ffeuroscience Vol. 39, No. 3, pp. 561-577, 1990 Printed in Great Britain

0306-4522/90$3.00+ 0.00 Pergamon Press plc 0 1990IBRO

CONVERGENCE OF CORTICAL AND CEREBELLAR PROJECTIONS ON SINGLE BASILAR PONTINE NEURONS: A LIGHT AND ELECTRON MICROSCOPIC STUDY IN THE RAT H. S. LEE and G. A. MIHAILOFF* Department of Anatomy, University of Mississippi, Medical Center, 2500 N. State St, Jackson, MS 39216, U.S.A. Abstract-A protocol that involved a combination of two orthogradely transported tracer substances, wheat agglutinin-horseradish peroxidase and Phaseolus vulgaris leucoagglutinin injected at separate locations in the same animal was utilized to investigate the possible congruence of axonal projection fields formed by the cerebral cortical and cerebellar aEerents to the basilar pontine nuclei. When large placements of tracer material were made in the cerebellar nuclei to label the cerebellopontine projections and a second tracer was injected in one of several cerebral cortical areas to visualize certain corticopontine projections, it was noted that axon terminal zones of the cortical and cerebeller systems occupied greater or lesser amounts of the same basilar pontine territory de~nding on the location of the cerebral cortical injection. Cere~llopontine terminal fields exhibited their greatest congruency with projections from the motor cortex containing the representation for facial musculature and with projections from the forelimb sensorimotor cortex. A lesser degree of overlap was observed when cerebellar projection zones were visualized in combination with basilar pontine projections from sensory face cortex, hindlimb sensorimotor cortex, visual cortex and auditory cortex. In addition, it was apparent that portions of the cerebellopontine and corticopontine terminal fields did not overlap at all. A related series of electron microscopic experiments was undertaken to establish that within the zones of overlapping cerebellar and cortical projections, there was in fact a convergence of the two afferent systems on single basilar pontine neurons. Boutons of the corticopontine system were labeled by the orthograde transport of wheat germ agglutinin-horseradish peroxidase injected into the sensorimotor cortex while cerebellopontine terminals were marked for electron microscopic identification in the same animal by transecting the brachium conjunctivum and allowing sufficient time for boutons in the pontine nuclei to exhibit degeneration. Although the number of definitive examples of convergence was small, nonetheless it was possible to observe single basilar pontine neuron dendrites receiving synaptic contacts from both the cortical and cerebellar afferents systems. Taken together these obervations indicate that some basilar pontine neurons receive a dual or convergent input from the cerebral cortex and cerebellar nuclei. It is difficult to estimate the prevalence of such convergence since cortical and cerebellar inputs typically contact distal and proximal pontine neuron dendrites, respectively, thus limiting the chances that both types of boutons can be observed in contact with a single basilar pontine neuron dendrite. Since the basilar pons contains both projection and local circuit neurons, and it has been shown that the cerebellopontine system includes excitatory and inhibitory components, it must remain for future electrophysiological studies to clarify the functional significance of this convergent circuitry.

The basilar pontine nuclei (BPN) have been considered as one of the important precerebellar nuclei involved in the prepro~amming and execution of volitional movement. It receives a major input from the cerebral cortex,8*9*2’*27~M as well as an unusually wide variety of ascending and descending subcortical projections arising throughout the central nervous system.‘*2g This suggests that at least some basilar pontine neurons might integrate these various inputs before transmitting a signal to the cerebellum. Such --_____ *To whom correspondence should be addressed. Abbreoiations: BPN, basilar pontine nuclei; DCN, deep cerebellar nuclei; FL, forelimb; HL, hindlimb; MF, motor face; PHA-L, Phaseoius vulgaris leu~a~lutinin; SC?, superior cerebellar peduncle; SF, sensory face; WGA-HRP, wheat germ agglutinin-ho~eradish peroxidase.

a notion is further supported by the observation of GABAergic local circuit neurons in the BPN of the rat5 as well as cat and p~mate.‘1,37 Moreover, it has also been shown that the basilar pons receives a dual feedback projection from the deep cerebellar nuclei (DCN) which is composed of glutaminergic6 and GABAergic4 fibers. Previous studies have more completely characterized the organization of the cerebellopontine system in terms of its origin from the major subdivisions of the DCN as well as its termination in specific regions of the BPN. ‘“~14~L7~18~38~40 An important question relative to the function of the BPN is whether the fibers of the cerebellopontine system terminate among BPN neurons in a manner that would facilitate an interaction with other afferent systems such as the projection from the cerebral cortex, or does the cerebellar system have its own 561

