THE JOURNAL OF COMPARATIVE NEUROLOGY 302:33&348

Afferent and Efferent Connectionsof the Odornotor Region of the Fadigid Nucleus in the Macaque Monkey HIROHARU NODA, SHOE1 SUGITA, AND YOICHI IKEDA Visual Science Department, School of Optometry, Indiana University, Bloomington, Indiana 47405

ABSTRACT Afferent and efferent connections of the fastigial oculomotor region (FOR) were studied in macaque monkeys by using axonal transport of wheat germ agglutinin conjugated horseradish peroxidase (WGA-HRP). When injected HRP is confined to the FOR, retrogradely labeled cells appear in lobules VIc and VII of the ipsilateral vermis and in group b of the contralateral medial accessory olive (MAO). In reference to the maps of topographical organization, the extent of the effective site in the fastigial nucleus (FN) could be assessed from the distributions of labeled Purkinje cells (P cells) in the vermis and labeled olivary neurons in the MAO. In contrast to the unilateral nature of the P-cell and climbing-fiber projections, those from the other brainstem regions to the FOR were bilateral. Following the injection of HRP into the FOR, the largest number of retrogradely labeled cells appeared in the pontine nuclei. Although the number of labeled cells was greater on the contralateral side in both the peduncular and dorsomedial pontine nuclei (DMPN), the number on each side was virtually identical in the dorsolateral pontine nucleus (DLPN). In the nucleus reticularis tegmenti pontis (NRTP), labeled cells were located only in its medial and dorsolateral portions bilaterally. In the vestibular complex, labeled cells appeared in the superior (SVN), medial (MVN), and inferior vestibular nuclei (IVN) bilaterally. The lateral vestibular nucleus (LVN), including y group and the ventrolateral vestibular nucleus, were free of labeled cells. Labeled cells appeared also in the perihypoglossal nucleus (PHN) bilaterally. In the pontine raphe (PR) and paramedian pontine reticular formation (PPRF), labeled cells appeared bilaterally in the caudal third of the area between the oculornotor and abducens nuclei. Labeled cells appeared also in the mesencephalic and medullary reticular formation. Tracing of anterogradely labeled axons demonstrated that most fibers from the FOR decussated within the cerebellum and entered the brainstem via the contralateral uncinate fasciculus. Some crossed fibers ascended with the contralateral brachium conjunctivum and terminated in the midbrain tegmentum. A small contingent of fibers advanced further to the thalamus. In the mesodiencephalic junction, labeled terminals were found contralaterally in the rostral interstitial nucleus of medial longitudinal fasciculus (riMLF) and a medial portion of Forel’s H Field. They appeared also in the central mesencephalic reticular formation (cMRF), the periaqueductal gray (PAG), the posterior commissure nucleus, and the superior colliculus. The oculomotor and trochlear nuclei, the red nucleus, and the interstitial nucleus of Cajal were free of labeled terminals. The projections from the FOR to the vestibular complex were bilateral and were mainly to the ventral portions of the LVN and IVN. The ipsilateral fibers emerged with the juxtarestiform body. The PHN, the SVN, and the MVN were free of terminals. Crossed fibers arising from the FOR terminated in a caudal portion of the PR, the PPRF, the DMPN, the NRTP, and the dorsomedial portion of the medullary reticular formation (DMRF). Labeled terminals in the DMRF were also found on the ipsilateral side, although the contralateral side was predominant. When the HRP injection involved a fastigial area rostral to the FOR, labeled terminals appeared also in the PHN, SVN, MVN, and DLPN of the contralateral side. Key words: HRP study, eye movements, cerebellum, fastigial afferents, fastigial efferents

Accepted August 14,1990.

o 1990 WILEY-LISS, INC.

(1990)

CONNECTIONS OF FASTIGIAL OCULOMOTOR REGION Recent microstimulation studies on monkeys have shown that the cerebellar cortex, which is related to saccadic function, is located in lobules VIc and VII of the vermis. This vermal area is designated as the oculomotor vermis and characterized by low thresholds for evoking saccades ( < 10 FA) and by saccade-related neuronal activity (Noda and Fujikado, '87b). The saccade evoked by the vermal stimulation has been shown to be the result of activation of P-cell axons (Noda and Fujikado, '87a). On the other hand, an anterograde WGA-HRP transport study has indicated that the P-cell axons arising from the oculomotor vermis terminate almost exclusively in an ellipsoidal region which protrudes caudally from the FN (Yamada and Noda, '871, a region which is designated as the fastigial oculomotor region (FOR). Microstimulation of the oculomotor vermis and the ventromedial aspect of the FOR results in saccades which differ in their horizontal directions, with vermal stimulation resulting in ipsilateral and fastigial stimulation resulting in contralateral saccades (Noda et al.,'88). When bicuculline was injected in the FOR, the saccades evoked by the vermal stimulation were suppressed for several hours (Noda et al., '88). Transient excitation of P cells during vermal stimulation causes an inhibition of tonic firing of fastigial neurons (Ohtsuka and Noda, '89). Furthermore, fastigial neurons bursting with saccades can be recorded only within the anatomical confines of the FOR (Ohtsuka and Noda, '90). These data are consistent with the concept that signals from the vermis are transmitted to the saccadic nuclei of the brainstem via the FOR. The control of

