuing branch of this projection was found to course ventrally, to then lie above the dorsal surface of the pons before turning medially to give rise to terminal labelling in the tegmental reticular nucleus. Labelling was also seen over the parabigenimal nucleus, ventral and dorsal nuclei of the lateral lemniscus, parabranchial nuclei, locus coeruleus, locus coeruleus pars a, and nucleus pontis oralis. It was difficult to determine the actual degree of terminal labelling in these areas due to the densely labelled fibres traversing them. Locus coeruleus was the most densely labelled caudal structure observed. Other labelled fibres left the PAG through a dorsolaterally orientated pathway and passed through the inferior colliculus to join the fibres in transit through the cuneiform nucleus. A further group left the deposit site at the level of the cuneiform nucleus and brachium conjunctivum to curve ventromedially through the dorsal region of nucleus pontis oralis where they contributed to light labelling observed there, as well as labelling the central superior nucleus.

201

observed in the medial accessory inferior olive. This loose bundle of fibres traversed the caudal medullary reticular formation and consolidated to form a compact bundle before descending through nucleus ambiguus and giving rise to labelled terminals within this nucleus (Fig. 6F). More caudally, labelled fibres were traced to the level of the ventral reticular formation of the medulla, then to the ventrolateral region of the white matter of the spinal cord. Comparison of projections from dorsal and lateral periaqueductal gray with those described for ventral periaqueductal gray

An impression of the differences in the patterns of labelling resulting from deposits of label in each of the three subdivisions can be appreciated by comparing Fig. 4 (for ventral deposit), Fig. 7 (for lateral deposit) and Fig. 8 (for dorsal deposit). Projections disclosed by a ventral deposit embraced all areas labelled with deposits anywhere within PAG. The extent of the projections from the other regions, and the structures they innervated, were considerably reduced on those found with deposits in ventral PAG. Quite some variation was noticed even in the main stream of fibres leaving PAG with dorsal and lateral deposits, compared with the projections detailed above for ventral PAG. In addition there was also some variance in the particular distribution of fibres in lesser streams and the areas they supplied.

Ventrocaudal projections

Projections determined periaqueductal gray

with deposits

in

lateral

It could be that the ventral fibres which were found to give rise to an ascending bundle of fibres located just dorsal to the cerebral peduncle also contributed to the group of caudally directed fibres confluent with those which accumulated dorsal to the pons. These ventrally directed fibres then curved caudalward and travelled either in a loose bundle through the reticular formation, giving rise to terminals in nucleus reticularis pontis caudalis and caudalis pars a, as well as to nucleus reticularis gigantocellularis, or coursed caudally in a position just dorsal to the trapezoid body (Fig. 6D), pyramids (Fig. 6B) and inferior olive (Fig. 6C). This latter bundle spread out as a sheet of fibres which contributed moderate-to-heavy labelling in the regions through which these fibres travelled. In addition, this complement of fibres supplied the raphe nuclei of the brainstem; raphe magnus, pallidus and obscurus. Additional labelling was observed at the dorsolateral aspect of the trapezoid body as well as on the ventrolateral and dorsal surfaces of the superior olive (Fig. 6E), the facial nucleus, nucleus reticularis paragigantocellularis pars CI and nucleus reticularis paragigantocellularis pars lateralis. Labelled fibres detected in the trigeminal nerve distributed to the principal sensory trigeminal nucleus and spinal trigeminal nucleus pars oralis, pars intermedia and pars caudalis. Some of the caudally directed fibres were responsible for the labelling

The main differences in the projections found with lateral or ventral PAG deposits were firstly that, although neither of the ascending bundles of fibres were as impressive as those resulting with deposits in ventral PAG, the periventricular bundle was noticeably stronger than that in the ventral bundle, the main source of the latter being through the deep layers of the superior colliculus. Despite the existence of these projections no labelling was seen over the posterior thalamic nucleus or nucleus suprageniculatus. Further, no fibres were observed to leave the posterior hypothalamus to travel ventrolaterally towards the lateral hypothalamus. It was concluded that labelling found within lateral and medial preoptic nuclei, and dorsal hypothalamus, resulted from fibres in the periventricular bundle. This would explain why the medial hypothalamus was labelled proportionally more than the lateral hypothalamus. The stria terminalis seemed again to derive its input from the ventral bundle. The fibres that formed the ascending ventrally directed bundle were found to be weakly labelled. This bundle did not continue more anteriorly than substantia nigra pars compacta or the parabigeminal nucleus. No labelled fibres were seen to pass through the subthalamic nucleus, and no fibres were distributed laterally through the substantia innominata to any of the regions labelled with ventral PAG deposits. Two

Fig. h

Efferents to periaqueductal gray in the rabbit

Fig. 7(A-K)-caption

overleaf.

Fig. 6. Examples of labelled fibres and terminals found at levels caudal to the deposit site. (A) Photomicrograph shows labelled fibres leaving the PAG through the cuneiform and parabrachial nuclei, and terminals distributed within them, together with streams of labelled fibres continuing towards the pons. (B-E) Descending fibres cut in cross-section. Labelled terminals spread laterally over the pyramids (B), inferior olive(C), trapezoid body (D), and superior olive (E). (F) Terminal labelling within the medulla (in nucleus ambiguus, and to the right in the ventral reticular nucleus). Scale bars = 100 pm.

203

204

S. T. MELLERand B. J. DENNIS

Fig. 7(LS). Fig. 7. Schematic outlines of the brain, from telencephalon to spinal cord, showing the distribution of labelled fibres and terminals following placement of label in lateral PAG. The relative intensity of labelling is indicated by the density of dots.

photomicrographs of examples of labelling seen as a result of a deposit in lateral PAG are given in Fig. SD, F. The second departure from the pattern of labelling seen with ventral deposits was that, with lateral deposits there was no evidence of a dorsal bundle of fibres contributing to descending projections.

Even so, the dorsal tegmental nucleus was lightly labelled. The third distinctive variation seen with lateral deposits was that more descending fibres left the deposit site through the inferior colliculus than through the cuneiform nucleus. These bundles merged in their caudoventral path passing through

Fig. 8(A-E).

Efferents to periaqueductal gray in the rabbit

205

0

Fig. 8(F-S). Fig. 8. Schematic outlines of the brain, from tekncephalon to spinal cord, showing the distribution of labelled fibres and terminals following placement of label in dorsal PAG. The relative intensity of labelling is indicated by the density of dots. the parabigeminal nuclei and dorsal pons over the lateral surface of the trapezoid body and pyramids before passing more caudally to the level of the inferior olive, although there was no lahelling in this nucleus, or in the spinal trigeminal nucleus, nucleus ambiguus or spinal cord. Additionally there was nothing equivalent to the loose projection of fibres seen to project through the medullary reticular formation with deposits in ventral PAG.

Projections determined periaqueductal gray

with

deposits

in dorsal

Whereas examples of deposits in ventral and lateral PAG were fairly well equated for size, those in the dorsal subdivision were centred in the rostra1 twothirds of the complex since no dorsal subdivision exists at the cat&al kvel of the PAG. The periventricular projection, diverging at the level of the posterior commissure, fed rostra1 sites in