H. S.

562

LEE and

G. A.

interpositus and lateral nuclei) by lowering the injection micropipette through the mid-region of vermal lobule V at 1.5, 2.5 and 3.8mm from the midline. The depth of the micropipette was always maintained at 4.3 mm from the surface of the cerebellum and aligned in the vertical plane. At each cerebellar injection site, 0.2 ~1 of 2.5% PHA-L was injected over a 20-min period. WGA-HRP injections in the cerebral cortex were made within several regions of the cerebral cortex including motor face (MF), sensory face (SF), forelimb (FL) sensorimotor and hindlimb (HL) sensorimotor areas, the visual cortex, and the auditory cortex at a depth of l&l .2 mm. The injection sites in the sensorimotor cortex were stereotaxically determined according the map of Hall and Lindholmzo while the placement of tracer material in the visual or auditory cortex was based on previous studies.“i2 To establish that injections were localized within the MF, SF, FL or HL areas of the sensorimotor cortex, the visual or auditory cortex, we examined orthograde terminal labeling patterns in the thalamus (Fig. 1). After a survival period of seven to 10 days for the transnort of PHA-L and two to three davs for WGA-HRP. animals were deeply anesthetized with sodium ~ntobarbital (0.15 ml/l00 g) and perfused using a combination of low pH (acetate buffer, pH 6.5) and high pH (borate buffer, pH 9.5) fixatives devised for PHA-L immunocytochemistry.‘9 Briefly, 150 ml of saline was perfused through the vascular bed, followed by 250 ml of 4% paraformaldehyde in 0.1 M sodium acetate buffer (uH 6.5. 4°C) and another 250 ml of 4% parafo~aldeh~de~~nd 0.05% ~lutaraldehyde in 0.1 M sodium borate b&r (PH 9.5,4”C).Perfusion of the fixative was followed bv 250 ml of borate buffer (DH 9.5. 4°C) to prevent excessive depression of WGA-HRPenzyme activity by aldehydes remaining in the brain tissue. After the brain was removed, the basilar pons was immersed in borate (PH 9.5, 4”C), whereas the injection sites (the DCN and the cerebral cortex) were stored in the borate buffer containing 10% sucrose. On the following day, the basilar pons was sectioned at 40pm on a Vibratome and the injection sites on a freezing microtome. Sections (the basilar pons and injections sites) were then processed for WGA-HRP histochemistry as described by Rye et ~1.~~One series of pontine sections reacted for WGA-HRP was subsequently utilized for the immunocytochemical localization of PHA-L as described by Mihailoff ef a1.30 Preliminary experiments indicated that extremely dense WGA-HRP reaction product within corticopontine axon terminals in the BPN hindered the visualization of subsequent PHA-L reaction product in the same tissue sections. Consequently both the volume and concentration of WGA-HRP injected into the cerebral cortex were limited to 0.16-0.20 ~1 and l%, respectively, in order to maximize the subsequent visualization of preterminal axons and bead-Iike boutons labeled with PHA-L.

restricted termination sites in the BPN? Comparison of the observations contained in several earlier studies cited above would suggest that cerebellopontine termination zones do indeed partially overlap with several other BPN afferent systems. In the present study, the possibility of convergence of the cerebellopontine and the corticopontine systems was investigated at the light microscopic level, by injecting two different orthograde tracers at different locations in the brain of the same animal. One tracer was injected into the DCN while a second was introduced into one of several functional areas of the cerebral cortex. Subsequently the overlap of the terminal zones of these projections within restricted areas of the BPN was examined with light microscopy. Realizing that the overlap of afferent projection zones may simply represent the termination of inputs on separate populations of BPN cells located in close proximity to one another, studies were performed at the ultrastructural level to determine if both systems actually formed synaptic contact with single basilar pontine neurons. Convergence of cerebellopontine and corticopontine systems within the BPN could represent the anatomical substrate for an important integrative function involving BPN neurons and these two important afferent projections. EXPERIMENTAL

MIHAILO~~

PROCEDURES

Twenty-one Long-Evans black-hooded rats ranging in weight from 280 to 320g were used. Animals were anesthetized using 3.6% chloral hydrate (0.1 ml per IOOg of body weight). Light microscopic studies

In order to investigate the possible convergence of terminal zones of the cerebellopontine system with those of the corticopontine system, two different orthograde tracers were employed. In a single animal, Phuseolus ~u~guris leucoagglutinin (PHA-L), was injected into the major subdivisions of the DCN and a conjugate of wheat germ agglutinin and horseradish peroxidase (WGA-HRP) was placed within one of several cortical areas such as the sensorimotor, visual or auditory cortex. In one animal, the location of the two injected tracer substances was reversed, but this led to substantial HRP uptake by damaged axons in the cerebellar white matter and subsequent retrograde transport by pontocerebellar neurons that obscured orthogradely labeled DCN projection zones. The rat was anesthetized with chloral hydrate and placed in the stereotaxic apparatus with the incisor bar set 3.0 mm below horizontal. Both PHA-L and WGA-HRP were pressure-injected using a glass micropipette (tip diameter 20-30 pm) attached to a 2-~1 Hamilton syringe. The DCN injection sites were located using stereotaxic coordinates derived from the atlas of Paxinos and Watson3’ In each animal, three injections were made into the DCN (medial,

Ultrasrrucrural studies

In order to investigate the convergence of cerebella- and corticopontine projections on single pontine neurons, double-labeling studies were performed at the electron microscopic level in six animds. Because the peroxidasebased WGA-HRP and PHA-L reaction products are not differentiable at the ultrastructural level, in a series of rats the superior cerebellar peduncle (SCP) was transected at its

-. Fig. 1. Orthograde terminal labeling in the thalamus is used to confirm the location of WGA-HRP injections in each functional area of the cerebral cortex. Injections in the MF or the SF cortex produce labeling in the caudal ventrolateral (A) or ventroposterom~ial (B) thalamic nucleus, respectively. Both the ventrolateral thalamic nucleus (C) and the medial portion of the ventroposterolateral nucleus (D) are labeled following an injection in the FL sensorimotor cortex. Rostra1 portions of the ventrolateral nucleus (E) and lateral area of the ventroposterolateral nucleus (F) are labeled after an injection in the HL sensorimotor cortex. The lateral (G) or medial (H) geniculate nucleus is labeled following injections in the visual or auditory cortex, respectively.

Cortical and cerebellar convergence in basilar pans

Fig. I. NSC 39,3---B

563

H. S. LEE and G. A. MIHAILO~F

564

entry into the brainstem. This results in orthograde degeneration of ~~~~o~ntine axons and terminals which marks these structures for electron microscopic id~tifi~tion. Subsequently, WGA-HRP was injected into the cerebral cortex (MF and FL sensorimotor areas) of the same animal to label corticopontine boutons. After a post-lesion survival period of seven to 10 days which included two to three days for the transport of WGA-HRP, animals were perfused with 15Oml of saline followed by 500 ml of fixative containing 4% glutaraldehyde and 1% paraformaldehyde in 0.1 M phosphate buffer (PH 7.4). Excessive fixative remaining in the tissue was