331

saccades by the cerebellum, therefore, depends on the signals conveyed by FOR neurons. The present investigation was initiated to understand the anatomical connections of the FOR with saccade-related structures in the brainstem. Previous degeneration studies have shown that the FN receives fibers from a large number of brainstem nuclei. The nuclei considered to send fibers to the FN include the IVN and MVN and several of the small subdivisions including groups f and x (Dow, '36; Brodal and Torvik, '57; Carpenter et al., '59; Carpenter, '60), the paramedian medullary reticular formation (Brodal and Torvik, '541, and parts of the PHN (Torvik and Brodal, '54; Carpenter et al., '59; Gonzalo-Ruiz et al., '88, HRP method). Fastigial efferent fibers have also been investigated extensively by degeneration (Rasmussen, '33; Rand, '54; Thomas et al., '56; Carpenter et al., '58; Cohen et al., '58; Carpenter, '59; Walberg and Pompeiano, '60; Walberg, '61; Walberg et al., '62a,b; Voogd, '64; Angaut and Bowscher, '70; Kievit and Kuypers, '72; Martin et al., '74) and autoradiographic methods (Batton et al., '77; Carpenter and Batton, '82; Sugimoto et al., '82; Asanuma et al., '831, and by anterograde or retrograde transport of HRP (Langer and Kaneko, '84; Kyuhou and Kawaguchi, '87; Gonzalo-Ruiz et al., '88). Although the general pattern of termination of fastigial fibers in the brainstem appears to be fairly well understood, data concerning the connections of the FOR, which designates the caudal portion of the FN, are still lacking. The degeneration studies agree on the widespread bilateral and asymmetrical distribution in the brainstem of the fibers

-

Abbreuiations

AIN B1

cc

cMRF CN cont contra DAO DLPN DMPN DMRF DN DSCP EPSPs F FEF FOR FN FR G HRP IC ICP INC Inj.

I0 ipsi.

IPSPS IVN JB L LL

Lt. LVN MA0 ML MLF MRF MVN N

anterior interposed nucleus bilateral crus cerebri central mesencephalic reticular formation cuneate nucleus contralateral contralateral dorsal accessory olive dorsolateral pontine nucleus dorsomedial pontine nucleus dorsomedial reticular formation dentate nucleus decussation of superior cerebellar peduncle excitatory postsynaptic potentials fastigial nucleus frontal eye field fastigial oculomotor region fastigial nucleus fasciculus retroflexus genu of facial nerve horseradish peroxidase inferior colliculus inferior cerebellar peduncle interstitial nucleus of Cajal injection site inferior olivary complex ipsilateral inhibitory postsynaptic potentials inferior vestibular nucleus juxtarestiform body lateral lateral lemniscus left lateral vestibular nucleus medial accessory olive medial lemniscus medial longitudinal fasciculus mesencephalic reticular formation medial vestibular nucleus total number of labeled neurons

N I11 n I11 N IV n N NVI nVI n.VII N XI1 NPS NRTP P cell PC PCN PHN PIN PL PN PPRF PR PY RF riMLF RN

Rt.

sc SCP so ST STT SVN TMB T SP TT UF

VL VN

VPL VPM WGA-HRP

oculomotor nucleus oculomotor nerve trocblear nucleus trocblear nerve abducent nucleus abducent nerve facial nerve hypoglossal nucleus nucleus parasolitarius nucleus reticularis tegmenti pontis Purkinje cell posterior commissure nucleus of posterior commissure nucleus prepositus hypoglossi posterior interposed nucleus pu1vin ar pontine nucleus paramedian pontine reticular formation pontine raphe pyramidal tract fasciculus retroflexus rostral interstitial nucleus of MLF red nucleus right superior colliculus superior cerebellar peduncle superior olivary nucleus solitary tract spinal trigeminal tract superior vestibular nucleus tetramethybenzidine tectospinal tract tectspinal tract uncinate fasciculus ventrolateral thalamic nucleus vestibular nucleus ventral posterolateral thalamic nucleus ventral posteromedial thalamic nucleus wheat germ agglutinin conjugated horseradish peroxidase

H. NODA ET AL.

332 arising from the FN. However, many fastigial fibers decussate in the cerebellum and take a variety of routes through the ipsilateral and contralateral FNs before leaving the cerebellum (Sugita et al., '88). Since even a perfect unilateral lesion of the FN may invariably interrupt the fastigital fibers which traverse parts of the nucleus, degeneration studies introduce variables which render interpretations of the findings difficult and sometimes unreliable. On the other hand, one problem with the tract-tracing methods is the diffusion of the tracer around the injection site. The spread of tracers becomes a critical factor, particularly when small experimental animals are used. In the present study we tried to place HRP at the caudal end of the FN. In this way, the diffusion of HRP involved the adjacent white matter, instead of the other fastigial portion. Even when the fibers running in the adjacent white matter are included in the HRP diffusion area, the target regions of the fibers descending in the vicinity of the caudal FN are well understood. It is generally accepted that the P-cell axons do not project to any region of the brainstem beyond the vestibular complex, with a possible exception of the PHN. Therefore, the labeled terminals which appear in remote locations are not caused by the uptake from the P-cell axons. On the other hand, the large brain of the macaque monkey substantially narrowed the relative size of the HRP injection in the FN as compared to injections in small animals. These advantages enabled us to study the topographical organization between the FOR and various portions of saccade-related nuclei in the brainstem. In order to identify the FOR and to evaluate the size of effective injection site, we referred to the maps of topographical organization for both the P-cell and olivocerebellar projections. In addition to the well-known corticonuclear projections, a precise topographical organization has been demonstrated from the caudal third of the MA0 to various parts of the caudal FN in the macaque monkey (Ikeda et al., '89). It was possible to assess the extent of the effective site from the distribution of retrogradely labeled P cells in the overlying cerebellar cortex and that of retrogradely labeled neurons in the inferior olive (10).The details of the olivocerebellar and cerebello-olivary connections studied in the same group of monkeys and two additional monkeys have been described elsewhere (Ikeda et al., '89).