206

S. T. MELLEK and B. J. DENNIS

a manner similar to deposits placed in ventral PAG, except for greater labelling throughout the thalamic component. Dorsal deposits were especially distinguished with respect to substantially more labelled fibres emerging from the PAG through the deep and middle layers of the superior colliculus and reticular formation. Terminal distribution of fibres via these routes differed with respect to their being, consistently, more labelling over the lateral preoptic nucleus, less over the stria terminalis, and none within the lateral septal nucleus, in response to labelling the dorsal division. In contrast to a degree of similarity of labelling of rostra1 structures when the label was placed in either dorsal or ventral PAG, the pattern of labelling of caudally directed projections, with dorsal deposits, resembled rather more those found with lateral deposits. There was no evidence of a dorsal bundle caudal to the level of the dorsal tegmental nucleus. This would explain the absence of labelling in the abducens nucleus and dorsal reticular formation of the medulla. The ventral caudal projection consisted mainly of fibres leaving dorsal PAG, via the cuneiform nucleus. The central superior nucleus and dorsal pontine reticular formation were supplied by this route (as with deposits in ventral PAG) although, once caudal to the pons, the distribution of labelled terminals approximated those seen with lateral PAG deposits. Projections described with anterograde horseradish peroxidase

transport

of

A previous report on the afferent connections of PAG’s depicts the deposit sites for HRP at rostra], midcollicular and caudal levels of PAG (see Fig. 2 in Ref. 98). On the whole, experiments employing anterograde transport of HRP from similarly situated deposit sites, having a degree of spread equivalent to that found with deposits in the autoradiographic experiments, essentially confirmed the basic observations reported above. Slight differences in the strength of projections observed with the autoradiographic or HRP procedures are not commented upon as it is impossible to match exactly the deposit sites, or to determine the relative amounts of each tracer transported. Such variations could account for occasional inconsistencies between similarly situated deposits. Comment is therefore made only where significant differences in the degree of labelling of projections were disclosed with respect to the two methods. With deposits of HRP in ventral PAG neither method revealed connections that were not established by the other. For lateral PAG deposits all areas that labelled in the autoradiographic experiment also labelled with HRP, yet in addition anterograde HRP labelling was observed in the ventral lateral geniculate nucleus, lateral septum, central superior nucleus, pontine reticular formation, inferior olive and nucleus ambiguus (which had labelled with

ventral PAG deposits of tritiated leucine, but not with lateral deposits). In the case of HRP deposits in dorsal PAG no labelling was detected in the amygdala, substantia innominata, the caudal extent of the bed nucleus of the stria terminalis, or the pontine reticular formation. whereas some labelling had been detected in these regions with the autoradiographic procedure. It is difficult to imagine what area the deposit would have spread so that could account for disparate observations should these regions not represent true connections of dorsal PAC. DlSCU!SION

This study has clearly recognized differential efferent projections from the dorsal, lateral and ventral regions of the PAG which follow particular rostrally directed (periventricular or ventrolateral bundles) or caudally directed routes (ventral and dorsal bundles). These bundles distribute to more than 80 regions throughout the central nervous system. Comparison with other studies

Although there has not previously been a comprehensive description of the efferent projections and terminal fields, a number of accounts (detailed below), have provided limited descriptions of fibre projections from the PAG through ascending, descending, or locally projecting sys~~~~,29~37.4~~61~62.~~~.~.l~.l~.l~~ Coll&vely,

these

Epoas

have described almost the entire course of projection fibres and terminal fields reported in the present study. In essence (as can be seen in Fig. 3). ascending fibres are comprised of a periventricular bundle which divides into thalamic and hypothalamic components, and a ventrally coursing group of fibres overlying the cerebral peduncle which continues through the lateral hypothalamus. Descending fibres either project caudally in the DLF, to end in the reticular formation, or course ventrally through the cuneiform nucleus to take up a position dorsal to the pons, before continuing more caudally in the ventral medulla, some axons even reaching into the spinal cord. The first description of PAG efferents was made by Mettler,im who lesioned “dorsal PAG”, almost certainly incorporating within it a part of what is now more generally regarded as an area belonging to lateral PAG. He found that ascending fibres left the PAG at the level of the posterior commissure and then divided to supply efferents to the thalamus and to the arcuate area of the hypothalamus. Descending connections were reported to leave the PAG via a ventral route to supply the tegmenturn. reticulotegmental nucleus and to pass as far caudal as the inferior olive. Additional projections and many more sites of innervation have been shown since then. With a lesion in the rostra] ventral PAG, Bucher and BurgiB described fibres which projected ventrally

Efferents to ~~aqu~uc~l to the substantia nigra and then passed through the subthalamus to the endopeduncular nucleus, with a complement of these fibres also being found to pass over the mammillary complex, distributing terminals to this region and to the perifornicai area. In terms of our description these fibres would appear to comprise part of the ventral bundle which leaves the PAG through the cuneiform nucleus. Also with lesions in ventral PAG, Kuypers79 found that the ventrally directed fibres travelled through the lateral hypothalamus and continued as far rostra1 as the preoptic area. In addition to these fibres he described a periventricular system supplying the lateral tuberal and iateral hypothalamic regions; and others radiating out to the mesencephalic reticular formation. Nauta,log with ventral PAG lesions, described a group of fibres in the DLF that supplied fibres and terminals to the intralaminar thalamus and to periventricular, medial and dorsal hypothalamus, thus drawing attention to another component. Ban’ later confirmed this evidence. Experiments by Hamilton,6’ and Hamilton and Skultety,@ indicated that their dorsal, lateral and medial PAG subdivisions gave rise to differential projections through both a periventricular and a ventrally directed system of fibres. Specifically, what they described was that fibres, dividing at the level of the posterior commissure, supplied branches to the anterior pretectal nucleus and lateral habenula nucleus, with the main bundle continuing rostrally to form the thalamic and hypothalamic divisions of the periventricular fibre system. They did not describe descending projections. Chi)’ extended knowledge of the number of areas receiving efferents from the PAG by describing a large number of thalamic and hypothalamic areas receiving projections via the periventricular system of fibres. Further to this, he identified a ventral projection from the PAG supplying the ventral tegmental area and zona incerta. Selective lesions provided evidence that projections from ventral PAG were more extensive than those from dorsal PAG. As is the case with other systems, the complement of PAG efferent connections revealed by lesion studies has now been found to be incomplete. The first report on PAG efferents established by autoradiography was that by Ruda”’ whose deposit sites were limited to regions dorsal and lateral to the aqueduct. Lateral deposit sites were found to provide more efferent projections than did dorsal deposits. Although no mention was made of any ventrally directed fibres, labelling was indicated in zona incerta and the fields of Forel, as though this labelling were derived as part of the periventricular bundle. The number of areas rostra1 to the PAG, that were described as being labelled in this brief report was not extended beyond those already described by Chi” in his lesion study. Labclling at caudal levels was only reported in reticular formation.

gray in the rabbit

207

Mantyh,89sw who used the autoradiographic procedure to describe the ascending and descending projections of the PAG in the monkey (with observations also in the rat and cat), extended, quite substantially, the number of areas shown to receive efferent fibres from the PAG. In contrast to our observations, and those of other authors, he contended that there were no differences in projections arising from deposits placed at different sites within the PAG. Those results led to the conclusion that the ascending fibres were carried almost exclusively in the periventricular bundle, although this is quite at variFurther, no branches ance with other reports. 37,61,62~109 of this bundle were found, in his study, to pass to pretectal and lateral habenula nuclei though this too had been described in earlier lesion studies.6’,62 No actual specification of an ascending ventrally directed bundle was given, although terminal distributions to zona incerta and in the mesencephalic reticular formation were noted. The most rostrally labelled area according to his observations was the anterior hypothalamus. Descending fibres from all deposit sites were described as being carried in a ventrally directed bundle formed from fibres passing through the deep layers of superior colliculus and cuneiform nucleus. An extension of this bundle was found to distribute fibres to reticular areas (the medial reticular areas being most strongly labelled, and lateral areas labelling less); and more caudally to the spinal cord. Eberhart et uI.,~’ though not describing efferents to caudal structures at all, provided a very full description of the ascending projections of the PAG, established by autoradio~aphic means, in the rat. They found, in agreement with many other studies,9~37~6’~62~79.89~‘~,109,117 that there was an ascending ventral projection, as well as a periventricular projection that divided into a dorsal component (which distributed fibres to the thalamus) and ventral component (which distributed fibres to the hypothalamus). However, the pattern of labelling in their experiments-in which deposit sites extended beyond the capsular boundaries of the PAG, and leakage from the pipette is likely to have involved the cortex-was so widespread that just about all structures in the brain were depicted as having labelled. While they acknowledged that cortical contamination could have contributed to the number of structures in which labelling was detected, they were of the opinion that deposits spreading into the territory beyond the bounds of the PAG would not have produced significant additional labelling. Additional structures in which they reported labelling in the rat over those reported here in the rabbit were: frontal cortex, olfactory tubercle, caudoputamen, ventral pallidurn, nucleus accumbens, lateroposterior, laterodorsal and anterioventral nuclei of the thalamus, the interpeduncular nucleus and dorsal lateral geniculate nucleus, and within the internal capsule. With the exception of those structures just listed, the results of our autoradiographic and HRP exper-

