Zonal organization of cortico-nuclear and nucleo-cortical projections of ...

7 downloads 0 Views 1MB Size Report
Zonal organization of cortico-nuclear and nucleo-cortical projections of the paramedian lobule of the cat cerebellum. 2. The C2 zone. Received: 28 April 1997 ...
 Springer-Verlag 1998

Exp Brain Res (1998) 118:316±330

RESEARCH ARTICLE

J.R. Trott ´ R. Apps ´ D.M. Armstrong

Zonal organization of cortico-nuclear and nucleo-cortical projections of the paramedian lobule of the cat cerebellum. 2. The C2 zone

Received: 28 April 1997 / Accepted: 22 July 1997

Abstract The cortico-nuclear (C-N) and nucleo-cortical (N-C) projections of the C2 cortical zone in pars anterior (pa) and pars posterior (pp) of the paramedian lobule (PML) in the posterior lobe of the cat cerebellum were investigated with a combined electrophysiological and neuroanatomical technique. In each experiment the mediolateral boundaries of the zone were localized on the cortical surface by recording field potentials mediated via climbing fibres and evoked in the zone by activity elicited in spino-olivocerebellar paths through percutaneous stimulation of the fore- and hindlimbs; a small (15±30 nl) injection of 1±2% WGA-HRP was then made into the zone. Distributions in the deep cerebellar nuclei were determined with light microscopy both for C-N terminal labelling due to anterograde axonal transport by Purkinje cells and for cell bodies labelled due to retrograde transport in N-C axons. The extent to which retrogradely labelled olivary neurones were confined to the part of the rostral medial accessory olive that innervates the C2 zone was estimated to provide an indication of the degree to which the injected tracer might have spread beyond the boundaries of the zone. The C-N projection from the part of the C2 zone in PML pa terminates almost exclusively (probably exclusively) in nucleus interpositus posterior (NIP) at all medio-lateral levels of the nucleus but most extensively at middle and lateral levels. At most levels the C-N termination territory forms a crescent with its outer curve following the caudal, dorsal and rostral borders of the nucleus and as a result it is mainly in the dorsal half of the nucleus. There is heavy overlap with the projection from the lobule V part of the C2 zone previously studied by us. The projection from the C2 zone in PML pp terminates entirely in NIP, but although at middle medio-lateral levels in the nucleus there is substantial overlap with the PML pa and lobule V projections, the projection territory is confined to the medial half of the nucleus. Evidence was obtained compatible with the view that throughout

)

J.R. Trott ´ R. Apps ´ D.M. Armstrong ( ) Department of Physiology, School of Medical Sciences, University of Bristol, University Walk, Bristol BS8 1TD, UK Fax: +44-117-928-8923

the C2 zone its lateral and medial parts project to different parts of NIP. In both PML pa and pp the C2 zone receives N-C projections from NIP. Most of the N-C cells concerned are in the dorsal half of NIP and the great majority lie within the corresponding C-N projection territory. However, the N-C projection to PML pa appears c. 6 times heavier than that to PML pp and the PML pa part of the zone also receives a minor additional projection from nucleus lateralis (NL). The findings are discussed in relation to the hypothesis of olivo-cortico-nuclear complexes or compartments, with particular reference to the internal organization of the C2 complex. Key words Purkinje cells ´ Cerebellar nuclei ´ Cerebellar zonation

Introduction It is now widely acknowledged that one fundamental organizational principle that applies to the cerebellar neuronal circuitry in mammals is its division into a number of ªparallelº functional units variously termed complexes, compartments or modules (e.g. Voogd and BigarØ 1980; Ito 1984). Each includes a narrow ªlongitudinalº zone of cortex that receives its climbing fibres from a localized subgroup of (contralateral) inferior olivary neurones. The Purkinje cells of the zone provide the (inhibitory) cortico-nuclear (C-N) projection to a particular ªdedicatedº deep cerebellar nucleus that also receives (excitatory) axon collaterals from the olivary region projecting to the cortical zone and in turn provides a nucleoolivary projection to the same part of the olive (for a study and earlier references see Dietrichs and Walberg 1981; also DeZeeuw et al. 1989; Nelson and Mugnaini 1989). In light of these connections the complexes are frequently described as ªolivo-cortico-nuclearº (O-CN) in nature. However, much remains to be defined concerning these functional units ± for example the full extent to which they differ in their mossy-fibre-mediated afferent