removed by rinsing the vascular system with 250ml of phosphate buffer. The brain was then stored in the same buffer overnight. Two series of alternate 40 and 9Opm Vibratome sections were collected through the basilar pons. Both series of sections were then processed for WGA-HRE’ hist~h~st~ as described by Rye er a!.% After HRP reaction, the series of 40-&msections was mounted for light microscopic observation of WGA-HRP labeling, whereas the series of !30-pm sections was further processed for electron microscopic study as described by Carson and Mesulam.13 Briefly, sections were osmicated using 1% 0~0, in 0.1 M phosphate buffer @H 7.4) for 1 h, dehydrated in ethanol and propylene oxide and wafer-embedded between plastic coverslips. One side of the coverslip was coated with a commercially available Teflon spray. Following polymerization of the plastic (Epon-Araldite), the Teflon-coated coverslip was separated from the wafer using a razor blade. Appropriate areas were selected from the wafers, cut out using a scaipel and glued to blank plastic blocks using a quickd~ng cyanoacryhte adhesive. Ul~at~n sections were then prepared, stained with lead and uranyl salts, and examined with an electron microscope. RESULTS To investigate the possible overlap and convergence of cerebral cortical and cerebellar projections within the BPN, studies were performed that employed light microscopic and electron microscopic tracing methods. In light microscopic studies, two different orthograde tracers were injected into the DCN and cerebral cortex and the potential overlap of their respective projection fields in the BPN was evaluated. At the ultras~ctural level, studies were performed to investigate the possibility that synaptic boutons formed by the cortico- and cerebellopontine tierent systems actually contacted single pontine neurons. Light microscopic observations Cerebellopontine projections. Since PHA-L injections into major subdivisions of the DCN (medial, interpositus and lateral nuclei) were common to each of the injection combinations described in the following sections and such injections produced a consistent pattern of terminal labeling, they are described only in this section. Cerebellopontine terminal labeling was present within each major subdivision of the rostra1 fourfifths of the contralateral BPN, including the medial, ventral, lateral and peduncular (dorsal and ventral) nuclei (Figs 2-7). In transverse brainstem sections, patchlike labeling appeared to represent an aggregate

of preterminal

axons and bead-like boutons corresponding to the lon~tu~nally (ros~~audaIly) oriented columns observed when autoradio~aphic tracing was employed.38 In a representative case (animal Rl748), only a small number of cerebellopontine terminal fibers were observed at the rostra1 pole of the BPN (Fig. 2Bl). Patches of dense PHA-L labeling were prominent in more caudal sections, especially in the medial and ventral BPN (Fig. 2B2-4). Patches of terminal labeling in the medial nucleus were present at rostra1 levels, whereas those in the ventral nucleus began to appear somewhat more caudally. A moderate number of preterminal axons and boutons were also diffusely dist~but~ throu~out the lateral BPN at these same levels. At mid-pontine levels small patches of PHA-L labeling were observed in the dorsomedial, midline, ventral peduncular and ventral regions of the ventral nucleus (Fig. 2B5-7). A thin band of PHA-L preterminal axons and boutons was often observed in the lateral BPN, adjacent to the lateral edge of the cerebral peduncle (Fig. 2B5-7). At caudal pontine levels, PHA-L terminal labeling became extremely sparse (Fig. 2B8-9). Overlap with corticopontine projections from the motor face area. Animal RI748 was chosen as a

representative case that illustrate the typical pattern of cerebellopontine and MF ~o~copontine terminal labeling (Fig. 2). A WGA-HRP injection involving areas of motor cortex containing the representation for musculature of the face resulted in the labeling of corticopontine terminal fields which exhibited a rostrocaudal columnar arrangement within the ipsilateral BPN. Corticopontine terminal labeling was observed within the medial portion of the ventral nucleus beginning at approximately 52Opm from the rostra1 pole of the BPN (Fig. 2B3). At midpontine levels, two major columns of corticopontine terminal labeling were observed, one located in the medial nucleus and the other in the ventral region of the ventral nucleus (Fig. 2B4-6). At more caudal levels a column of corticopontine labeling was still present in the medial nucleus (Fig. 2B7), but in the most caudal sections no terminal labeling was observed (Fig. 2B8-9). Rather extensive overlap was observed between the pontine terminal zones formed by afferents from the MF cortex and DCN (Fig. 2B3-5). At rostra1 pontine levels such overlap was located in an area extending into the medial and ventral BPN (Fig. 2B3-4, see also Fig. 8A, B). At mid-pontine levels, the overlap region was divided into two separate zones: one located that the ventral portion of the medial nucleus and the other more ventrally in the ventral nucleus (Fig. 2B5). No overlap was observed at caudal levels (Fig. 2B6-9). Overlap with projections from the sensory face cortex. Because the cortical area comprising the

somatosensory representation for the face (SF) is injections rather extensive,*’ multiple WGA-HRP

Cortical and cerebeliar convergence in basilar pons

565

R-1748 Motor Face Cx. /

Cerebellar

Nuclei

B. \

Fig. 2. A dorsal view of the rat brain indicating the locations of HRP placement in MF cortex and PHA-L injections in the cerebellar nuclei is shown in A. A rostrocaudal series of transverse sections through the BPN (B) demonstrates the terminal labeling produced by the cortical and cerebellar injections. Arrows in sections 3-5 indicate the overlap of terminal zones.

were usually employed in these cases. In a representative case (auimal R1738) the corticopontine terminal labeling was quite dense and organized as several lon~tudinally oriented columns (Fig. 3). At rostra1

pontine levels, a column of corticopontine terminal labeling was observed in the lateral portion of the ventral nucleus (Fig. 3W-4), while another column of terminal labeling began to appear more caudally in

H. S. LEEand G. A. MIHAILOFF

566

R-1738 Sensory

Face Cx. /

Cerebellar

Nuclei

Fig. 3. HRP is injjected in the SF cortex, while PHA-L is injected in the cerebellar nuclei (A). A rostrocaudal series of transverse sections through the BPN demonstrating the tenninal overlap (arrows) is shown in B.