MAflERIALSANDMETHoDS Distributions of retrogradely labeled neurons and anterogradely labeled axons were investigated in the cerebellar cortex and the brainstem after WGA-HRP injections. These studies were conducted in eight adolescent pig-tailed monkeys (Macaca nemestrina), 4-5 kg body weight. We also used the brains of three other monkeys, cut in either parasagittal, frontal, or horizontal planes and stained with cresylviolet, as a reference for the identification of brainstem nuclei and regions in unstained HRP sections. In order to compare afferent fibers to the FOR with those to the oculomotor vermis, the data from our previous experiment were also used. The data, including the description of injection site, have been published (Yamada and Noda, '87). The methods of WGA-HRP injection and the volume of the enzyme injected in each monkey, the procedures of histological processing, and the analysis of data have been described in detail (Ikeda et al., '89). In brief, a mixture containing 10%WGA-HRP and 10%HRP in 0.05 pl Ringer solution (0.1pl in monkey Hy and 0.5 p1 in monkey Pn) was

injected into the FOR. After 72 hours, each monkey was deeply anesthetized (using the maximum evaporation rate of methoxyflurane) and perfused with warm (36°C) Ringer solution followed first by 4 liters of 1% paraformaldehyde and 1.25%glutaraldehyde in a 0.1 M phosphate buffer (pH 7.4) and then by 4 liters of 10%sucrose (pH 7.4) at 4°C. The brain was sliced into 80 pm parasagittal sections (100 pm coronal sections in monkey Hy) on a freezing microtome. The sections were treated with tetramethylbenzidine (TMB) following the method described by Mesulam ('82),and some sections were counterstained with neutral red. The location of cell bodies and axons labeled with HRP was carefully mapped onto large drawings with the aid of a drawing tube at magnifications of 1 0 0 - 2 5 0 ~ . The boundary of each brainstem nucleus was carefully marked on the cameralucida drawings in reference to the photomicrographs from Nissl-stained parasagittal sections of the brains of other monkeys.

RESULTS Injection sites and distributionsof retrogradely labeled P cells and olivary neurons In four of the eight monkeys, the center of HRP injection was successfully placed in the caudal FN (monkeys Tk, Hy, Pn, Op). The center was found in the white matter between the FN and the posterior interposed nucleus (PIN) in two monkeys (monkeys Tn, Ty), while it was placed in the PIN in the remaining two monkeys (monkeys Gt, Om). The location and extent of effective site for each monkey were determined by the distribution of retrogradely labeled P cells in the overlying vermis and that of labeled cells in the 10. These data have been described in detail (Ikeda et al., '89). In brief, the effective site of the HRP injection (0.05 p1) was confined to the FOR in monkey Tk. Labeled P cells were found only in the oculomotor vermis of the ipsilateral side and in group b of the MA0 of the contralateral side. In monkey Hy, the HRP injection (0.1 p1) was placed in the FOR, but it encroached anteriorly upon the dorsocaudal half of the FN. Additional labeling appeared in a narrow longitudinal zone within 0.6 mm of the midline in vermal lobules 111-VI. Cells in group b of the caudal MA0 were labeled. Some labeled cells appeared also in groups a and c and in a caudal part of the dorsal accessory olive (DAO). A large mount of HRP (0.5 pl) was successfully injected into the FOR, but the effective site involved the adjacent areas in monkey Pn. It did not spread beyond the midline. In addition to the oculomotor vermis, labeling appeared in a lateral part of lobule VIII and in a narrow longitudinal zone of lobule IX of the vermis. A few labeled P cells were also found in the midline lobules V-IX, in the medial portion of crus I1 and paramedian lobule. In the 10, labeled cells were found in the central MA0 and a wide area of the DAO, in addition to the main labeling in the caudal MAO. In monkey Tn (0.05 pl), the HRP injection included a region anterior to the FOR. Labeled P cells appeared in lobule VI and labeled olivary neurons appeared in group a. In monkey Op, HRP (0.05 p1) was injected in the caudal FN but the center of effective site was ventral to the FOR. Labeled P cells were found in vermal lobules VIII and IX. Labeled olivary neurons appeared in caudal parts of groups a and b, in addition to group c. In monkey Gt, HRP (0.05 p1) was placed in the medial portion of the PIN. The P cells were labeled mainly in the paramedian lobule and in a lateral portion of vermal lobules IX and X. Some labeled P cells

CONNECTIONS OF FASTIGIAL OCULOMOTOR REGION were also found in a narrow longitudinal zone that extended throughout the anterior lobe. Labeled olivary neurons were found in the rostral two-thirds of the MA0 and the DAO. Only a part of group a was labeled in the caudal MAO. In monkey Om, the HRP injection (0.05 +l) wasplaced in the lateral portion of the PIN. The distribution of labeled P cells were consistent with that of monkey Gt, and there was no sign of HRP spread into the FN. The data from these four monkeys-in which HRP was injected in the adjacent structures, but the effective site did not include the FOR-were used as controls.