317

inputs and in their efferent projections to extra-cerebellar targets other than the inferior olive; also the extent to which their neurones differ in their activity during movements. In addition, there are significant gaps in our knowledge of their internal organization, especially outside the vestibulo-cerebellum. Even among the complexes most thoroughly investigated, i.e. the C1, C2 and C3 that collectively make up the paravermal part of the cerebellum, most studies (e.g. Groenewegen et al. 1979; Oscarsson 1979, 1980; Dietrichs 1981a; Trott and Armstrong 1987; Trott et al. 1990; Armstrong 1990) have focussed on the parts of their cortical zones within the anterior lobe, although substantial parts lie in medial parts of the crura and in the paramedian lobule (PML). We have recently shown, however, in complete accord with the O-C-N hypothesis (Trott et al. 1997), that the parts of the pars anterior (pa) and pars copularis (pc) divisions of PML that belong to the cortical zone of the C1 compartment, provide C-N projections directed overwhelmingly (perhaps exclusively) to nucleus interpositus anterior (NIA), the deep nucleus that is also the C-N target of the C1 zone in the anterior lobe. However, we found also that, even apart from its known division (see for example Oscarsson 1969; Cooke et al. 1972; Armstrong et al. 1973; Inui 1989) into a ªforelimbº region (which includes the parts in lobule V and rostral PML) and a ªhindlimbº region (in lobule IV and caudal PML), the C1 zone is not homogeneous in its connections throughout its length because the part in PML pa differs conspicuously from the parts in lobule V and PML pc in receiving nucleo-cortical (N-C) afferents; moreover, the majority of these originate not from NIA but from nucleus interpositus posterior (NIP) which belongs to the C2 compartment. Because of these findings and because of a relative paucity of evidence (see Discussion) regarding its C-N projections we report here a study in the cat of the neighbouring C2 zone in PML pa and in the pars posterior part of PML (PML pp). The approach adopted is identical to that used by Trott et al. (1997): electrophysiological identification of the zone was followed by its injection with small amounts of wheatgerm agglutinin±horseradish peroxidase (WGA-HRP) and subsequent study of the distributions in the deep nuclei of orthogradely labelled C-N terminals and retrogradely labelled N-C cells. In lobule V we have previously shown (Trott et al. 1990) that the C-N projection of the C2 zone is to NIP and that this nucleus is overwhelmingly the source of N-C afferents to the zone. Here, we have sought to determine whether the same is true for PML parts of the zone. Because the C2 zone receives its climbing fibres only from the rostral part of the medial accessory olive (rMAO; e.g. Voogd and BigarØ 1980; Trott and Apps 1993) the possibility of spread of tracer beyond the confines of the zone was controlled for by determining the distribution of retrogradely labelled neurones in the contralateral inferior olive (cf. Trott et al. 1997). Finally, other studies have shown that the B and C3 (and probably also the C1) complexes are each divided in-

to a number of smaller (and perhaps equally fundamental) subunits: in their cortical zones longitudinal subzones or ªmicrozonesº have been convincingly demonstrated. These differ in the somatic afferent receptive fields of their climbing fibres and there is evidence that their Purkinje cells project to distinct subgroups of deep nuclear neurones (Andersson and Oscarsson 1978; Ekerot and Larson 1979a, b; Ekerot et al. 1991; Garwicz and Ekerot 1994; Garwicz et al. 1996). However, the C2 zone differs in that its main spino-olivocerebellar path (the lateral funiculus path; LF-SOCP) exhibits very substantial peripheral convergence (cf. Larson et al. 1969; Cooke et al. 1972; Inui 1989) and the existence of subzones has not to date been demonstrated. Nevertheless, if subzones exist it is possible that WGA-HRP injection sites involving different ªwidthsº within the zone might generate different distributions of C-N terminal labelling in the target deep nucleus. We have therefore also sought to test this prediction.