the medial nucleus (Fig. 3B4). This terminal labeling gradually shifted in a mediodorsal direction and eventually was located in the dorsomedial BPN region (Fig. 3B6). The single aggregate in the ventral

nucleus (Fig. 3B3-4) became subdivided into several small columns each of which extended further caudally; one column was located ventral to the cerebral peduncie (Fig. 3B5) and two other columns

Cortical and cerebellar convergence in basilar pons

561

R-1746 Forelimb

Cx. /

Cerebellar

Nuclei

a 1720 Cm

9 1960 pm

Fig. 4. In animal R1746, HRP is injected in the FL sensorimotor cortex and PHA-L is injected in the cerebellar nuclei (A). Extensive overlap of terminal zones (arrows) is observed (B).

occupied medial and lateral positions within the ventral portion of the ventral nucleus (Fig. 3B6). Finally, at caudal pontine levels, a crescent-shaped longitudinal column enveloping the ventrolateral and dorsolateral aspects of the cerebral peduncle was present in the lateral nucleus (Fig. 3B7-9). Unlike the

previous injection case which combined the MF cortex and DCN (Fig. 2), there was only a minimal degree of overlap between the terminal zones from the SF cortex and DCN (Fig. 3B4,6,7). Rostrally, a small overlap zone was located in the ventral portion of the ventral BPN (Fig. 3B4). At mid- to caudal

568

H. S. LEE and G. A. MMAILOFF

pontine levels, a column of overlapping projections (Fig. 3B6-7) was observed in approximately the same area as the one located more rostrally (Fig. 3B4). Another small overlap area was located in the dorsomedial BPN (Fig. 3B6). Overlap with projections from the forelimb sensorimotor cortex. Animal R1746 was selected as a representative case that exhibited the characteristic patterns of FL corticopontine and ~r~~llopontine terminal labeling (Fig. 4). A WGA-HRP injection involving portions of both the FL sensory and motor areas produced several longitudinally oriented columns in the BPN. At mid-pontine levels, dense WGA-HRP terminal labeling was located in the dorsal portion of the ventral nucleus, just ventral to the cerebral peduncle (Fig. 4B4). At more caudal levels, relatively small patches of WGA-HRP terminal labeling were observed in the ventral and dorsal peduncular regions (Fig. 4B5-6). At its caudal extent, dense corticopontine labeling was present in two locations: one in the medial nucleus and the other in the ventral nucleus (Fig. 4B6-8). Extensive overlap was observed between the terminal zones from the FL sensorimotor cortex and DCN (Fig. 4B4-7). A dense overlap one was located in the dorsal portion of the ventral nucleus at mid-pontine levels (Fig. 4B4; see also Fig. 8C,D). There were also two small overlap zones: one located rostrally in the ventral peduncular region and the other more caudally in the dorsal peduncular region (Fig. 4B5-6). More caudally, additional overlap was apparent in an area extending into the lateral portion of the medial nucleus and the medial portion of the ventral BPN (Fig. 4B7). Overlap with projections from the h~dlimb sensorimotor cortex. In a representative case (animal Rl735), corticopontine projections originating from the HL area produced several longitudinal columns of terminal labeling (Fig. 5). WGA-HRP labeling was present in the ventral nucleus just ventral to the cerebral peduncle at a rostra1 pontine level (Fig. 5B3). At mid- to caudal pontine levels terminal labeling was observed in two discrete regions: one located near the midline region of the medial nucleus and the other in the lateral nucleus (Fig. 5B6-7). Corticopontine terminal labeling in the lateral nucleus was dense and widespread, whereas the projection to the medial nucleus was relatively weak. Labeling was also observed in the ventral, as well as the medial and lateral nuclei (Fig. 5B8). More caudally, terminal zones in the medial, ventral and lateral nuclei appeared to merge (Fig. 5B9). Rather sparse overlap between terminal zones of the HL sensorimotor cortex and those of the DCN was observed only at rostra1 pontine levels, particularly in the dorsal portion of the ventral BPN (Fig. 5B3). No overlap was observed at mid- to caudal pontine levels; PHA-L labeled cerebellopontine projections were observed in the dorsomedial, midline, ventral peduncular, ventral and lateral nuclei at

mid-pontine levels (Fig. 5B4-6), whereas WGA-HRP terminal labeling from the HL sensorimotor cortex was mainly located at caudal pontine levels (Fig. 5B7-9). Overlap with projections from the visual cortex. Case R1732 was selected to represent the pontine terminal zones that are labeled by an injection in the visual cortex (Fig. 6). These projection fields were mainly observed in the rostra1 two-thirds of the BPN in the ventrolateral portion of the lateral nucleus (Fig. 6B2-5). More caudally another small area of termination occupied the dorsolateral portion of the lateral nucleus (Fig. 6B6-7). A rather small but distinct area of overlap involving the terminal zones of the visual corticopontine and cerebellopontine projections was located in the ventral portion of the lateral nucleus at rostra1 pontine levels (Fig. 6B2-4; see also Fig. gE,F). More caudally, overlap was also observed in the dorsolateral portion of the lateral BPN (Fig. 6B6). Ouerl~p with projections from the auditory cortex. In a representative case (animal R1752), corticopontine terminal labeling originating from the auditory cortex was observed in the ventrolateral portion of the lateral nucleus at rostra1 to mid-pontine levels (Fig. 7B3-5). These projection zones occupied the same region of the lateral BPN as those from the visual cortex, but, in addition, they extended more medially into the ventral nucleus. At more caudal levels, the dorsolateral BPN also contained terminal labeling (Fig. 7B6-7). The overlap between the auditory corticopontine projections and cerebellopontine terminals was minimal (Fig. 7B4-7). A small zone of overlap was located in the ventral region of the lateral BPN at mid-pontine levels (Fig. 7B4-4). More caudally, overlap was also observed in the dorsolateral portion of the lateral BPN (Fig. 7B6-7). Ultrastructural observations. The protocol that was followed in a single animal, combined the placement of a lesion in the SCP and an injection of WGA-HRP into the contralateral sensorimotor cortex (MF and FL sensorimotor areas). The purpose of these experiments was to determine if in fact both the cerebella- and corticopontine systems actually form convergent synaptic boutons with single pontine neurons. Sub~quently the BPN neuropil was searched for instances in which both types of Iabeled boutons contacted the same dendrite or separate dendrites that arose from the same neuron. The light microscopic studies described earlier in this report were utilized to identify those pontine regions that contained overlapping projection zones and thus were to be investigated with electron microscopy. A large number of examples were found, in which cerebellopontine and corticopontine boutons formed synaptic contact with separate dendritic profiles located within close proximity to one another. WGA-HRP-ladle corticopontine terminals containing spheroidal vesicles were small and formed as~met~c synaptic contact with small (< 1 pm)