208

S. T.

MELLERand

B. J. DENNIS

Deposits of the label in lateral PAG showed a few notable differences in both its efferent fibre prqjection and terminal distribution compared with that seen with the more extensive projection from ventral PAG. With lateral PAG deposits no fibres or terminals were observed in the subthalamic nucleus, substantia innominata, amgydaloid nuclei, or in the suprageniculate thalamic nuclei (all of which were structures Comparison of projections found in autoradiographic which labelled with rostra1 projections from deposits and horseradish peroxidase experiments in ventral PAG); neither was a dorsal, caudally directed bundle apparent, and only one ventrally Very few differences were noted in the distribution directed route was revealed. It is only in considering patterns of labelled terminals and projections estabsuch differences as these, and matching them to the lished by autoradiographic and anterograde HRP functions in which they are shown to be involved, labelling experiments. that we can hope eventually to be able to specify more There is the potential for problems in interpretation of anterograde labelling studies using HRP as closely the neural basis for the involvement of PAG in its various roles. a tracer. It has been proposed that unless histochemiTo this end, it is possible that some clues to cal processing is optimal in the HRP procedure only terminal fields, and not axonal projections, would be function may be inferred from differences in fibre visualized.13’ It is also known that terminals become trajectories. Substantially more fibres were found to labelled before parent axons show an appreciable leave the lateral deposit site through the inferior degree of labelling. 13’Either way the full complement colliculus, and fewer through cuneiform nucleus, in comparison with deposits in ventral PAG. Both of projections would then not be divulged. Difficulties caudal fibre projections and their associated terminal may also arise in determining the efferent connections distributions were essentially similar to the correof a region by the use of anterogradely transported HRP where retrograde labelling of collateral axons sponding pattern with ventral PAG deposits as far also occurs, leading to false conclusions. Also the caudally as the level of the inferior olive, yet no HRP chromogen reaction-product, accumulated labelled fibres or terminals were detectable more within fine dendrites, may be confused with anterocaudally, presumably reflecting the different function of this division. grade labelling. For these various reasons, the results In the case of distribution of fibres from a dorsal dependent on HRP transport to trace efferent condeposit, again a somewhat different pattern was nections of PAG were used only to confirm autofound. The thalamic component of the periventricuradiographic observations. lar bundle was more heavily labelled, and more Comparison of projections from difSerent periaqueduc terminals were seen in the posterior thalamus, yet tal gray deposit sites fewer in the bed nucleus of the stria terminalis. The descending component resembled more closely the The description of projections from PAG is comdistribution found for lateral PAG deposits, once plicated by the fact that various authors have found cause to subdivide it somewhat differently,22@*“o~‘U more indicating that the ventral PAG would appear to be the main source of efferent outflow to the caudal resulting in some variance in descriptions of the regions of the brain. course of fibres leaving each area. In the experiments detailed in our results, deposits were placed in dorsal, Subdivisions of periaqueductal gray lateral or ventral regions of the PAG, in several Our reasons for subdividing PAG into the four instances with no apparent spread of the label into regions outlined in Fig. 2 are based on a statistical any of the neighbouring “subdivisions” or into study of Nissl-stained material.% Since the grounds nearby territory beyond the PAG. Essentially similar for those conclusions are the subject of another results were obtained in each case where comparable paper it is not appropriate to detail the particulars deposits were placed at similar sites. The larger the here. In the present study we have found partial deposit, the heavier was the labelling. At rostrocaudal differentiation of efferent projections with respect levels of the PAG where all subdivisions are repto PAG regions. A previous report on the afferent resented, there were no appreciable differences in labelling compared with midcollicular deposits. We connection of PAG showed likewise.98 Whatever the functional significance of these findings may did not manage to confine deposits within the narrow be it is not to be explained away as a species medial division surrounding the aqueduct since difference, peculiar to the rabbit, since a comparspread of label occurred from this region into both the ventricular system and into other divisions of able arrangement is in evidence also, at least, in the rat.22,23 PAG. However, although the end result of the patThere is close agreement between the regions tern of labelling varied with deposits in each of the defined in our studies of the rabbitsw and the other three sites, no other variant was detected conclusions reached by Beitz22 in respect to the whether or not the medial division was also involved.

iments in the rabbit serve to confirm the areas which have been shown, in less complete studies by others, in a variety of species, to receive efferent projections from the PAG. In addition this study has distinguished the course of ascending and descending projections from deposit sites dorsal, lateral and ventral to the aqueduct.

Efferents to periaqueductal gray in the rabbit

subdivisions proposed for the rat. Despite the fact that some others have failed to demonstrate cytologically significant regions within the PAG8g~~J’0J41J42 or substantial differences between projections related to these regions,8g*90similar demonstrations in two species (Beitz for the rat, ours for the rabbit), enforce the requirement that boundaries defined on anatomical grounds should be considered seriously when behavioural studies are formulated. Using other criteria, such as histochemical evidence,3g or in regard to specific functional zones within this territory,‘* there is also evidence that the PAG is able to be subdivided. Technical considerations

A number of factors must be borne in mind in attempting to describe efferent projections by autoradiography. Should the concentration, or volume, of tracer provided at the deposit site be insufficient, the full complement of connections would not be revealed. At best there would be risk of confusing projection fibres with terminals where axons, lightly labelled, were cut in cross-section. In an attempt to eliminate this problem moderately high concentrations of label were used to obtain heavily labelled axons. In order to decrease diffusion from the deposit site and thus restrict the tracer to the smallest possible area, relatively small volumes have been employed (0.05-0.5 ~1) and the tracer injected at a relatively slow rate. The outer limit of an injection site has, here, been estimated as the area where the density of grains over the somata equalled that over the neuropil. This criterion has been used by other authors,4*x4’yet it may in fact provide an overestimate of the effective size of the deposit site. This means, in the present context that, as a conservative approach has been taken in delimiting the spread of the label, there should be no doubt that deposits did fall well within the regions indicated. Although there may be some potential benefits in using a “cocktail” of tritiated amino acids as labels,24v4git was decided not to incorporate other labels into the injection bolus since leucine has been proven to function well in the central nervous system as a tracer and has a number of distinct advantages over other precursors (proline being the other most commonly used). Leucine is taken up equally well by all neurons (whereas proline may not be utilized as effectively by larger neurons24s4g and although terminal fields are well labelled by proline, projection fibres are not). Leucine has the advantage that both fibres and terminals are well labelled by it, and these are usually readily distinguishable from one another.4g Further, proline diffuses more readily from the injection site, and is known to be transported transynaptically when large deposits and long survival times are employed. In using leucine alone as the label in these experiments, it was in the expectation that this tracer would be transported by all fibres, and

209

therefore, in showing no preferential uptake or undue enhancement of labelling of any area, would give more clear-cut results. In the case involving lateral PAG, where anterograde transport of HRP appears to have labelled structures not disclosed by autoradiographic analysis, it is possible that labelling may not have been observed if the deposit had been of insufficient quantity to permit detection above background levels. It is odd that this was a consistent finding in each experiment in which either one or the other label was used: five with HRP and five with tritiated leucine. No explanation can be offered for this anomalous finding. Reciprocal concentrations of the periaqueductal gray

On the whole, as one might anticipate, there is a considerable degree of reciprocity of connections between regions found to contribute input to the PAG, and those which receive input from it. While our study of afferent connections of PAGg8 provided evidence of input from several areas of cortex to all parts of PAG (in agreement with Benz”), we have no evidence that the PAG provides input to any part of cortex in the rabbit. The only report to provide evidence of such connections in the rat has largely been discounted by its authors.47 We also have no evidence of input to cerebellum, nor to a relatively small number of other structures in the hindbrain which were previously shown to provide input to the PAG.‘* While there is some reciprocity of connection between hypothalamus and thalamus, each receiving projections from the PAG and contributing to it, there appears to be a significant absence of reciprocity between various nuclei within them. Some assessment of this can be made in comparison of the figures showing efferent, and afferent, connections for each deposit site in this and our previous paper.‘* The difference is especially remarkable in the case of the thalamus, where midline and intralaminar nuclei were found to receive input from PAG, but not to have a significant return connection. Presumably these areas must have special significance in the way they process information. Without considerably more evidence, it would be futile to guess just which neural circuits are related to particular functions in which PAG has been implicated. However, certain functional relationships to the PAG have been suggested as outlined below. Functional implications

A large number of recent reports have pointed to the relevance of PAG in information processing and behavioural responsiveness, and drawn attention to different functions with regard to certain areas specified as having connections with PAG. Even so, this exercise still falls short of providing an understanding of meaningful neural circuitry, and presumably will do for some time to come. Since each

210

S.