Brainstem afferent fibers to the fastigid oculomotor region Retrogradely labeled P cells appeared only in the cerebellar cortex of the injection side, while neurons in the I 0 were labeled only on the contralateral side. In addition, as will be described later, anterogradely labeled terminals appeared only in the contralateral brainstem, except for the vestibular complex and the DMRF that receive bilateral projections from the FOR. It is most probable, therefore, that the HRP injections made in the present experiment were strictly confined to the ipsilateral side. It is important to note that retrogradely labeled cells appeared in the brainstem bilaterally in all monkeys, despite the unilateral injections of HRP. Projections fkom the vestibular complex and PHN. Following HRP injections in the FOR, retrogradely labeled neurons appeared bilaterally in the SVN; the MVN; the IVN, including the minor cell groups f, x, z; and the PHN (Figs. 1, 2). The labeled neurons aggregated in the caudal half in both the MVN and IVN. The LVN, including y group, and the ventrolateral VN were free of labeled cells. Thus, the vestibulofastigial connections were quite different from those found in the other regions of the brainstem. In the I 0 or the NRTP, for example, the location of labeled terminals was always found in a part of the region where retrogradely labeled neurons aggregated, following the HRP injections into the FOR. However, the FOR projected only to the LVN and IVN of both sides. The MVN; the SVN, including y group; the ventrolateral VN; or the PHN did not receive FOR fibers. The distribution of retrogradely labeled neurons in the vestibular complex did not differ significantly even when the effective site included a region anterior to the FOR (monkeys Hy, Pn, Tn) or a region ventral to the FOR (monkey Op), although the density of labeled cells was higher in monkeys P n and Hy. Therefore, topographical organization between the E" and the vestibular complex was less impressive than that for the olivofastigial projection. In Table 1, the numbers of retrogradely labeled neurons found in different subnuclei of the vestibular complex and the PHN were compared among three cases of FOR injection and a case of oculomotor vermis injection. Confirming the above notion, the numbers did not differ significantly among the three FOR-injection cases. As far as the vestibular complex is concerned, the numbers were surprisingly similar to those found in monkey Jo, in which HRP was injected in the oculomotor vermis but was not limited to one side of the midline (Yamada and Noda, '87). Labeled neurons were found in the LVN in monkey Jo, but the same nucleus was free of labeling in the monkeys with FOR injections. The parasagittal zone B projects to the LVN (Voogd and Bigark, '80). The labeling of neurons in the LVN in monkey J o might have resulted from the spread of HRP into a part of the B-zone, but it did not occur in the

333

FOR-injection cases. A significant difference is seen in the numbers of labeled cells found in the PHN between the FOR- and vermal-injection cases. Projections from the NRTP. The HRP injections into the FOR caused retrograde labeling of neurons in the bilateral NRTP. The distribution of labeled cells was nearly symmetrical and quantitatively similar on each side. When they were plotted in a three-dimensional map of the nucleus, the neurons aggregated in the medial third, the caudal K,and the dorsal half of the NRTP. The pattern of distribution of labeled cells was similar even when the effective site involved a fastigial area rostral to the FOR (monkeys Hy, Tn). In monkey Pn, however, the labeling showed a different pattern. As labeled neurons appeared also in the central portion of the NRTP, the dorsoventral gradient of labeling, which was commonly observed in all cases of FOR injection, became less impressive. The effective site in monkey P n involved a part of the PIN, as evidenced by the labeling of P cells in the medial portion of c m s I1 and paramedial lobule. The spread of HRP into the PIN caused the labeling of the NRTP cells in the central portion. In monkey Gt, for example, labeled cells were found in the central portion and caudal % of the NRTP (Fig. 3B-D) .

Figure 4A shows the distribution of retrogradely labeled neurons found in different parasagittal planes of monkey Tk. There were more labeled cells on the injection side within 1 mm of the midline, but the total number on each side was comparable. Comparing the total numbers of the labeled NRTP neurons found in monkeys Tk and Jo (Fig. 4B), one can see only a fraction of the NRTP neurons, which projects to the oculomotor vermis, sends axons to the FOR. Immediate comparison of these figures is not possible, because the injection in monkey J o was bilateral but was confined to lobule VII, while that in monkey Tk was strictly unilateral but included a fastigial area which received P-cell projection from lobule VIa. Nonetheless, it may be possible to estimate that less than 3%of the NRTP neurons projecting to the oculomotor vermis sent axons to the FOR. Projections from thepontine nuclei. Following HRP injection into the FOR in monkey Tk, labeled neurons appeared in bilateral PNs, but they were more numerous on the contralateral side. On both sides, the labeled neurons were widely scattered but tended to aggregate in a dorsal portion of the DMPN (Figs. 1B-D, 2B,C) and in the DLPN (Figs. lF, 2B,C). Labeled neurons were distributed in the dorsal portion of the peduncular nucleus at the same pontine levels. Labeling of pontine neurons in monkey Hy was less numerous than others for unknown reasons (Fig. 2). However, the tendency of distributions in all monkeys of fastigial and PIN injections was not significantly different. The labeled neurons in the PNs had small cell bodies, which characterized the majority of pontine cells in the Nisslstained sections. Figure 4C shows the mediolateral distribution of labeled pontine neurons for monkey Tk. The numbers in the DMPN and peduncular nucleus on the contralateral side were larger as compared to the ipsilateral side, but those in the DLPN were comparable. A possibility of involving the PIN in the effective site can be denied because no anterogradely labeled axons were found in the ipsilateral SCP in monkey Tk. The total number of labeled neurons following injection into the FOR was 785, while that for vermal lobule VII was 7,731. When one makes a rough estimate, there-

-

4 L

CONNECTIONS OF FASTIGIAL OCULOMOTOR REGION fore, approximately 10% of pontine cells projecting to the oculomotor vermis send axons to the FOR. The percentage may be higher if one considers only the contralateral peduncular nuclei. Projection from the pontine raphe. Labeled cells appeared in this midline structure mainly in the caudal third of the area between the oculomotor and abducens nuclei. The boundaries between NRTP and PR and between PR and PPRF were obscure, but we identified the PR by the ventrodorsal direction of fibers in parasagittal sections. Various sizes of labeled cells were distributed throughout the PR which corresponds to the location of omnipause neurons. Projections from the PPRF. Relatively large multipolar cells were labeled in the PPRF, in the caudal third of the area between the oculomotor and abducens nuclei. The labeled cells were found bilaterally, but the percentage of labeling was very small and was estimated to be less than 10%of the labeling found in the same region following HRP injections in the oculomotor vermis (Yamada and Noda, '87). Large cells outnumbered small labeled cells, and both groups of cells were intermingled throughout the extent of PPRF. At the rostral level of the PHN, small labeled cells were interspersed among the axon bundles of the MLF, in the interfascicular nucleus. There appeared t o be no projections from this region to the PIN, because labeled cells were not seen in monkeys Gt (Fig. 3) and Om. Projections from the midbrain. The FOR seemed to receive a modest projection from the contralateral midbrain. Ipsilaterally, however, less than a dozen neurons were retrogradely labeled in the entire midbrain in each monkey, suggesting that the ipsilateral projection from this region is limited. Faintly labeled small neurons scattered in the ventral PGA on the contralateral side. The labeling in this region was consistent in all FOR-injected monkeys. Labeled neurons were found also in the MRF, lateral aspect of the interstitial nucleus of Cajal (INC), and in the deep layers of the SC and inferior colliculus (IC) of monkey Hy (Fig. 2A,B) and monkey Pn. Although labeled neurons were scattered throughout the MRF, the locations of labeled neurons were inconsistent among different monkeys. Therefore, except for the projection from the ventral PAG, it was difficult to conclude that any particular mesencephalic region projects to the FN.