Cortical and cerebellar convergence in basilar pons

569

R-1735

240

pm

8 1680 pm

9 1920em

Fig. 5. Orthograde tracers, HRP and PHA-L, are injected in the HL sensorimotor cortex and cerebellar nuclei, re.spectively(A). Only a small amount of overlap is observed at rostra1 pontine levels (B). dendritic spines or distal dendritic shafts (Fig. 9B,C). In contrast, degenerative cerebellopontine terminals were often relatively large (24pm) and formed glomerular synaptic complexes with several claw-like dendritic structures (Fig. 9A,C). Degenerative cerebellopontine boutons exhibited either a dark

appearance resulting from a large accumulation of synaptic vesicles (Fig. 9A), or contained an array of swollen vesicles and tubulovesicular profiles (Fig. QC). In addition, both types of degenerative boutons also exhibited small, round, clear profiles (Fig QA,C, arrowheads) that appear to be membrane.

H. S. LEEand G. A. MIHAILOFF

R-1732 Visual

Cx. /

Cerebellar

Nuclei

\ \ 2 240 pm

Fig. 6. A dorsal view of the rat brain indicating HRP placement in the visual cortex and PHA-L injections in the cerebellar nuclei (A). In a rostrocaudal series of BPN sections (B), overlap of terminal zones (arrows) is observed

in the ventral

and dorsolateral

bound. These structures are interpreted to be glial processes that are invaginating and engulfing the degenerative boutons. Situations in which identified boutons of both afferent systems contacted a single dendrite were

portions

of the lateral

nucleus.

infrequent. However, one clear example of cortical and cerebellar convergence is illustrated in Fig. 10. In the center of the field is a BPN neuron dendrite that receives synaptic contact from both a WGA-HRPlabeled cortical bouton (open block arrow) and a

571

Cortical and cerebeliar convergence in basilar pons PHA -L

A.

R-1752 Auditory

Cx. /

Cerebellar

Nuclei

[D

Fig. 7. HRP is injected in the auditory cortex and PHA-L is injected in the cerebellar nuclei (A). Overlap of terminal zones (arrows) is similar to that observed in combined visual cortex and cerebellar nuclear injections (B). shrunken, glial enwrapped cerebellar terminal (solid block arrow). Although, as indicated in Fig. 9C, it

was common to observe cortical and cerebellar boutons close to one another in the neuropil, it proved impossible in these situations to determine if both types of boutons might have contacted separate dendrites of the same BPN neuron.

DISCUSSION

Projections from the cerebellar nuclei to the basilar pons have been demonstrated in all species examined~‘“~‘4~‘7*3*~40 however, the functional properties of this feedback system have not yet been made clear. Although previous studies indicated that the

512

H. S. LEEand G. A.

MIHAILOFF

Fig. 8. Low (left column: A, C, E) and high (right column: B, D, F) magnification views of the overlap between WGA-HRP-labeled corticopontine terminals (black granules) and PHA-L-labeled cerebellopontine preterminal axons and boutons (brown fibers with varicosities). (A and B) Combination of MF cortex and cerebellar nuclear injections (animal R1748). (C and D) Combination of FL sensorimotor cortex and cerebellar nuclear injections (animal R1746). (E and F) Combined injections of visual cortex and cerebellar nuclei (animal R1732). Areas surrounded by arrowheads represent overlap zones of corticopontine and cerebellopontine terminals.

Cortical and cerebellar convergence in basilar pons

513

Fig. 9. A representative example of a degenerative cerebellopontine bouton is shown in A. Such boutons often exhibited an increased electron density suggestive of dark degeneration. Also present are lucent glial processes surrounding the reactive bouton and invaginating into it at three locations (arrowheads). As in control material, such boutons often formed synaptic contact with several dendritic profiles (*c). In B a representative WGA-HRP labeled corticopontine bouton (block arrow) forms an asymmetric synaptic contact with a dendritic profile. As shown in C, cortical (open arrow) and cerebellar (closed arrow) boutons were often observed in close proximity within the neuropil. With regard to the degenerating cerebellar bouton (closed arrow) note the presence of lucent glial processes (gl) around and within (arrowheads) the bouton as well as the swollen vesicles and unusual tubulovesicular structures within the bouton. Synaptic contacts are formed with two of three nearby dendritic profiles (*). Scale bar = I pm. The magnification factor is the same for A, B and C.

cerebellopontine projection might provide an excitatory input to the BPN,’ immunocytochemical studies suggested that the cerebellopontine system is composed of inhibitory4 as well as excitatory6 components. The present study employed anatomical methods to explore the ~ssibility that cerebellar and cortical afferents might converge on single BPN neurons. Observations from both light and electron

microscopic studies support the notion that some BPN neurons do in fact receive synaptic input from both the cortico- and cerebellopontine systems. Light microscopic observations

The present findings demonstrated that in the rat, portions of the ~re~llopontine terminal zones occupied some of the same regions of the BPN as did