T. MELLER and

observation puts a new facet on our understanding of this complex structure it is important for anyone engaged in behavioural studies to attempt to characterize their observations with due regard to the anatomy and histochemistry of the regions concerned. Sexual behaviour It is well accepted that the lordosis reflex in sexually receptive female rodents is elicted through the agency of the medial preoptic nucleus and the ventromedial hypothalamus, both of which have been established as having afferent and efferent connections with all parts of PAC. It is also recognized that the PAG is in a position to integrate that reflex by feeding forward ascending somatosensory information to hormone-sensitive regions governing descending influences that control spinal cord output via direct, and indirect pathways. x’J’s Thus the cuneiform nucleus and other structures in the ventral reticular formation, where lesions have been shown to disrupt lordosis, appear to be involved in the actual maintenance of lordosis,64 rather than being involved in the elicitation of this response. Endocrine effects have been seen in relation to agents infused into the PAC. Gonadotrophin-releasing hormone potentiates lordosis and beta-endorphin inhibits it.“’ Luteinizing hormone-releasing hormone which has a major influence in potentiating the lordosis reflex, has been implicated in exerting its effect expressly through the lateral PAG.“8”9’2’ Relevancy of pathways, involving septum and lateral habenula, are now supported by the demonstration of efferent as well as afferent connections of PAG with these forebrain structures. Locomotion

Experimental evidence implicates at least two regions within PAG in locomotion. One is situated in the outer part of lateral PAG, and the second in ventral PAG, However, any involvement of PAG in locomotor control is considered to be only as part of a more general control mechanism involving other neighbouting structures. Thus one “locomotor region” is understood to incorporate ventral PAG, cuneiform nucleus and the brachium conjuctivum.” Another system, involved in the generation of circling behaviour, is served through the agency of “the angular complex” (which incorporates the deep layers of the superior colliculus. lateral PAG, and the mesencephalic reticular formation.“2) Its effect is said to over-ride striatal and substantia nigra motor control. The phenomenon of circling, and the development of body asymmetry, have also been evoked by the injection of GABA agonists and antagonists into the PAG and deep layers of the superior collicu1~s.‘~ The lateral PAG has also been implicated in head movement, and ventral PAG in locomotion, through tectal projection to the pontomedullary reticof these ulospinal neurons. ‘I6 Again connection

8. J. D~NKIS

regions with PAG is supported by our studies of’ afferent and efferent connections. Vocalizarion

Knowledge of the involvement of PAG in vocalization stems largely from behavioural responses found with electrical stimulation of the complex, and in lesion studies. Although involving a motor response, vocalization elicited by PAG stimulation, usually has a large emotional component associated with it.” It is assumed therefore that it integrates a range of emotional states (expressed as rage or fear), with pertinent behavioural responses (reflected in associated autonomic changes, characteristic motor responsesdefensive actions and vocalization). However, it is also involved in vocalization where there are no emotional overtones, as with speciesspecific calls. Electrophysiological recording from PAG has demonstrated that it is crucial for the generation of calls, for previously quiescent cells within it discharge immediately before vocalization starts and cease at the onset of the ~all.~ Vocalization has been produced by stimulation of the PAG in a variety of mammals26~7’~7’~80~RR and non-mammalian ~pecies.~~~‘~~ Other regions, which on stimulation evoke cries do so through their effect on the PAG.‘8.69.7’Even with removal of the entire forebrain. species-specific vocalizations persist, whereas lesions in the PAG produce akinetic mutism.‘~7~*~74’2* Lesions caudal to the PAG have been found to either stop species-specific calls or change them to uncharacteristic sounds.” All the evidence is thus indicative of the importance of the PAG as the locus of integration for inputs necessary in the normal manifestation of this phenomenon.“,” The efferent projections important for the motor aspect of vocalizations are relayed via nucleus ambiguus.” a nucleus to be shown here as receiving a strong projection from ventral PAC. Autonomic

responses and the defence reaction

That a range of autonomic reactions (including pupillary dilatation, piloerection, temperature regulation, bladder and stomach tone, changes in respiration, and cardiovascular responses), occur with stimulation or lesion of the PAC. has often been cited. Many of these effects occur in association with the “defence reaction”. The involvement of several functional circuits, with certain of these responses, was referenced in our discussion of afferent connections of PAG in the rabbit9* and the same regions have now been found to be involved in reciprocal circuitry, so enhancing the commitments made then. A few additional points are worthy of note. Rises in blood pressure have been observed with stimulation of various regions of the PAG.72.87.V2 An increase in renin secretion following midbrain stimulation is thought to be due to a central renin-angiotensin system.% Angiotensin receptors

Efferents to periaqueductal gray in the rabbit

211

have been localized in both the hy~thalamus component of the pain reaction. De&do” recogand PAG,” lending support to the notion that nized that “noxious stimulation evokes a series of hypothalamic-PAG involvement is essential for the protective reflexes which include vocalization, typical expression of central pressor responses. It has been facial expressions, motor, autonomic and psychologisuggested that this is perhaps achieved via nucleus cal manifestations accompanying the sensations of paragigantocellularis since cardiovascular effects (repain”; and this is not to be doubted.& A pharmacosulting from midbrain stimulation) can be abolished logical means by which the two effects can be dissowith bilateral lesions of this nucleus,*6 for which an ciated has recently been demonstrated.‘03 Before behavioural ma~fes~tion implicated the efferent connection has been demons~at~ in this study, as also has connection with the nuclei of the PAG in pain perception, el~trophysiolo~~l evisolitary tract, further substantiating the claim that dence had provided a clue as to its involvement in nociception,43,63~7’well in advance of neuroanatomical this region is involved in cardiovascular and respiratsupport for such connections being demonstrated.76*M ory components of the defence reaction.1~‘2~32 Changes in body temperature cause changes in the Current thought emphasizes a descending modulatory role for PAG in antinociception. However, it firing rate of neurons in the midbrain and it has been proposed that PAG may act through’opiate-depenshould not be forgotten that ascending information, dent mechanisms, coupled with the effects of thyroid channelled particularly through the limbic connections of PAG, could still have a place in explaining releasing hormone and neurotensin to produce this effect.138.‘39 the affective quality of pain, at least in humanss3 The extent to which an animal’s lack of responsiveness to Intestinal motility, lzo besides stomach ton‘e,12ghas noxious stimuli could be attributed to a reduction in been reported to be influenced by PAG stimulation, again through an opiate system. The intimate re- emotional overtones, as much as through activation of centrifugal systems, awaits investigation. Disparity lationship between hypothalamus and PAG, and in the two systems must surely mark a point of PAG input to various medullary nuclei, could again be called into effect to explain these findings, yet in departure in the way in which pain is perceived by animals in contrast to man. Whether regional bias so doing it raises once more the question as to how this structure can be instrumental in achieving so within the PAG complex would be reflected in respect to any form of antinociceptive control achievable via many different effects. the limbic system in the human might have to remain Autonomic effects are associated with a general unresolved unless clinical evidence throws light on the response termed the ‘“defence reaction”10J1~31*32~92J24 subject. whose expression is generated through regions localSince the initial report by Reynolds of potent ized in ventral and lateral PAG. These reactions differ analgesia being generated in the rat during stimufrom the “fear-like”, “pain-like”, aversive, and flight lation of the PAG,‘13 a vast number of reports have responses that are observed predominantly with stimulation of dorsal PAG.~~~~,~,1~*~~3,~~ Electrical appeared concerning the phenomena of opiatestimulation of the PAG (particularly the dorsal reinduced and stimulation-produced analgesia (SPA), gion) induces aversive, switch-off behaviour in which in a wide range of species (see Refs 13, 53, 67, 83, 85, 114, 115, 140, 144). animals will try to escape from the stimulation.40 Evidence that there are different regions in PAG Basically these effects differ, in that stimulation of capable of generating SPA is to be found in reports dorsal PAG causes a suppression of the hypothalamic-elicited flight behaviour, whereas ventral PAG which show that more powerful stimulation is needed stimulation facilities it.27J07 This provides another to elicit analgesia from dorsal and lateral PAG, than example of some degree of specificity of function with stimulation in ventral PAG.s*s’ “Pure” analgebeing associated with one or other region within sia is there said to be elicitabIe only with stimulation PAG, and if not with all of that subdivision at least within ventral PAG, whereas analgesia, accompanied with a specific zone within it. GABA,““ and seroby various other behavioural manifestations, has tonin have been proposed as transmitters involved resulted with stimulation in numerous regions within in the generation of aversive behaviour. Serotonin the remainder of the PAG. Somatotopic organization injected into dorsal PAG raised the threshold for for SPA and PAG has also been inferred.13’ Analgesia induced escape behaviour, implying that this system elicited by the application of opiates within PAG would normally mitigate against the expression of shows regional specificity largely with respect to aversive responses. Iz6 ventral PAG,‘9*~~5z8’~g1 which effect is antagonized by naloxone. 2*6,1g*20 Opiates administered into other Pain and antinociception regions of PAG show much less of an effect,8’J4s in Current researchers in the field dissociate becorrelation with the greater density of p-opioid haviour which can be interpreted as reflecting the binding sites in ventral PAG.lo5 “defence reaction” from any expression of the frank Since the most powerful descending fibre pathways experience of pain, whereas earlier, Spiegel et a1.‘33 from PAG have been shown in this study to arise in had suggested that the vocal, autonomic, rage and ventral PAG (the parallel pathways from the dorsal defence reactions should rightly be considered as a and lateral subdivisions being quite weak in com-