Materials and methods Injections of the bidirectional axonal transport tracer substance WGA-HRP were made into the C2 cerebellar cortical zone in PML using methods identical to those used and described in detail by Trott and Apps (1993; see also Trott et al. 1990, 1997). Description is therefore confined to the most essential points. Operative procedures and electrophysiology Nine purpose-bred adult cats weighing between 2.5 and 5.5 kg were used. In each an aseptic operation was carried out with full general anaesthesia induced and maintained with sodium pentobarbitone (Sagatal, BDH; 40 mg/kg plus maintenance doses). Small craniotomies were made over one or both paramedian lobules, the dura mater was incised and reflected and the positions of the borders between the cortical zones were determined electrophysiologically. Percutaneous electrodes were used to deliver brief (0.1 ms) rectangular electrical pulses to both ipsilateral and both contralateral limbs and the resultant field potentials mediated via climbing fibres (and therefore via the SOCPs) were mapped on the surface of the PML folia using glasscoated tungsten microelectrodes, as described and illustrated by Trott and Apps (1993; see their Fig. 1) and also by Trott et al. (1997; their Fig. 1). Use of the same technique in lobule V of the anterior lobe has been described by Trott et al. (1990; see their Figs. 2±4). The C2 zone is characterized by the appearance of SOCP responses that are evoked from all four limbs owing to extensive peripheral convergence in the main path (LF-SOCP) that targets the zone (cf. Larson et al. 1969; Inui 1989). By contrast, the C1 and the C3 and D1 zones receive input mediated from only the ipsilateral limbs. In all cases responses indicative of both the C1 and C2 zones were found and the border between them was taken as occurring at the point across the width of the folium (i.e. at the point along its long axis) at which their responses were of equal amplitude (cf. Trott and Apps 1993). At such points both responses were of course much smaller than in the middle of the zones. In some cases in PML pa, C3 zone responses were also present so that the lateral boundary of the C2 zone could be localized in the same way as the medial. However, in other pa cases and all pp cases such responses were not detectable. In some of these cases (e.g. 34R, 35R and 23R) the C2 zone was flanked laterally by a zone displaying ipsilateral responses evoked at a latency on average 5 ms longer than the C1 zone responses in the same folium. These responses were evoked from the fore- or hindlimb depending on the folium studied and were taken to indicate the presence of the D1 zone (Trott and Apps 1993). In such cases the width of the C2 zone may have been