514

H. S. LEEand G. A. MIHAILOFF

Fig. 10. Illustrated here is an example of cerebellar and cortical convergence onto a BPN neuron. In the center of the field a transversely sectioned dendrite (Dd) receives synaptic contact from a WGA-HRP labeled corticopontine bouton (open block arrow) and a degenerative cerebellopontine terminal (solid block arrow) which is shrunken and partially enwrapped by lucent glial processes (*). Fragments of the degenerative bouton (arrowheads) are completely engulfed by the glial process.

projections from various areas of the cerebral cortex, such as the sensorimotor, visual, and auditory cortices (Figs 2-7). Overlap existed when a patch of PHA-L-labeled cerebellopontine preterminal axons was located within the perimeter of a punctate, WGA-HRP-labeled corticopontine terminal zone. The location of one or two threads of diffuse PHA-L labeling within a corticopontine terminal area, was not considered to be evidence of overlap. Once the overlap zones were identified, they were used in subsequent ultrastructural studies as areas to be investigated for the presence of synaptic contacts linking the cortical and cerebellar afferents with the same BPN neuron. Light microscopically observed overlap of terminal zones, formed by cerebellopontine inputs and sensorimotor corticopontine projections from the MF, SF, FL and HL areas, was mainly located in restricted regions of the medial, ventral and peduncular BPN (Figs 2-7). Brodal et a[.” indicated that portions of the cerebellopontine projection area in the cat overlap with termination sites of corticopontine fibers from cortical regions including SM I and SM II, particularly in the paramedian and dorsolateral BPN. Similarly, the present study in the rat described overlap between cerebellopontine and the sensorimotor corticopontine terminal zones in the medial nucleus (corresponds to the paramedian nucleus in the cat), whereas the degree of overlap in the rat dorsolateral BPN (Figs 2-7) was less extensive than in the cat. In similar studies in the opossum, Yuen et ~1.~ also described a partial overlap between the cerebellar and sensorimotor cortical BPN afferents, but they illus-

trated virtually no confluence between visual or auditory cortical projection fields and those of the cerebellar nuclei. In contrast, the present anatomical study in the rat described a small area of moderately dense overlap involving cerebellar projection zones and those of visual and auditory cortex that was consistently observed in ventral and dorsolateral portions of the lateral BPN (Figs 6,7). Our observations also demonstrated that each region of the sensorimotor cortex (MF, SF, FL and HL areas) exhibited a variable degree of terminal zone overlap with projections from the DCN (Figs 2-7). There was an extensive overlap of the cerebellopontine system with projections from the MF cortex (Fig. 2), while overlap with the SF cortex was relatively limited (Fig. 3). A similar finding was previously suggested in an autoradiographic study of rat cerebellopontine projections.38 Extensive terminal overlap between the cerebcllopontine system and the FL (and face) corticopontine system has also been described in the opossum.ZS~40 Moreover, in the rat, overlap of terminal zones involving cerebellopontine afferents and FL sensorimotor corticopontine projections was prominent (Fig. 4), whereas cerebellar overlap with HL sensorimotor corticopontine projections was relatively minimal (Fig. 5). FL motor and FL sensory cortices were not investigated independently because even though there is only a partial overlap of the FL motor and sensory cortices, the non-overlapping fields are quite small and injections restricted to such zones would be difficult to achieve even under electrophysiological control. Although previous reports indicated that a large proportion of

Cortical and cerebellar convergence in basilar pons cerebellopontine fibers arise as collateral branches of cerebellothalamic projections,24 at this point it is unclear whether the convergence described in the present study might occur between the cerebellopontine system and the oerebellar efferent-related corticopontine system (involving the cerebellothalamo-cortico-pontine pathway) or those corticopontine projections that are not influenced by the cerebellar efferent system. In addition to the overlap zones, PHA-L-labeled ~re~llopontine fibers were also distributed to other regions of the BPN. For example, areas within the medial BPN region at rostra1 pontine levels, particularly dorsomedial and midline regions (e.g. Fig. 2B2-3), exhibited extremely dense patches of cerebellopontine labeling (Figs 2-7). The presence of cerebellopontine projections in areas that are completely devoid of any co~icopontine terminals was also illustrated in the opossum.@ It appears that in the rat BPN, one of the areas (dorsomedial region) that receives cerebellar but not cortical projections might also receive input from the medial and lateral mammillary nuclei. I6 Therefore, it remains to be determined whether such areas of the BPN that appear to receive projections from the DCN but not the cerebral cortex might actually represent a site of convergence involving the cerebellar input and one of the numerous other BPN afferent systems.

The ultrastructural portion of this study described a protocol that marked the synaptic boutons of the corticopontine system with orthogradely transported WGA-HRP while the cerebellopontine boutons were identified by degeneration in response to transection of their parent axons in the superior cerebellar peduncle. It was not uncommon to observe both types of boutons in the same field and in close proximity to one another but in most instances such boutons were not in synaptic contact with the same postsynaptic profiie. However, in several instances when a fairly long obliquely or longitudinally sectioned dendrite was present in the field, it was possible to establish that WGA-HRP-labeled cortical boutons and degenerative cerebellar boutons were in synaptic contact with that dendrite. Also several examples were observed in which a transversely sectioned dendrite was contacted by both a cortical and a cerebellar axon terminal. These observations confirm the convergence of cortical and cerebellar inputs at the level of the BPN. The infrequent observation of convergent cerebellopontine and corticopontine terminals on single BPN dendritic profiles in the present study might, in part, result from the differential distribution of the termination sites of the two inputs along different portions of the dendritic surface of pontine neurons. Previous studies in the rat3’ and also in the opossum2* and monkey” have, in fact, shown that corticopontine boutons primarily form synapses with dendritic spines and distal dendritic shafts of BPN