FOR projectionstobrainstemnuclei Intracerebellarpathways ofFOR fibers. When HRP was injected in the FOR, anterogradely labeled axons were found in both the contralateral uncinate fasciculus (UF) and the ipsilateral juxtarestiform body (JB). This finding was consistent in all four FOR-injected monkeys. The locations of the labeled axons within the cerebellum were also consistent in these monkeys, except for monkey Pn, in

Fig. 1. Camera-lucida drawings of parasagittal sections of the brainstem of monkey Tk, showing locations of retrogradely labeled cells following HRP injection into the FOR of the right FN. The order of the drawings is arranged in such a way that the changes in the locations of labeled cells can be easily seen by following the bold arrows between the figures. The open arrow (B + A) indicates the shift of parasagittal planes laterally from the injection site (O), whereas filled arrows indicate the shifts of the plane laterally on the contralateral side. The side and distance from the midline are shown at the right lower corner of each drawing. Scale bar shown in F applies also to A-E.

335

which some fibers in the ipsilateral brachium conjunctivum (superior cerebellar peduncle, SCP) were also labeled. The labeling of the SCP axons in monkey P n resulted from the spread of HRP into the PIN. The effective site in monkey P n involved a part of the PIN, as indicated by the retrograde labeling of the P cells in the medial portion of cms I1 and paramedian lobule. It is generally agreed that only fibers from the FN decussate within the cerebellum and those of the anterior interposed nucleus (AIN), the PIN, and the dentate nucleus (DN) do not. Instead, the fibers from these nuclei leave the cerebellum with the ipsilateral SCP. This notion was confirmed in monkeys Om and Gt in which labeled fibers were found only in the ipsilateral SCP and both the UF and J B were free of labeled axons. Projections to the vestibular complex. Following HRP injections in the FOR, anterogradely labeled terminals were found bilaterally in the vestibular complex. The labeled axons which descended in the ipsilateral J B entered the LVN and terminated in the ventral % of the nucleus. Some fibers passed through the LVN and terminated in the rostroventral part of the IVN adjacent to the ventrolateral VN. However, no terminals were found in the MVN, the SVN, the ventrolateral VN, or the PHN of the injection side (Fig. 5B-D). Many fastigial fibers which entered the brainstem via the contralateral UF terminated in the ventral part of the LVN (Fig. 5B). A few terminals were found also in the contralatera1 IVN in the portion corresponding to the small cell groups f, x, and z. The MVN, the ventrolateral VN, and the SVN, including y group, were also free of terminals on the contralateral side (Fig. 5). The pattern of labeling in the vestibular nuclei in monkey Hy was almost identical with that in monkey Tk, except that the labeling was stronger. In monkey Op, in addition to the vestibular labeling seen in monkey Tk, a small number of labeled terminals was found also in a rostral portion of the ipsilateral SVN and a ventrolateral portion of the h" of both sides. In monkey Pn, labeled terminals were more or less in all vestibular subnuclei, except for y group. The labeling in the SVN and h", which contain most of the eye-movement related neurons of the VNs, appeared only when the HRP injection involved a fastigial area rostral to the FOR (monkeys Op and Pn). It seemed most likely that the FOR neurons do not project to the vestibular region for the vestibulo-ocular reflex. Projections to the PHN. The efferent fibers of the FOR do not terminate also in the PHN. The PHN was free of labeled terminals in monkey Tk (Fig. 7G) and Op. However, if the effective site included a fastigial area rostral to the FOR (monkeys Hy, Pn), labeled terminals appeared at the ventral border of the caudal third of PHN (Fig. 5D) and at the rostral end of the PHN, immediately caudal to the abducens nucleus. The PHN was free of labeled terminals also in monkeys Om and Gt. Projections to mesodiencephalic junction and midbrain tegmentum. In contrast to the bilateral projections to the vestibular complex, the FOR projection to the midbrain tegmentum and the thalamus was strictly contralateral. The terminal labeling was derived from the decussated FOR fibers which run medial and dorsal to the contralateral SCP. In the midbrain, these fibers did not decussate again but remained in a region dorsolateral to the MLF. Labeled terminals scattered widely from the caudal end of the trochlear nucleus to the rostral end of the riMLF. In

H. NODA ET AL.

336

F

5mm

HY Fig. 2. Camera-lucida drawings of coronal sections of the brainstem of monkey Hy, showing the locations of retrogradely labeled cells following HRP injection into the FOR of the right FN in frontal sections. The center of injection site (0) is seen in E. Scale bar shown in F applies also to A-E.

CONNECTIONS OF FASTIGIAL OCULOMOTOR REGION TABLE 1. Number of Retrogradely Labeled Cells in the Vestibular Complex and Perihypglossal Nuclei Following HRP Injeetions Into the Caudal FN and Vernal Lobule VII Monkey Tk

Monkey Op

Ipsi.