318 slightly over-estimated because of the possible presence, between the C2 and D1 zones, of a C3 zone that could not be revealed either because it was very narrow or (less probably) because the peripheral stimuli failed to activate its SOCPs. In other cases no responses typical of either the C3 or the D1 zone could be detected lateral to the C2 zone and in such cases the lateral border of the zone was taken to lie at the point where C2 responses became too small to be detected. One folium was chosen for hydraulic injection into the centre of the C2 zone of a small volume (15±30 nl) of 1±2% WGA-HRP delivered 0.5 mm below the pial surface with a glass micropipette. During the histological investigation (see below) the location of the injected folium was determined relative to the rostro-caudal division of PML into pars anterior, pars posterior and pars copularis (pa, pp and pc respectively). All injections were into folia belonging to pa or pp; pc was not investigated. In one case (37L) the injected folium belonged morphologically to pp but in terms of connectivitiy it appeared appropriate to group it with the pa cases (see Results and Discussion). Note that although all C2 zone recording loci, whether in pa or pp, exhibited responses to stimulation of all four limbs, C1 responses were evoked in PML pa only from the ipsilateral forelimb (because the part of the C1 zone in rostral PML is ªforelimb-relatedº; Trott et al. 1997 and cf. Cooke et al. 1972; Inui 1989). In some PML pp folia C1 responses were similarly elicited only from the forelimb but in others only from the hindlimb. This is because it is within PML pp that the transition occurs to the hindlimb region of the zone that is represented in the caudal part of PML (cf. Cooke et al. 1972; Inui 1989). Histological procedures After 66±94 h survival each animal was deeply anaesthetized and perfuse-fixed. Cerebellum and medulla oblongata were removed, blocked separately and frozen sectioned at 50 m, the former sagittally, the latter coronally. Two series of alternate sections were prepared: one stained with cresyl violet and the other processed histochemically for HRP as described by Mesulam (1982). Bright- and dark-field light microscopy was employed to define all injection sites and the distribution patterns in the deep cerebellar nuclei of C-N terminals and N-C cell bodies HRP-labelled as a result of respectively anterograde axonal transport by Purkinje cells and retrograde transport by N-C neurones. Table 1 Injection site details for the 11 injections of wheatgerm agglutinin±horseradish peroxidase (WGA-HRP) into the cortex of paramedian lobule (PML) in the present study. Left column shows case identifier. Second and third columns show respectively the nanolitres of 1±2% WGA-HRP injected and hours survival time. Fourth column shows PML folium injected, with 1 as most rostral folium in the lobule; parentheses indicate another folium involved in the injection site due to injectate spread. Single asterisk indicates minor damage to folial surface near injection site. Double asterisk indicates injection was

The rostro-caudal and dorso-ventral extent of each site was measured and the medio-lateral (M-L) width was estimated from the number of sections in which labelled axons issuing from the site were present in the folial white matter at a level just deep to the locus of injection. Over much the greatest part of each site HRP was present co-extensively in the three cortical layers (cf. Trott et al. 1997). Patterns of C-N and N-C labelling were routinely represented on a series of standard diagrams representing the appearance in sagittal section of the cerebellar deep nuclei [nucleus lateralis (NL), NIA, NIP and nucleus fastigius (NF)], at 32 medio-lateral levels equally spaced 0.25 mm apart. These diagrams were prepared by averaging the nuclear outlines in six cases including some reported here (cf. Trott and Apps 1993). C-N labelling was represented by stippling and an outline was then drawn enclosing the stippled area(s). Note that although local variations in the density of labelled terminations were evident they were difficult to represent and in the illustrations no attempt has been made to do so (see Results and Discussion). NC labelling was shown on the same diagrams with each labelled cell body represented by a large dot. Note that because only alternate sections were used for histochemistry the true numbers of N-C neurones are underestimated, though presumably to the same degree in different cases. At each M-L level the area of each nucleus and the area(s) within it occupied by labelled C-N terminals were measured. This allowed numerical comparisons between different C-N projection territories (Fig. 3, Table 2) but it is emphasized that these are only semi-quantitative because variations in termination density were ignored (cf. Trott et al. 1997). The distribution of retrogradely labelled neurones in the contralateral inferior olive was also charted and estimates were made (Trott et al. 1997) of the relative sizes of the territories in which they were present in rMAO and in other olivary subnuclei [i.e. rostral dorsal accessory olive (rDAO) and principal olive (PO)].