575

neurons, whereas only a relatively small proportion of cerebellopontine boutons are in synaptic contact with the distal portion of the dendritic tree. The majority of cerebellar boutons form a glomerular type arrangement with dendritic spines and the shafts of intermediate or proximal dendrites.39 Interestingly, those cerebellar boutons that do form synapses with distal dendritic shafts are among the smallest of the cerebellar types. The frequent observation of cerebellopontine and corticopontine terminals in close proximity within the neuropil might actually represent a ~rcumstance in which the boutons of the two afferent systems are terminating on different regions of the dendritic tree of the same BPN neuron. Of course, it remains conceivable that in this situation, the cortical and cerebellar boutons are seeking out two different classes of BPN neurons, the dendrites of which are not spatially separated. Since there is now fairly strong evidence that BPN GABA neurons are local circuit neurons,5,7,“,26,37it could be that such neurons and the BPN projection neurons receive segregated cortical and cerebellar inputs. Alternatively one type of BPN neuron (but not the other) might receive convergent inputs. Future studies will need to address these important questions. Although the question of how cerebella- and corticopontine projections might function in the regulation of motor behavior requires further electrophysiological evaluation, the integrative capacity of single BPN neurons has been established for other combinations of pontine afferent systems. Potter et a1.33reported that in the rat, slightly less than half of those BPN neurons that responded to electrical stimulation of sensorimotor, visual, or auditory cortex also displayed a convergent input from two or more cortical areas. Convergence of multiple cortical inputs on a single BPN neuron was also reported in the cat” and the monkey. 34A recent anatomical study in our laboratory indicated that in the rat partially overlapping terminations are observed in the caudal half of the BPN between projections from the nucleus cuneatus and the FL sensory cortical area, as well as between projections from the nucleus gracilis and the HL sensorimotor cortical area.22 Parallel studies involving single-unit recording demonstrated that approximately 40% of cells recorded in the rat BPN received convergent peripheral (probably via the dorsal column system) and cortical inputs, especially from the forepaw and FL somatosensory cortex.23 Thus, it appears that one of the fundamental functional operations performed in the BPN might involve the processing of various convergent afferent signals, as shown both in the present studies of cerebral cortical and cerebellar nuclear afferent combinations, as well as in previous studies involving two different areas of the cortex,33-35 or the cortex and somatosensory peripheral input combinations.22*23 This notion is further supported by recent studies in the rat2’ and cat’ that revealed a greater diversity

H. S. LEEand G. A. MIHAILOFP

576

of afferent systems reaching previously been demonstrated.

the BPN than had

CONCLUSION On the basis of the present light and electron microscopic studies, it appears that circuitry exists that would allow BPN neurons to receive convergent inputs from the cerebral cortex and cerebellar nuclei. Preliminary single-unit recording has revealed that in the rat, some basilar pontine neurons do receive convergent cerebellar nuclear and cerebral cortical inputs (Azizi and Kosinski, personal communication). For some BPN neurons, both inputs are excitatory, while for others the cortical input was excitatory and the cerebellar input inhibitory. The precise functional role of the cerebellopontine system

remains relatively unclear. However, since the input from the cerebral cortex appears to be a major excitatory driving force for BPN neurons, the cerebellopontine fibers, via a convergent synaptic input mechanism, might serve to either decrease or enhance the magnitude of the cortical input to a given BPN neuron. Alternatively, the cerebellar feedback might serve to select (or deselect) the mossy fiber input arising from a subset of pontocerebellar neurons. While it is difficult to assess the behavioral consequences of these functions, such circuitry could be utilized in both the planning and updating of volitional movement as a result of the widespread projections of the pontocere~llar system to virtually all areas of the cerebellar cortex. Acknowledgement-The

support provided by NINDS (NS12644) is gratefully acknowledged.

REFERENCES 1. Aas J. E. (1989) Subcortical projections to the pontine nuclei in the cat. J. camp. Neurol. 282, 331-354. 2. Allen G. I. and Tsukahara N. (1974) Cerebrocerebeller communication systems. Physiol. Rev. 54, 957-1006. 3. Azizi S. A., Burne R. A. and Woodward D. J. (1985) The auditory corticopontocerebellar projection in the rat: inputs to the paraflocculus and midvermis. An anatomical and physiological study. Expf Brain Res. 59, 3649. 4. Border B. G., Kosinski R. J., Azizi S. A. and Mihailoff G. A. (1986) Certain basilar pontine afferent systems are GABAergic: combined HRP and immun~ytffihemical studies in the rat. Brain Res. Bun! 17, 169-179. 5. Border B. G. and Mihailoff G. A. (1985) GAD-immunoreactive neural elements in the basilar pontine nuclei and nucleus reticularis te~enti pontis of the rat. I. Light microscopic studies. Exp/ Brain Res. 59, MxMl4. 6. Border B. G. and Mihailoff G. A. (1987) Evidence for the existence of ~utaminergic basilar pontine afferent systems in the rat. Neurosci. Abstr. 13, 1263. 7. Border B. G. and Mihailoff G. A. (1990) GABA-ergic neural elements in the rat basilar pons: electron microscopic immunochemistry. J. camp. Neural. 295, 123-135. 8. Brodal P. (1978) Principles of organization of the monkey corticopontine projection. Bruin Res. 148, 214-218. 9. Brodal P. (1982) The cerebropontocerebellar pathway: salient features of its organization. Expl. Bruin Res. 6, 108-133. 10. Brodal P., Destomb-es J., Lacerde A. M. and Angaut P. (1972) A cerebellar projection onto the pontine nuclei. An experimental anatomical study in the cat. Expl Bruin Res. 16, 543-558. Il. Brodal P., Mihailoff G. A., Border B. G., Ottersen 0. P. and Storm-Mathison J. (1988) GABA-containing neurons in the pontine nuclei of the rat, cat and monkey. An immunocytochemical study. Neuroscience 25, 2745. 12. Burne R. A., Mihailoff G. A. and Woodward D. J. (1978) Visual corticopontine input to the paraflocculus: a combined autoradiographic and horseradish peroxidase study. Bruin Res. 143, 139-146. 13. Carson K. A. and Mesulam M.-M. (1982) Electron microscopic tracing of neural connections with horseradish peroxidase. In Tracing Neural Canneetions with horseradish Peroxiduse (ed. Mesulam M.-M.), pp. 153-184. John Wiley New York. 14. Chan-Palay V. (1977) Cerebellar Dentate Nucfeusr ~rg~i~ution, Cytology, and ~ra~mitters, pp. 297-326. Springer, New York. 15. Cooper M. H. and Beal J. A. (1978) The neurons and synaptic endings in the primate basilar pontine gray. J. cump. Neural. ISO,17-42. 16. Cruce J. A. (1977) An autoradiographic study of the descending connections of the mammillary nuclei of the rat. J. camp. Neural. 176, 631644. 17. Faull R. L. M. (1978) The cerebellofugal projections in the brachium conjunctivum of the rat. II. The ipsilateral and contralateral descending pathway. J. camp. Neural. 178, 519-536.