Contra.

Ipsi.

Contra.

5 0 56 52 0

17 0 79 51 0

12 0

YgrOUP

8 0 68 50 0

PHN

15

11

18

Nucleus

Monkey Hy

Imi. ~

SVN

LVN MVN

m

78 66 0 19

Contra.

Monkey Jo'

Right

Left

12 0 101 62

60 32 108

0

51 298

76 23 56 117 0

~

13 0 83 63 0 17

20

96

258

'HRP was i n j d into vernal lobule VII The enzyme encroached upon the contralateral CerebeUer cortex (FromYanlada andNoda, '871

medial parasagittal sections, the labeling of terminals was found in the ventral part of the PAG and a part of the dorsal tegmental nucleus. However, both the oculomotor and trochlear nuclei were free of labeled terminals. The major labeling appeared in sections 1-2 mm lateral to the midline. Rostrally, the terminal labeling extended beyond the fasciculus retroflexus (FR) into an area where fibers of the MLF were no longer visible. In the parasagittal sections of the medial INC, dense aggregates of labeled terminals were found in a region rostrodorsal to the INC (Fig. 6B,C). Correlating with Nissl-stained sections, the rostral half of this labeling was found to correspond to the riMLF, while the caudal half was most probably the lateral part of the nucleus of Darkschewitsch. Labeled terminals were distributed also in a medial portion of the Forel's H Field. The INC and the red nucleus were, however, free of labeled terminals (Fig. 6B,D). A less dense but wide distribution of labeled terminals was found in a region lateral to the INC. The labeled terminals scattered among numerous labeled axons of fastigial origin (Fig. 6D). Many of these fibers terminated in the MRF but some ascended further to the thalamus. A large part of the terminal labeling corresponded to the region known as the central MRF which was recently identified physiologically by virtue of yielding horizontal saccades when electrically stimulated (Cohen et al., '82). A considerable number of labeled terminals was also found in the nuclear complex of the posterior commissure (Figs. 5D, 6C). Projections to the pontine reticular formation. In the paramedian pons, labeled terminals were found in clusters only in the ventrocaudal part of the PR, in a region rostromedial to the abducens nucleus (Fig. 7B,C). The labeling coincided with the location of omnipause neuron. In more lateral pons, the distribution of the terminal labeling extended more rostrally and the density of labeling increased. Labeled terminals aggregated in the dorsal half of the pontine RF, but the regions near the NRTP and PN were free of labeled terminals (Fig. 7E,F). The entire labeling lateral to the PR coincided with the PPRF. Consistent with the projections to the MRF, the FOR projection to the pontine RF was also contralateral. The distributions of labeled terminals in the pontine RF of monkey Hy (Fig. 5) and monkey P n (not shown) were similar to that of monkey Tk. The projection was strictly contralateral but the rostrocaudal extent of the distribution in the PPRF was somewhat larger in monkeys Hy and P n than that in monkey Tk. In monkeys Om and Gt, in which the HRP injections were confined to the PIN, no labeling of terminals was found in the PPRF, although there was an ample projection to the contralateral NRTP. Projections to the medullary RF. The strongest projection from the FOR was found in the medial part of the

337

medullary RF (Fig. 7G-I), particularly in a region caudoventral to the abducens nucleus. Although the number of terminals was reduced, the labeling continued to cover the dorsal half of the medullary RF down to the level of the PHN. The labeling was found only in the medial half (within 2 mm of the midline) and the lateral half was free of labeled terminals. Since the labeling was found only in the medial and dorsal half of the medullary RF, we used the functional terminology the dorsomedial RF (DMRF). Although the primary FOR projection to the DMRF was contralateral by the fibers emerging from the UF, there were some terminals on the ipsilateral side that arose from the ipsilateral JB. The ipsilateral projection was abundant, particularly in monkey Hy (Fig. 5D) and monkey Pn. These terminals were supplied by fibers which emerged via the ipsilateral JB, passed through the LVN and IVN, and entered this region. In monkeys Om and Gt, the DMRF was free of labeled terminals. Projections to the NRTP. The NRTP receives fibers from the contralateral FN. It receives fibers also from the PIN but the distribution of labeled terminals was markedly different from that of the FOR projection. When HRP was injected into the FOR, labeled terminals were found in the medial and dorsal portions of the contralateral NRTP (Figs. 5A, 7C,D), while the central portion of the NRTP was free of labeling. Similar patterns of terminal labeling were found also in monkeys Op and Pn. Monkey P n had additional labeling in the processus tegmentosus lateralis. The axons which terminated in the NRTP left the cerebellum with the contralateral SCP, traveled along its medial border, turned ventrally in the pons, and entered the NRTP after passing through the region of the PPRF. In contrast, the course of fibers and the topography of terminal distribution of the PIN projection to the NRTP were quite different from those of the FOR projection. Instead of decussating within the cerebellum, the PIN fibers left the cerebellum with the ipsilateral SCP and crossed the midline in the decussation of the SCP (DSCP). After crossing the midline in the midbrain, the fibers descended directly into the NRTP without passing through the region of the PPRF. They terminated in the central and lateral portions of the NRTP. The ipsilateral NRTP and the medial and dorsal portions of the contralateral NRTP were free of terminals (not illustrated). Projections to the pontine nucleus (PN). Fibers from the FOR terminate in the DMPN of the opposite side of injection. The labeling in the DMPN was found in all four monkeys of FOR injection. On the other hand, the labeled terminals in the DLPN were found only in monkeys Hy (Fig. 5A) and Pn, indicating that the target of the FOR fibers is primarily the DMRF and that the DLPN is the target of the FN neurons located rostral to the FOR. In monkey Om, labeled terminals were found only in the DLPN of the contralateral side. Projections to the nucleus parasolitarius (NPS). Labeled terminals were found in the contralateral NPS in all four monkeys in which the effective site involved the FOR. The fibers terminating in this nucleus passed through the caudoventral portion of the IVN, turned medially, and reached the NPS. However, there was no projection from the PIN, as indicated by the finding that the NPS was free of labeled terminals in monkeys Om and Gt. Projections to the inferior olivary complex (IO). The details of the fastigio-olivary projections have been reported in a preceding paper (Ikeda et al., '89). In brief, the fibers from the FOR terminate in the Z-portion (an area of