212

S. T. MELLEK

and B. J. DENNIS

therefore further implicating these particular strucparison), it is not surprising that ventral PAG is tures in antinociception and, in some instances (by the region where stimulation has most often been specification of the neurotransmitters involved), indieffective in promoting what has been interpreted as cating more of the mechanism by which they are a degree of analgesia in experimental animals. The entrained.33.‘“.‘43 sites of stimulation which have proven to be most The conclusion to be drawn is that despite there efficacious in demonstrating antinociceptive responbeing a considerable amount of evidence available siveness in animals favour the caudal levels of ventral PAG”“,‘02,‘22,‘23 whereas in humans stimulation at that does implicate PAG in a host of functions, few specific connections have yet been established which rostra1 sites has been shown to be more likely to bring define the relevant anatomical substrate. It has generrelief without introducing distressing side-effects, and aversive reactions.“’ ally not proven possible to rationalize behavioural It is an essential design feature of any descending findings in such a way. Most correlations that have been attempted have only been expressed very control system which should limit, but not exclude, vaguely, indicative of our present state of ignorance, the passage of ascending information, that this should be capable of being turned on or off dependthough the time is approaching when such considering on requirements. In this instance that would seem ations should be integral to any behavioural study. A high degree of collateralization of projections is to be achieved by most of the input, directed to the spinal cord from the PAG, being exerted indirectly. suggested by the fact that so few neurons (those occupying this relatively small volume of the brain) Fibre projections relay first within the reticular formation, (particularly in nucleus raphe magshould supply so many different areas of the brain nus’3,20.49,13s and lateral reticular nucleus,34.45B59,‘“)beThis observation, coupled with the fact that the PAG fore descending to the spinal cord to suppress the has been implicated in so many functions, points to upward passage of messages coded to signal pain. the need for every endeavour to be made to enunciate Lesions of these areas disrupt SPA elicited by stimuunderlying neural circuitry. lation of ventral PAG. Involvement of the hindbrain Apart from emphasizing a need to pay attention to nuclei, reticularis gigantocellularis, magnocellularis the subdivisions indicated here when behavioural and paragigantocellularis, as sites of relay for systems experiments and manipulative procedures are carried from PAG have been regarded as grounds for supout, no particular significance can be drawn from posing that they too are instrumental in exerting them at present, but it is partly against such differcentrifugal control over ascending sensory ences that understanding of functional relevancy input. 3.14.15,58.79 must eventually be made. It is anticipated that the revelation of the patterns of neurotransmitter distriWithout dwelling on details which are not particubution within these systems will also be useful in larly appropriate in an anatomical study such as this, it is significant to note that numerous pharmacounravelling the puzzle as to why so many regions of the brain have connections with this complex. logical tests involving the administration of various transmitters, or their depletion, restoration, or antagin_ onism by other drugs, 2,4,5,16,30,38.56.65,66,78.101,102,132.143 Acknowledgement-The authors wish to thank Dale Caville terferes with SPA or prolongs its effect. Lesions of of the Department of Pathology for his valuable assistance in the preparation of the photographic plates. appropriate nuclei”’ and pathwaysI do likewise,

REFERENCES 1. Adametz J. and O’Leary J. L. (1959) Experimental mutism resulting from periaqueductal lesions in cats. Neurology 9, 636-642. 2. Adams J. E. (1976) Naloxone reversal of analgesia produced by brain stimulation in the human. Pain 2, 161-166. 3. Akaike A., Shibata T., Satoh M. and Takagi H. (1978) Analgesia induced by microinjection of morphine into, and electrical stimulation of, the nucleus reticularis paragigantocellularis of the rat medulla oblongata. Neuropharmacology 17, 775-778. 4. Akil H. and Liebeskind J. C. (1975) Monoaminergic mechanisms of stimulation-produced analgesia. Bruin Res. 94, 279-296. 5. Akil H. and Mayer D. J. (1972) Antagonism of stimulation-produced analgesia by p-CPA, and serotonin synthesis inhibitor. Brain Res. 44, 692-697. 6. Akil H., Mayer D. J. and Liebeskind J. C. (1976) Antagonism of stimulation produced analgesia by naloxone, a narcotic antagonist. Science 191, 961-962. 7. Bailey P. and Davis E. W. (1942) Effects of lesions of the periaqueductal gray matter in the cat. Proc. Sot. exp. Biol. Med. 51, 305-306. 8. Bailey P. and Davis E. W. (1944) Effects of lesions of the periaqueductal gray in Mocaca mulatta. J. Neuropath. exp Neural. 3, 69-72. 9. Ban T. (1964) The hypothalamus, Med. J. Osaka Univ. 15, l-83.

especially on its fiber connection, and the septopreoptico-hypothalamic

system.

10. Bandler R., Depaulis A. and Vergnes M. (1985) Identification of midbrain neurons mediating defensive behaviour in the rat by microinjections of excitatory amino acids. Behuv. Brain Res. 15, 107-I 19. 11. Bandler R., Prineas S. and McCulloch T. (1985) Further localization of midbrain neurons mediating the defence reaction in the cat by microinjections of excitatory amino acids. Neurosci. Left. 56, 31 l--316.

Efferents to periaqueductal gray in the rabbit

213

12. Bandler R. and Tork I. (1987) Midbrain periaqueductal gray region in the cat has afferent and efferent connections with solitary tract nuclei. Neurosci. Letl. 74, l-6. 13. Basbaum A. I. and Fields H. L. (1978) Endogenous pain control mechanisms; review and hypothesis. Ann. Neurol. 4, 451-462. 14. 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. camp. Neural. 187, 513-532. 15. Basbaum A. I., Clanton C. H. and Fields H. L. (1976) Ascending projections of nucleus gigantocellularis and nucleus raphe magnus in the cat. An autoradiographic study. Anar. Rec. 184, 354. 16. Basbaum A. I., Marley N. J. E., G’Keefe J. 0. and Clanton C. H. (1977) Reversal of morphine and stimulus-produced analgesia by subtotal spinal cord lesions. Pain 3, 43-56. 17. Basbaum A. I., Moss M. S. and Glazer E. J. (1983) Gpiate and stimulation-produced analgesia: the contribution of the monoamines. In Advances in Puin Research and Therapy (eds Bonica J. J., Lindblom U. and Iggo A.), Vol. 5, pp. 324-339. Raven Press, New York. 18. Bazett H. C. and Penfield W. G. (1922) A study of the Sherrington decerebrate animal in the chronic as well as the acute condition. Bruin 45, 185-265. 19. Behbehani M. M. (1981) Effect of chronic morphine treatment on the interaction between the periaqueductal gray and the nucleus raphe magnus in the rat. Neuropharmucology 20, 581-586. 20. Behbehani M. M. and Fields H. L. (1979) Evidence that an excitatory connection between the periaqueductal gray and nucleus raphe magnus mediates stimulation produced analgesia. Bruin Res. 170, 85-93. 21. Beitz A. J. (1982) The organization of afferent projections to the periaqueductal gray of the rat. Neuroscience 7, 133-159. 22. Beitz A. J. (1985) The midbrain periaqueductal gray in the rat. I. Nuclear volume, cell number, density, orientation, and regional subdivisions. J. camp. Neural. 237, 445-459. 23. Beitz A. J. and Shepard R. D. (1985) The midbrain periaqueductal gray in the rat. II. A Golgi analysis. J. camp. Neural. 237, 460-475. 24. Berkley K. J., Graham J. and Jones E. G. (1977) Differential incorporation of tritiated proline and leucine by neurons of the dorsal column nuclei in the cat. Bruin Res. 132, 485-505. 25. Brown J. L. (1972) An exploratory study of vocalization areas in the brain of the Redwinged Blackbird (Ageluius phoeniceus). Behuviour 39, 91-127. 26. Brown T. G. (1915) Note on the physiology of the basal ganglia and mid-brain of the anthropoid ape, especially in reference to the act of laughter. J. Physiol., Lond. 49, 195-207. 27. Brutus M., Shaikh M. B. and Siegel A. (1985) Differential control of hypothalamically elicited flight behavior by the midbrain periaqueductal gray in the cat. Behuv. Bruin Res. 17, 235-244. 28. Bucher V. M. and Burgi S. M. (1953) Some observations on the fiber connections of the di- and mesencephalon in the cat. Part III. The supraoptic decussations. J. camp. Neural. 98, 355-379.