Results This report is based on the histological findings in nine cats in which 11 injection sites were produced by making a small injection of WGA-HRP into a selected folium in most rostral folium of PML pars posterior (pp) but case was grouped as a PML pa case. Fifth and sixth columns respectively show C2 zone and injection site dimensions in the medio-lateral plane determined as in Materials and methods. Note folial damage in 37R precluded a zone width estimate though lateral funiculus path (LFSOCP) responses were recorded at (and on either side of) the injection site. Right column shows lobular and zonal location of the second (contralateral) sites present in some cases. Note the 37R and L and 23R and L pairs of cases each involved the same animal

Injectate volume (nl)

Survival time (h)

Folium injected

Width of C2 zone (mm)

Injection site width (mm)

Locus of contralateral site

Pars anterior 39L 30L 23L 37R* 55 34R 37L**

30 25 15 25 15 15±20 25

94 72 91 66 90 93 66

2 3 3 3 3 4 4

(+3) (+2) (+2)

1.0 1.1 1.4 ? 0.8 2.0 1.6

3.1 3.8 1.8 0.9 1.7 1.3 2.2

None lob. V C2 PML pp C2 PML pa C2 None None PML pa C2

Pars posterior 35R 21R 36R 23R

12±15 15 15 15

93 68 69 91

5 5 6 6

(+4) (+6) (+5) (+5,7)

2.0 0.9 1.4 1.3

1.8 2.4 1.5 1.5

None PML pa C1 None PML pa C2

Case

(+4)

319

Fig. 1A±G Distributions of cortico-nuclear (C-N) terminal and nucleo-cortical (N-C) retrograde labelling in nucleus interpositus anterior (NIA) and nucleus interpositus posterior (NIP) in the six PML pa cases plus case 37L. Each horizontal row shows the results from one case (see identifier). Figurine on the left of each row shows left paramedian lobule (PML) with wheatgerm agglutinin±horseradish peroxidase WGA±HRP injection site shown as the black area. Remaining outlines represent the ipsilateral cerebellar nuclei as they appear in sagittal sections spaced 0.5 mm apart, of which level 7 is the most lateral shown. Note calibration bar applies only to nuclei, not PML figurines; PML folia were usually 3.0 to 3.5 mm wide. C-N terminal labelling is represented by areas of stippling. Each large dot represents a retrogradely labelled N-C neurone. Note that in cases 39, 55 and 37L a few additional neurones present at more lateral levels are omitted for reasons of space; also that presentation is confined to alternate levels in the deep nuclear sequence. NL Nucleus lateralis

within the PML, in either pa or pp. Table 1 details the folia involved; also injectate volumes and survival times. Each injection was made into the centre of the C2 cortical zone after the positions of its lateral and medial boundaries had been determined electrophysiologically (see Materials and methods and also Trott et al. 1990, 1997 and Trott and Apps 1993). Table 1 shows the estimated width of the zone in each case and also, for comparison, the effective width of the corresponding injection site determined histologically (see Materials and methods). Note that R in six of the case identifiers in Table 1 indicates that in those cases the injection was into the right-side PML; however, in Figs. 1, 2 and 4 all PML figurines are shown left-sided to aid comparisons between cases. In the rostro-caudal plane and in depth all except two sites ranged in maximum size from 0.8 to 2.2 mm and from 0.8 to 2.0 mm respectively. The exceptions were cases 39L and 30L where the sites were larger (respectively 2.2

and 3.0 mm rostro-caudally and 2.5 and 2.6 mm in depth). All sites were largest in width (and rostro-caudal extent) at the cerebellar surface, presumably because of the shallow positioning of the pipette tip. In each case light microscopy was used (see Materials and methods) to map the distributions in the deep cerebellar nuclei of HRP labelling due to orthograde axonal transport in the C-N projections and retrograde transport in the N-C projections. In five animals the injection shown on the left in Table 1 was the only one made, but in the remaining four animals an injection was also made contralaterally at the location shown on the right in Table 1. These contralateral sites pose no interpretative difficulty in regard to the C-N projection because many previous studies have shown this to be strictly ipsilateral (cf. Trott et al. 1997). However, the N-C projections have sometimes been reported to include a minor crossed component so interpretation of the N-C labelling in the ªbilateralº cases includes consideration of a possible contribution from the contralateral sites (see Discussion). Six of the injections were placed into folia which were verified histologically as belonging to PML pa and the remaining five into PML pp folia. However, in Table 1 and elsewhere one PML pp case (case 37L) is grouped with the pa cases. For further comment see Discussion. Cortico-nuclear projections Pars anterior cases In Fig. 1 horizontal rows A±F each represent one of the six PML pa cases and row G represents case 37L. The fig-