H. NODA ET AL.

338

4

CONNECTIONS OF FASTIGIAL OCULOMOTOR REGION group b) of the contralateral MAO. When the effective site included a fastigial area rostral to the FOR, labeled terminals appeared in group a. When it included an area ventral to the FOR, the labeling appeared in caudal parts of groups a and b, in addition to group c of the caudal MAO. HRP injection into the PIN resulted in labeling of caudal part of the central MA0 and the medial part of the central portion of the DAO. Projections to the superior colliculus (SC). Labeled terminals were found in the intermediate layer of the contralateral SC in all cases of FOR injection. Unlike the terminals commonly observed in the brainstem, those found in the SC formed finger-like boutons. These terminals were derived from axons which ascended medial to the contralateral SCP. There was no quantitative difference in the labeling between monkey Tk (HRP was confined to the FOR) and monkey Pn (HRP injection involved caudal % of the FN), indicating that those axons arose from a relatively circumscribed region in the caudal FN. No labeled terminals were found in the PIN-injection cases. Projections to the thalamus. The terminal distributions showed markedly different patterns between the cases of FOR- and PIN-injection. When the effective site included the FOR, labeled terminals appeared in the contralateral ventral posterolateral nucleus (VPL). The terminals were confined to the dorsolateral portion of the nucleus in monkey Tk, while they were found also in the medial and ventral portions in monkeys Hy and Pn. On the other hand, when HRP was injected into the PIN, labeled terminals were found not only in the VPL but also in the ventral posteromedial nucleus (VPM) and ventrolateral nucleus (VL).

DISCUSSION Brainstemprojectionsto the fastigial oculomotor region In conjunction with the findings from a previous retrograde-labeling study on monkeys in which HRP was injected into vermal lobule VII (Yamada and Noda, '871, the present study has provided new information on the anatomical basis for cerebellar oculomotor control. An important finding was that the distributions of retrogradely labeled cells in the brainstem that coincided remarkably well with the results obtained after the HRP injection into lobule VII (Yamada and Noda, '87). The topographical organization found between the cerebellar vermis and the brainstem regions has been confirmed also in the brainstem projections to the FOR. The data have indicated that a vermal lobule and a fastigial region, which are topographically connected by the P-cell axons, receive their afferent fibers from the same brainstem area. Thus, the present study supports the well-accepted hypothesis that the brainstem inputs to the FN may arise as collaterals of mossy fiber afferents (for review Blanks, '88). The present study has shown that the distribution of retrogradely labeled cells coincided with the location of

Fig. 3. Distribution of retrogradely labeled cells found in the brainstem of monkey Gt following HRP injection into the right PIN. The center of injection site (0) is seen in A. A, B: Parasagittal planes of the ipsilateral side. C-F: Parasagittal planes of the contralateral side. Scale bar shown in F applies also to A-E.

339

anterogradely labeled terminals discovered after the HRP injection into the FOR. The distributions of the retrograde and the anterograde labeling were, however, not necessarily the same, because of the difference in sensitivity to the tracer depending on which portion of the neuron was inside the effective site. Nonetheless, it was consistently found that the distribution of labeled terminals was always smaller than that of labeled neurons and that the labeled terminals appeared in the region where the densest population of retrogradely labeled neurons was found. However, an exception to the reciprocal connections was found in the fastigiovestibular connections. The FOR received afferent fibers from the MVN, the SVN, and the IVN, but it projected to the LVN and the IVN. In the retrograde HRP study, Carpenter and Batton ('82) found that the principal afferents to the FN arose bilaterally from the dorsal and caudal parts of the MVN and IVN. In monkeys with HRP injections confined to the FN, no neurons in the LVN, the SVN, and y group were retrogradely labeled on either side (Carpenter and Batton, '82), although the projection from the SVN to the FN has been described in the cat (Kotchabhakdi and Walberg, '78). Although the sources of afferent fibers described here agree with many of the previous observations, there were certain quantitative differences. While most of the previous studies considered the inputs to the entire FN, a portion of the FN, identified only recently in macaques, has been focused in the present study. The quantitative difference also reflects the functional characteristics of the FOR as opposed to the entire FN. It is generally agreed that the vestibular complex is the principal source of the input to the FN. According to Carpenter and Batton ('82), the largest number of fastigial afferent fibers arise from cells in the caudal part of the MVN. In the present study, the largest number of labeled cells was found in the PNs, while the second largest number was the labeling in the NRTP. In contrast, the fibers of vestibular origin shared only a small fraction of the afferent fibers terminating in the FOR. In monkey Tk, for example, labeled neurons in the VNs were only 12.6% of the total labeled neurons of the brainstem. The paucity of the vestibular projections also characterizes the mossy-fiber inputs to vermal lobule VII (Yamada and Noda, '87). The present study has shown that the FOR receives afferent fibers from the brainstem structures where neurons discharge with saccades. A considerable number of fibers were shown to arise from the NRTP. The same region of the NRTP in monkeys contains neurons that discharge during saccades (Crandall and Keller, '85).They were directionally selective and discharged long before (20-30 msec) the onset of a saccade. The mossy fibers arising from the NRTP region, therefore, may transmit saccade-related signals to the FOR. Large retrogradely labeled neurons were widely scattered in the PPRF which is considered as the final common pathway for all rapid eye movements (Bender and Shanzer, '64; Luschei and Fuchs, '72; Keller, '74; Henn and Cohen, '76). It is generally accepted that the so-called medium-lead burst neurons of the PPRF provide the input which drives oculomotor neurons during saccades. These units show directional selectivity primarily in the horizontal direction. Pontine raphe neurons in the rostral aspect of the abducens nucleus were also labeled. This area coincides with the location of omnipause neurons (Keller, '74; Raybourn and Keller, '77; Nakao et al., '80; Langer and Kaneko, '83, '84). Labeled neurons were found