29. Bucher V. M. and Burgi S. M. (1953) Some observations on the fiber connections of the di- and mesencephalon in the cat. Part IV. The ansa lenticularis, pars ascendens mesencephalica, with observations on other systems ascending from and descending to the mesencephalon. J. camp. Neural. 99, 415-435. 30. Calcutt C., Hangley S., Sparkes C. and Spencer P. (1973) Roles of noradrenalin and 5-hydroxytryptamine in the antinociceptive effects of morphine. In Agonists and Antagonist Actions of Narcotic Analgesic Drugs (eds Kosterlitz H., Collier M. and Villareal J.), pp. 176-191. University Park Press, Baltimore. 31. Carobrez A. P., Schenberg L. C. and Graeff F. G. (1983) Neuroeffector mechanisms of the defense reaction in the rat. Physiol. Behuv. 31, 439-444. 32. Carrive P., Bandler R. and Dampney R. A. L. (1988) Anatomical evidence that hypertension associated with the defence reaction in the cat is mediated by a direct projection from a restricted portion of the midbrain periaqueductal gray to the subretrofacial nucleus of the medulla. Bruin Res. 460, 339-345. 33. Carstens E., Fraunhoffer M. and Zimmerman M. (1981) Serotonergic mediation of descending inhibition from the midbrain periaqueductal gray, but not reticular formation, of spinal nociceptive transmission in the cat. Pain 10, 149-167. 34. Carstens E. and Watkins L. R. (1986) Inhibition of the responses of neurons in the rat spinal cord to noxious skin heating by stimulation in midbrain periaqueductal gray or lateral reticular formation. Bruin Res. 382, 266-277. 35. Changaris D. C., Severs W. B. and Keil L. C. (1978) Localization of angiotensin in rat brain. J. Histochem. Cyfochem. 26, 593-607. 36. Changiz G. and Asdourian D. (1984) Circling and bodily asymmetry induced by injection of gamma-aminobutyric acid agonists and antagonists into the superior colliculus. Phurmuc. Biochem. Behuv. 21, 853-858. 37. Chi C. C. (1970) An experimental silver study of the ascending projections of the central gray substance and adjacent tegmentum in the rat with observations in the cat. J. camp. Neural. 139, 259-272.

38. Cicero T. J., Meyer E. R. and Smithloff B. R. (1974) Alpha adrenergic blocking agents: antinociceptive activity and enhancement of morphine-induced analgesia. J. Phurmuc. exp. Ther. 189, 72-82. 39. Conti F., Barbaresi P. and Fabri M. (1988) Cytochrome oxidase histochemistry reveals regional subdivisions in the rat periaqueductal gray matter. Neuroscience 24, 629-633. 40. Delgado J. M. R. (1955) Cerebral structures involved in transmission and elaboration of noxious stimulation. J. Neurophysiol. 18, 261-275. 41. Dennis B. J. (1985) Centrifugal control mechanisms: Somaesthesia. Proc. Aust. physiol. phurmuc. Sot. 16, 23-41. 42. Dennis B. J., Dodman A. P. and Kerr D. I. B. (1983) A rabbit head holder for attachment to a conventional stereotaxic machine. Bruin Res. Bull. 10, 723-726. 43. Dennis B. J. and Kerr D. I. B. (1961) Somaesthetic pathways in the marsupial phalanger, Trichosurus vulpecula. Aust. J. exp. Biol. 39, 29-42. 44. Di Scala G., Schmitt P. and Karli P. (1984) Flight induced by infusion of bicuculline methiodide into periventricular structures. Bruin Res. 309, 199-208. 45. Duggan A. W. (1983) Injury, pain and analgesia. Proc. Ausr. physiol. pharmuc. Sot. 14, 218-240. 46. Duggan A. W. (1984) The suppression of pain. Proc. Aust. physiol. pharmuc. Sot. 15, 25-44.

S.

214

T.

MELLER and B. J. DENNIS

47. Eberhart J. A., Morrell J. I., Krieger M. S. and Pfaff D. W. (1985) An autoradiographic study of projecttons ascending from the midbrain central gray, and from the region lateral to it, in the rat. J. romp. Neural. 241, 285-310. 48. Edwards S. B. and De Olmos J. S. (1976) Autoradiographic studies on the midbrain reticular formation: ascending projections of nucleus cuneiformis. J. camp. Neural. 165, 417-432. 49. Edwards S. B. and Hendrickson A. (1981) The autoradiographic tracing of axonal connections in the central nervous system. In Neuroonaromicul Tracr-rracing Method7 (eds Heimer L. and Robards M. J.). pp. 171-205. Plenum Press.

New York. 50. Fardin V., Oliveras J.-L. and Besson J.-M. (1984) A reinvestigation

51.

52. 53. 54. 55. 56. 57. 58.

of the analgesic effects induced by stimulation of the periaqueductal gray matter in the rat. 1. The production of behavioral side effects together with analgesia. Brain Rex 306, 105-123. Fardin V., Oliveras J.-L. and Besson J.-M. (1984) A reinvestigation of the analgesic effects induced by stimulation of the periaqueductal gray matter in the rat. II. Differential characteristics of the analgesia induced by ventral and dorsal PAG stimulation. Bruin Res. 306, 125 -139. Fields H. L. and Anderson S. D. (1978) Evidence that raphespinal neurons mediate opiate and midbrain stimulation-produuzd analgesias. Pain 5, 333. 349. Fields H. L. and Basbaum A. 1. (1978) Brainstem control of spinal pain-transmission neurons. A. Ret:. Physiol. 40, 2 I7--248. Ganong W. F. (1977) The renin-angiotensin system and the central nervous system. Fedn Proc. 34, 1771-I 775. Garcia-Rill E., Skinner R. D., Gilmore S. A. and Chvings R. (1983) Connections of the mesencephalic locomotor remon (MLR). II. ABerents and efferents. Brain Res. Bull. 10, 63-71. Gebhart G. F. (1982) Opiate and opioid peptide effects on brainstem neurons: relevance to nociception and antinociceptive mechanisms. Puin 12, 93-140. Gebhart G. F. and Toleikis J. R. (1978) An evaluation of stimulation-produced analgesia in the cat. Expl Newof. 62, 570-579. Gray B. G. and Dostrovsky J. 0. (1983) Descending inhibitory influences from periaqueductal gray, nucleus raphe magnus, and adjacent reticular formation. I. Effects on lumbar spinal cord nociceptive and non-nociceptive neurons.

J. Neurophysiol. 49, 932-941. 59. Hall J. G., Duggan A. W., Morton C. R. and Johnson S. M. (1982) The location of brainstem neurons tonically inhibiting dorsal horn neurons of the cat. Brain Res. 244, 215-222. 60. Hamilton B. L. (1973) Cytoarchitectural subdivisions of the periaqueductal gray matter in the cat. J. camp. Neural.