320

Fig. 2A, B Overall distributions in NIA and NIP of C-N and N-C labelling as obtained by pooling cases. A The results of pooling the seven cases shown in Fig. 1. Black area in PML figurine represents the total extent of the lobule involved in the injection sites. Stippling and large dots at each level show respectively the total area of NIP and NIA ªoccupiedº by C-N terminals and the locations of N-C cells. Note that at levels 10 to 13 and 5 where cells were especially dense their spread is indicated but some are omitted to avoid overcrowding. B is similar to A but derives from the four cases shown individually in Fig. 4

urines to the left show the position and approximate extent of the injection site whilst the distribution of C-N labelling in the deep nuclei is shown by the outlined stippled areas superimposed on diagrams of the nuclei as they appear in sagittal sections 0.5 mm apart. Note that each case was initially mapped at 0.25 mm intervals but in the interests of compression only alternate (odd-numbered) levels have been shown in Fig. 1; also that presentation is confined to levels 7±17 in the complete sequence of 32 needed to cover the whole of the deep nuclear complex from the lateral pole of NL (level 1) to the medial pole of NF (level 32). NIP in fact extends from level 5 to level 20 and NIA from 9 to 20: at levels 7 and 8 NIA is fused with the dorsal part of NL (region NL/ NIA). The complete sequence of 32 levels is given in Fig. 1 of Trott et al. (1990).

In all seven cases a substantial C-N termination territory was found within NIP where it lay primarily in the dorsal half of the nucleus and frequently took the form of a crescent with the tips pointing ventrally. In three cases (37R, 34R and 37L; see Fig. 1D, F and G, respectively) C-N labelling was entirely confined to NIP. In the remaining four cases (39L, 30L, 23L and 55) there was also a small-

Fig. 3A, B The medio-lateral distributions in NIP of the C-N terminal labelling present in groups of cases involving tracer injections into different parts of PML. In A open circles show what percentage of the total C-N projection territory labelled by the injections of Fig. 1 and Fig. 2A is present at each medio-lateral level in NIP; filled circles are similar but for the PML pars posterior (pp) cases of Fig. 4 and Fig. 2B. B The percentage of NIP present at each medio-lateral level in the nucleus (crosses connected by broken lines). Open circles show the percentage of the nucleus at each level that was ªoccupiedº by the C-N terminal territories for the C2 zone in PML pars anterior (pa) (plus the cortex involved in case 37L); filled circles are similar but for the four PML pp cases shown in Fig. 4 and Fig. 2B

er territory in NIA, mainly in its caudalmost part. The labelling in NIA always involved lower densities of termination than in NIP and in two cases (39L and 30L) its extent in the medio-lateral, dorso-ventral and rostro-caudal planes was very small. Because each injection site was confined to one or two folia the cases were pooled with the intention of providing a closer approach to defining the total extent of the C-N projection from the C2 zone. The result is shown in Fig. 2A, where it is noticeable that the crescent form of the stippled C-N territory in NIP is maintained (or even at some levels enhanced), as is the marked size difference between the NIP and NIA territories. As described in Materials and methods, a semi-quantitative approach to the C-N projections was also adopted and the results relating to NIP are shown in Fig. 3. The open circles in Fig. 3A show the extent of the termination territory at each medio-lateral level where it was present, expressed as a percentage of the total pro-

321

31%). In total, the C-N territory embraced only 11% of the nucleus. Pars posterior cases