H. NODA ET AL.

340

[1 A

NRTP Inj . i t Fastigial Oculornotor Region

N=336

60

Monkey Tk

Pontine Nuclei In] L t Fastigial Oculomotor Region

N = 785

50

I n

6o

Monkey Tk 150

40

30

20 10

0

Proc Tegment L o t

NRTP inj.: B i Lobule V l i of Verrnis

rli

DLPN

Peduncular N

DMPN Peduncular N

DLPN

1

500

Pontine Nuclei Inj El Lobule VII o f Verrnis

Monkey Jo

-/ 300

t t..

n

Monkey J o

n

i

300

n

I200

I00

0

Fig. 4. Comparison of the retrogradely labeled cells in the NRTP and the pontine nuclei following HRP injections into the FOR (monkey Tk) and into vermal lobule VII (monkey Jo). Each column represents the number of cells for each nucleus counted from three consecutive TMB-reacted,80-pm thick sections. Filled columns, large cells. Hatched

columns, intermediate cells. Open columns, small cells. The data for the vermal injection were taken and rearranged by combining the data from the histograms shown in Figures 10 and 11of a previous study (Yamada and Noda, '87).

also in the dorsal and medial portion of the medullary reticular formation. Neurons in this portion are related only to ipsilateral fast eye movements and exert an inhibitory influence on contralateral abducens motoneurons (Hikosaka et al., '77, '78; Yoshida et al., '82; Strassman et al., '86).

DMRF, and the riMLF (including the medial portion of Forel's H Field). PPRF. Anterogradely labeled terminals were found throughout the dorsal part of the nucleus reticularis pontis oralis and caudalis, the PPRF, in a functional term. The involvement of the PPRF in the generation of saccades correlated well with the results of clinical observations (Christoff, '741, stimulation (Bender and Shanzer, '64) and lesion experiments (Goebel et al., '71; Cohen and Komatsuzaki, '72). The cells in the PPRF are active immediately before a saccade. The size and the direction of the subsequent saccade could be exactly predicted from the burst of activity (for review Fuchs et al., '85). The earlier observations on the fastigial projection to the PPRF (Walberg et al., '62a,b; Batton et al., '77; Chan-Palay, '77) are now supported by autoradiographic studies in the cat (Graybiel, '77a,b) and monkey (Asanuma et al., '831, and a retrograde HRP study in New World monkey (Gonzalo-Ruiz et al., '88). The present study has confirmed this projection by demonstrating that the fibers arising in the FOR terminate in the PPRF. RiMLF. The heavy labeling of terminals found in the riMLF, rostra1 to the FR, coincided with the anatomical confines of the riMLF. The labeled terminals were found also in a medial portion of Forel's H Field. This region was outlined in the cat by Graybiel('77b) and in the monkey by Biittner-Ennever ('77) and is considered to contain the premotor neurons which directly activate the vertical ex-

Brainstem projectionsfrom the fastigial d o m o t o r region The present study has shown that neurons in the FOR send their axons to specific regions in the brainstem that are related to the oculomotor function. The characteristics of the projection can be summarized as follows: 1) The FOR fibers terminate primarily in the regions of the medial brainstem RF that have direct projections to the extraocular motor nuclei. 2) They project to the regions containing neurons which play a well-defined role in the generation of saccades, as demonstrated either by the presence of saccaderelated neuronal activity or by virtue of yielding saccades when electrically stimulated. 3) They project back to the brainstem regions which are known to provide the cerebellum with saccade-related input signals. 4) On the other hand, they send virtually no terminals to the brainstem regions known to be involved in the maintenance of eye position. First, three regions of the medial brainstem reticular formation are known to have direct projections to the extraocular motor nuclei: these regions are the PPRF, the

CONNECTIONS OF FASTIGIAL OCULOMOTOR REGION

341

D

5 rnm

Fig. 5. Distribution of anterogradely labeled axons from the FOR and their terminals in the brainstem of monkey Hy, illustrating the pathways and terminals of the fastigial fibers in the coronal planes. Fine dotted lines indicate anterogradely labeled axons from the FN and

larger dots indicate locations of labeled terminals. Each dot represent approximately 20 labeled terminals found in large camera-lucida drawings. Scale bar shown in D applies also to A-C.

traocular motoneurons during saccades. Lesions involving this region are associated primarily with downward gaze paralysis (Pierrot-Deseillingny et al.,'82). Direct connections of the riMLF with the vertical oculomotoneurons are suggested by the presence of monosynaptic EPSPs and

IPSPs evoked in trochlear motoneurons by stimulation of the riMLF (Nakao and Shiraishi, '85). Nakao and Shiraishi ('83, '85) have shown further that stimulation of this region evokes EPSPs in the motoneurons directly innervating the inferior rectus muscle. M e r e n t s to the riMLF have been

H. NODA ET AL.

342

C

\

I

,, /

-. .

.:

,/'

. ..' ....., ,, .,.'

..., ... .. _ _ _ ..,.: .$'

, r _ - ~ - -

.i.

,,'

pc

%'

,,.:..> .I.

'%'

\ '

I '

\

i

\, . ...

5..