149. l-28. 61. Hamilton B. L. (1973) Projections of the nuclei of the periaqueductal gray matter in the cat. J. camp. Neural. 152, 45558. 62. Hamilton B. L. and Skultety F. M. (1970) Efferent connections of the periaqueductal gray matter in the cat. J. camp. Neural. 139, 105-l 14. 63. Hara T., Favale E., Rossi G. F. and Sacco G. (1961) Responses in mesencephalic reticular formation and central gray matter evoked by somatic peripheral stimuli. ExpI Neural. 4, 297-309. 64. Havens M. D. and Rose J. D. (1984) Impaired lordosis response in golden hamsters with lesions in the ventromedial midbrain. Expi Neurol. 66, 583-589. 65. Hayes R. L., Newlon P. G., Rosecrans J. A. and Mayer D. J. (1977) Reduction of stimulation-produced analgesia by lysergic acid diethylamide, a depressor of serotonergic neural activity. Bruin Res. 122, 367-372. 66. Hosobuchi Y. and Wemmer J. (1977) Disulfiriam inhibition of development of tolerance to analgesia induced by central grey stimulation in humans. Eur. J. Pharmac. 43, 385-387. 67. Hosobuchi Y., Adams J. E. and Linchitz R. (1977) Pain relief by electrical stimulation of the central gray matter in

humans and its reversal by naloxone. Science 197, 183-186. 68. Hunsperger R. W. (1956) Role of the suhtantia g&a centralis mesencephali in electrically-induced rage reactions. In Progress in Neurobiology (ed. Ariens-Kappers J.), Vol. I, pp. 289-294. Elsevier, Amsterdam. 69. Jurgens U. (1983) ABerent fibres to the cingtdar vocalization region in the monkey. Expl Neural. SO, 395-409. 70. Jurgens U. and Muller-Preuss P. (1977) Convergent projections of different limbic vocalizational amas in the squirrel monkey. Expl Brain Res. 29, 75-83. 71. Jurgens U. and Pratt R. (1979) Role of the periaqueductal gray in vocal expression of emotion. Bruin Res. 167, 367-378. 72. Kabat H., Magoun H. W. and Ranson S. W. (1935) Electrical stimulation of points in the forebrain and midbrain: the resultant alterations in blood pressure. Archs Neurof. Psychiuf. 34, 931-955. 73. Kanai T. and Wang S. C. (1962) Localization of the central vocalization mechanism in the brain stem of the cat. Expl Neural. 6, 426-434. 74. Kelly A. H., Beaton L. E. and Magoun H. W. (1946) A midbrain mechanism for facie-vocal activity. /. Neurophysiol. 9, 181-189. 75. Kerr D. 1. B., Haugen F. P. and Melzack R. (1955) Responses evoked in the brain stem by tooth stimulation. Am. J. Physiol. 183, 253-258. 76. Kerr F. W. (1975) The ventral spinothalamic tract and other ascending systems of the ventral funiculus of the spinal cord. J. camp. Neural. 159, 335-356. 77. Kiser R. S., Lebovitz R. M. and German D. C. (1978) Anatomical and pharmacologic differences between two types of aversive midbrain stimulation. Bruin Res. 155, 331-342.

78. Kuraishi Y., Harada Y., Satoh M. and Takagi H. (1979) Antagonism by phenoxybenzamine of the analgesic effect of morphine injected into the nucleus reticularis gigantoccllularis of the rat. Neuropharmaco~ogy 18, 107-I 10. 79. Kuypers H. G. J. M. (1956) Certain fiber connections of the mesenphalic central gray matter. In Progress in Neurobiology (ed. Ariens-Kappers J.), Vol. I, pp. 264-272. Elsevier. Amsterdam. 80. Larson C. R. and Kistler M. K. (1984) Periaqueductal gray neuronal activity associated with laryngeal electromyography and vocalization in the awake monkey. Neurosci. Lea. 46, 261-266 81. LewisV. A. and Gebhart G. F. (1977) Evaluation of the periaqueductal gray (PAG) as a morphine-specific locus of action and examination of morphine induced and stimulation produced analgesia at coincident PAG loci. Brain Res. 124. 283-303.

Efferents to periaqueductal gray in the rabbit

215

82. Liebeskind J. C. and Mayer D. J. (1971) Somatosensory evoked responses in the mesencephalic central gray matter of the rat. Brain Res. 27, 133-151. 83. Liebeskind J. C., Mayer D. J. and Akil H. (1974) Central mechanisms of pain inhibitions: studies of analgesia from focal brain stimulation. Ado. Neural. 4, 261-269. 84. Liebman J. M., Mayer D. J. and Liebeskind J. C. (1970) Mesencephalic central gray lesions and fear-motivated behaviour in rats. Brain Res. 23, 353-370. 85. Long D. M. (1978) Electrical stimulation of the nervous system for pain control. In Contemporary Clinical Neurophysiofogy (eds Cobb W. A. and Van Duijn H.), pp. 343-348. Elsevier, Amsterdam. 86. Lovick T. A. (1985) Projections from the diencephalon and mesencephalon to nucleus paragigantocellularis lateralis in the cat. Neuroscience 14, 853862. 87. Lovick T. A. (1985) Ventrolateral medullary lesions block the antinociceptive and cardiovascular responses elicited by stimulating the dorsal periaqueductal gray matter in rats. Pain 21, 241-252. 88. Magoun H. W., Atlas D., Ingersoll E. H. and Randon S. W. (1937) Associated facial, vocal and respiratory components of emotional expression: an experimental study. J. Neural. Psychopath. 17, 241-255. 89. Mantyh P. W. (1983) Connections of midbrain periaqueductal gray in the monkey. I. Ascending efferent projections. J. Neurophysiol. 49, 567-58 1. 90. Mantyh P. W. (1983) Connections of midbrain periaqueductal gray in the monkey. II. Descending efferent projections. J. Newophysiol. 49, 582-594. 91. Mayer D. J. and Liebeskind J. C. (1974) Pain reduction by focal electrical stimulation of the brain: an anatomical and behavioral analysis. Brain Res. 68, 73-93. 92. McDougall A., Dampney R. and Bandler R. (1985) Cardiovascular components of the defence reaction evoked by

excitation of neuronal cell bodies in the midbrain periaqueductal gray of the cat. Nenrosci. Left. 60, 69-76. 93. Meessen H. and Olsxweski J. (1949) A Cytoarchitectonic Atlas of the Rhombencephalon of the Rabbit. Karger, Basel. 94. Mehler W. R. (1957) The mammalian “pain tract” in phylogeny. Anal. Rec. 127, 332. 95. Meller S. T. and Dennis B. J. (1981) A posterior approach to periaqueductal gray in a labelling study. Proc. Aust. physiol. pharmac. Sot. 12, 167P. 96. Meller S. T. and Dennis B. J. (1986) Autoradiographic mapping of efferent connections of the periaqueductal gray with the prosencephalon, in the rabbit. Proc. Aust. physiol. pharmac. Sot. 17, 46P 97. Meller S. T. and Dennis B. J. (1986) Descending efferent projections of periaqueductal gray in the rabbit. Proc. Aust. physiol. pharmac. Sot. 17, 165P. 98. Meller S. T. and Dennis B. J. (1986) Alferent projections to the periaqueductal

99. 100. 101.

102. 103.

104.

gray in the rabbit. Neuroscience 19, 927-964. Meller S. T. and Dennis B. J. (1991) Quantitative Nissl study of the neuronal types, and recognition of cytoarchitectural subdivisions, with the rabbit periaqueductal gray. J. camp. Neural. (in press). Mettler F. A. (1944) The tegmento-olivary and central @mental fasciculus. J. camp. Neural. 80, 149-175. Mohrland J. S. and Ge.bhart G. F. (1980) Effects of focal electrical stimulation and morphine microinjection in the periaqueductal gray of the rat mesencephalon on neuronal activity in the medullary reticular formation. Brain Res. 201, 23-37. Mohrland J. S. and Gebhart G. F. (1981) Effect of morphine administered in the periaqueductal gray and at the recording locus on nocisponsive neurons in the medullary reticular formation. Brain Res. 225, 40-412. Morgan M. M., Depaulis A. and Liebeskind J. C. (1987) Diazepam dissociates the analgesic and aversive effects of periaqueductal gray stimulation in the rat. Brain Res. 423, 395-398. Morton C. R., Duggan A. W. and Zhao Z. Q. (1984) The effects of lesions of medullary midline and lateral reticular areas on inhibition in the dorsal horn produced by periaqueductal gray stimulation in the cat. Brain Res.