18. Faull R. L. M. and Carman J. B. (1978) The cerebellofugal projections in the brachium conjunctivum of the rat. 1. The contralateral ascending pathway. J. camp. Neurol. 178, 495-518. 19. Gerfen C. R. and Sawchenko P. E. (1984) An anterograde neuroanatomical tracing method that shows the detailed morphology of neurons, their axons, and terminals. Bruin Res. 290, 219-238. 20. Hall R. D. and Lindholm E. P. (1974) Organization of motor and somatosensory neocortex in the albino rat. Brain Res. 66, 23-38. 21. Hartmann-von Monakow K., Akert K. and Kunzle H. (1981) Projections of precentral, premotor, and prefrontal cortex to the basilar pontine gray and to NRTP in the monkey (Macaca fasc~cu~uris). Archs Suisses Neural. ~earo~hir. psychiur. 129, 189-208. 22. Kosinski R. J., Azizi S. A., Border B. G. and Mihailoff G. A. (1986) Origin and ultrastructural identification of dorsal column nuclear synaptic terminals in the basilar pontine gray of rats. J. camp. Neurol. 253, 92-104.

23. Konsinski R. J., Azizi S. A. and Mihailoff G. A. (1988) Convergence of cortico- and cuneopontine projections onto components of the pontocerebellar system in the rat: and anatomical and electrophysiological study. Expl Brain Res. 71, 541-556. 24. Lee H. S., Kosinski R. J. and Mihailoff G. A. (1989) Collateral branches of cerbellopontine axons reach the thalamus,

superior colliculus, or inferior olive: a double-fluorescence Neuroscience 28, 725-734.

and combined fluorescence-HRP

study in the rat.

Cortical and cerebellar convergence in basilar pons

577

25. Martin G. F., King J. S. and Dom R. M. (1974) The projections of the deep cerebellar nuclei of the opossum, Diaklphb marsupialis virginiana. J. Himforsch. 15, 545-573. 26. Mihailoff G. A. and Border B. G. (1990) Evidence for the presence of presynaptic dendrites and GABA-immunogold labeled synaptic boutons in the monkey basilar pontine nuclei. Brain Res. 516, 141-146.

27. Mihailoff G. A., Bume R. A. and Woodward D. J. (1978) Projections of the sensorimotor cortex to the basilar pontine nuclei in the rat: an autoradiographic study. Brain Res. 145, 347-354. 28. Mihailoff G. A. and King J. S. (1975) The basilar pontine gray of the opossum; a correlated light and electron microscopic analysis. J. camp. Neurol. 159, 521-552. 29. Mihailoff G. A., Kosinski R. J., Axiti S. A. and Border B. G. (1989) Survey of noncortical a&rent projections to the basilar pontine nuclei: a retrograde tracing study in the rat. J. camp. Neural. Zsz, 617643. 30. Mihailoff G. A., Lee H., Watt C. B. and Yates R. (1985) Projections to the basilar pontine nuclei from face sensory and motor regions of the cerebral cortex in the rat. J. camp. Neurol. 237,251-263. 31. Mihailoff G. A., Watt C. B. and Bume R. A. (1981) Evidence suggesting that both the corticopontine and cerebellopontine systems are each composed of two separate neuronal populations: an electron microscopic and horseradish peroxidase study in the rat. J. camp. Neurol. 195, 221-242. 32. Paxinos G. and Watson C. (1986) The Rat Brain in Stereotaxic Coordinates, 2nd edn. Academic Press, Australia. 33. Potter R. F., Ruegg D. G. and Wiesendanger M. (1978) Responses of neurons of the pontine nuclei to stimulation of the sensorimotor, visual and auditory cortex of rats. Brain Res. BUN. 3, 15-19. 34. Ruegg D. G., Seguin J. J. and Wiesendanger M. (1975) Cortical effects from somatic areas on single neurons of the pontine nuclei. Can. Physiol. 6, 52. 35. Ruegg D. G. and Wiesendanger M. (1975) Corticofugal effects from sensorimotor area I and somatosensory area II on neuron of the pontine nuclei on the cat. J. Physioi. 247, 745-757. 36. Rye D. B., Saper C. B. and Wainer B. H. (1984) Stabilization of the tetramethylbenxidene reaction product. J. Histochem. Cytochem. 32, 1145-I 153. 37. Thier P. and Koehler W. (1987) Mo~holo~, number and dist~bution of putative GABA-ergic neurons in the pontine basilar grey of the monkey. J. camp. Neurot. 265, 311-322. 38. Watt C. B. and Mihailoff G. A. (1983) The cerebellopomine system in the rat. I. Auto~dio~aphic studies. J. camp. Neurol. 215, 312-330. 39. Watt C. B. and Mihailoff G. A. (1983) The cerebellopontine system in the rat. II. Electron microscopic studies. J. camp. Neurol. 216, 429437. 40. Yuen H., Dom R. M. and Martin G. F. (1974) Cerebellopontine projections in the American opossum. A study of their origin, distribution and overlap with fibers from the cerebral cortex. J. camp. Neural. 154, 257-285. (Accepted 15 May 1990)