301, 121-130. 105. Moskowitz A. S. and Goodman

R. R. (1984) Light microscopic autoradiographic localiitions of p and 6 opioid binding sites in the mouse central nervous system. J. Neurosci. 3, 1437-1449. 106. Moss M. S., Glazer E. J. and Basbaum A. I. (1983) Light and electron microscopic observations of leucine enkephalin in the cat periaqueductal gray. Ado. Pain Res. Ther. 5, 265-271. 107. Nakao H., Yoshida M. and Sasaki T. (1968) Midbrain central gray and switch-off behavior in cats. Jup. J. Physiol. 18, 462-470.

108. Nashold B. S. Jr, Wilson W. P. and Slaughter D. G. (1969) Sensations evoked by stimulation in the midbrain of man. J. Neurosurg. 30, 14-24. 109. Nauta W. J. H. (1958) Hippocampal projections and related neural pathways to the mid-brain in the cat. Brain 81, 319-341.

110. Olsxweski J. and Baxter D. (1954) Cytoarchitecture of the Human Brain Stem. J. P. Lippincott, Philadelphia. 111. Prieto G. J., Cannon J. T. and Liebeskind J. C. (1983) N. raphe magnus lesions disrupt stimulation-produced analgesia from ventral but not dorsal midbrain areas in the rat. Brain Res. 261, 53-57. 112. Reavill C., Muscatt S., Leigh P. N., Jenner P. and Marsden C. D. (1984) Enhanced GABA function in the angular complex (lateral periaqueductal gray matter and adjacent reticular formation) alters the postural component of striatal- or nigral-derived circling. Expl Bruin Res. 56, l-11. 113. Reynolds D. V. (1969) Surgery in the rat during electrical analgesia induced by focal brain stimulation. Science 164, 444-44s. 114. Richardson D. E. (1979) Central gray stimulation for the control of cancer pain. In Aduances in Pain Research and Theraw (eds Bonica J. J. and Ventafridda V.). Vol. 2. DD. 487-492. Raven Press. New York. 115. Richardson E. D. and Akil H. (1977) Pain relief by elkE&al brain stimulation in man. I. Acute administration in periaqueductal and periventricular sites. J. Neurosurg. 47, 178-183. 116. Rose J. D. (1985) Dorsal-ventral differences in the midbrain distribution of single neurons with head movementcorrelated and locomotion-correlated firing in the golden hamster. Expf Neural. 87, 225-234. 117. Ruda M. A. (1975) An autoradiographic study of the efferent connections of the midbrain central gray in the cat. Anat. Rec. 181, 468-469. 118. Sakuma Y. and Pfaff D. W. (1980) Convergent effects of lordosis-relevant somatosensory and hypothalamic influences on central gray cells in the rat mesencephalon. Expl Neural. 70, 269-28 1.

S. T. MELLER and B. J. DENNIS

216

119. Sakuma Y. and Pfaff D. W. (1983) Modulatjon of the lordosis reflex of female rats by LHRH, its antiserum and analogs in the mesencephalic central gray. Neuroendocr~no~ogy 36, 218-224. 120. Sala M., Parolaro D., Crema G., Spazzi L.. Giagnoni G., Cesana R. and Gori E. (1983) Involvement of periaqueductal gray matter in intestinal effect of centrally administered morphine. Eur. J. Pharmac. 91, 251-254. 121. Samson W. K., McCann S. M., Chud L., Dudley C. A. and Ross R. L. (1980) Intra- and extrahypothalamic luteinizing hormone-releasing hormone (LH-RH) distribution in the rat with special reference to mesencephaton sites which contain both LHRH and single neurons responsive to LHRH. N~roe~docr~no~ogy 31, 66-72. 122. Sandkiihler J. and Gebhart G. F. (1984) Characterization of inhibition of a spinal nociceptive reflex by stimuiation medially and laterally in the midbrain and medulla in the pentobarbital-anaesthetized rat. Brain Aes. 305, 67-76. 123. Sandkiihler J. and Gebhart G. F. (1984) Relative contributions of the nucleus raphe magnus and adjacent reticular formation to the inhibition by stimulation in the periaqueductal gray of a spinal nociceptive reflex in the pentobarbital-anaesthetized rat. Brain Res. 305, 77-87. 124. Schenberg L. C., De Aguiar J. C. and Graeff F. G. (1983) GABA modulation of the defense reaction induced by brain electrical stimulation. Physics. Behav. 31, 429-437. 125. Schmidt R. S. (1966) Central mechanisms of frog calling. Behaviour 26, 251-285. 126. Schutz M. T. B., De Aguiar J. C. and Graeff F. G. (1985) Antiaversion role of serotonin in the dorsal periaqueductal gray. Psychopharmacology 85, 340-345. 127. Sirinathsinghji D. J. S. (1985) Modulaton of lordosis behaviour in the female rat by corticotrophin releasing factor, beta-endorphin and gonadotropin releasing hormone in the mesencephalic centrai gray. Brain Res. 336, 45-55. 128. Skultety F. M. (1958) The behavioral effects of destructive lesions of the ~~aqu~uc~l gray matter in adult cats. J. camp. Neuroi. 110, 337-365. 129. Skultety F. M. (1959) Relation of periaqueductal gray matter to stomach and bladder motility. Neurology 9, 190-198. 130. Skultety F. M. (1963) Stimulation of the periaqueductal gray and hypothalamus. Archs Neural. 8, 609-620. 131. Soper W. Y. and Melzack R. (1982) Stimulation produced analgesia: evidence for somatotopic organization in the midbrain. Brain Res. 251, 301-311. 132. Sparkes C. and Spencer P. (1971) Antinoci~ptive activity of morphine after injection of biogenic amines in the cerebral ventricles of the conscious rat. Br. J. Pharmac. 42, 230-241. 133. Spiegel E. A., Kletzkin M. and Szekely E. G. (1954) Pain reactions upon stimulation of the tectum mesencephali. J. Neuropath. exp. Neural. 13, 212-220.

134. Taber E. (1961) The cytoarchitecture of the brain stem of the cat. I. Brain stem nuclei of the cat. J. camp. Neurnl. 116, 27-69. 135. Tsubokawa T.. Yamamoto T., Katayama Y. and Moriyasu N. (1981) Diencephaiic m~ulation of activities of raphe--spinal neurons in the cat. Exp/ Neurol. 74, X-572. 136. Valenstein E. S. (1965) Independence of approach and escape reactions to electrical stimulation of the brain. J. camp. Physiol. Psychol. 60, 20.-30. 137. Warr W. B., de Olmos J. S. and Heimer L. (1981) Horseradish peroxidase: the basic procedure. In Neuroanatomical Tract-tracing Methods (eds Heimer L. and Robards M. J.), pp. 207-262. Pienum Press, New York.

138. Widdowson P. S., Griffiths E. C. and Slater P. (1983) Body temperature effects of opioids admi~ister~ into the periaqueductal gray area of rat brain. Regul. Pept. 7, 259-267. 139. Widdowson P. S., Griffiths E. C., Slater P. and Yajima H. (1983) Effect of mammalian and avian neurotensins and body neurotensin ragments on wet-dog shaking and body temperature in the rat. Regul. Pept. 7, 357-366. 140. Willis W. D. (1982) Control of Nociceptiue Transmission in the Spinal Cord. Springer, Heidelberg. 141. Winkler C. and Potter A. (1911) An Anatomicai Guide to Experimental Researches on the Rabbit Brain. W. Versluys, Amsterdam. 142. Winkler C. and Potter A. (1914) An Anatomical Guide to Experimental Researches on the Cat’s Brain. W. Versiuys, Amsterdam. 143. Yaksh T. L. (1979) Direct evidence that spinal serotonin and noradrenaline terminals mediate the spinal antinociceptive effects of morphine in the periaqueductal gray. Brain Res. 160, 180-185. 144. Yaksh T. L. and Rudy T. A. (1978) Narcotic analgetics: CNS sites and mechanisms of action as revealed by intracerebral injection techniques. Pain 4, 299-359. 145. Yaksh T. L., Yeung J. C. and Rudy T. A. (1976) Systematic examination in the rat of brain sites sensitive to the direct application of morphine: observation of differential effects within the periaqueductal gray. Brain Res. 114, 83-103. (Accepted 30 May 1990)