Fig. 4A±D Distribution of C-N terminal labelling and N-C labelling in four PML pp cases. Each row represents the case shown by the identifier. Conventions as in Fig. 1

jection; note that no less than 64% occurred between levels 9 to 16, i.e. in the middle part of the medio-lateral extent of NIP. Further information is provided by Fig. 3B, where the crosses show the proportion of NIP present at each level and the open circles show the percentage of the nucleus at each level that was ªoccupiedº by C-N terminals, which was c. 50% or more at each of levels 10 to 16. In all, the projection ªoccupiedº approximately 43% of NIP. Figure 3 does not include data for NIA but it may be noted that the occupancy exceeded 16% at two levels only (levels 10 and 11 where it was respectively 28% and Fig. 5A±E Distributions in NIP of the C-N projections arising from different portions of the C2 cortical zone. Note for reasons of space NIA is omitted and presentation confined to alternate sagittal levels. A Black areas represent the projection territory for lobule V b/c as deduced from cases (n=5) of Trott et al. (1990) in which WGAHRP was injected into the C2 zone. B Black areas show the extent of overlap between the CN projections arising from the parts of the C2 zone in PML pa and PML pp (cf. Fig. 2A and B). C as B but showing overlap between the PML pa and the lobule V projections. D as B and C but showing overlap between the PML pp and the lobule V projections. E Black areas show the overall C2 zone C-N projection territory in NIP deduced by pooling the PML pa, PML pp and lobule V territories

The C-N termination territories found for the injections into the C2 zone in PML pp (excluding case 37L shown in Fig. 1G) are shown in Fig. 4 in a format similar to that of Fig. 1 except that only levels between 13 and 19 are represented because in all cases the termination territories extended less far laterally in nucleus interpositus than in PML pa cases. In two cases (35R and 36R) C-N labelling was entirely confined to NIP while in the two remaining cases (21R and 23R) NIA also contained patches of terminal labelling. It was, however, lighter and confined to only three or two levels respectively. When the four cases of Fig. 4 were pooled Fig. 2B was obtained. In NIP the C-N territory at levels 12 (not shown) to 16 was, like that for PML pa cases, roughly crescent shaped and mainly in the dorsal half of the nucleus. When the semi-quantitative approach was applied it provided the results shown in Fig. 3A and B by filled circles. Fig. 3A further emphasizes that the C-N territory was restricted to the medial half of NIP and Fig. 3B shows that its percentage occupancy of the nucleus exceeded 50% at each of levels 15 to 19. Overall, the territory involved 31% of the total extent of NIP. By contrast, the percentage occupancy of NIA by C-N labelling (not shown in Fig. 3) was appreciable (cf. Fig. 2B) only at levels 17 and 18 (where it was 29% and 25% respectively). Overall, label was confined to only 4% of NIA.

322

Comparisons between the C-N projections to NIP from PML pa, PML pp and the C2 zone in lobule V Trott et al. (1990) established, using WGA-HRP as tracer, that the part of the C2 cortical zone in the tips of the folia of lobule V b/c in the anterior lobe projects to NIP. Therefore, for purposes of comparison, the findings from their five cases were pooled in the same way as for the present PML cases and the result is shown by the filled areas in Fig. 5A. Overall, the projection territory occupied 35% of NIP and it resembled the PML pa territory in being represented at most medio-lateral levels of the nucleus. Like both the PML territories it constituted at most levels a crescent-shaped area mainly in the dorsal half of the nucleus. It is obvious from comparison of Fig. 2A and B and Fig. 5A that each of the three C-N territories in NIP overlapped partially with the other two and the three areas of overlap are isolated for inspection in Fig. 5B, C and D (for Fig. 5E see Discussion). It is interesting to note that at most levels where an overlap occurred it formed a crescent not grossly different in size to the two projection territories under comparison. This implies that at such levels the two territories were rather similar in location and extent. When the overlaps were investigated semi-quantitatively (see Materials and methods) Table 2 was obtained. Note that overlap between the two PML territories was certainly not greater (indeed, slightly less) than between either of them and the lobule V C2 territory. Topographical localization within the projections from PML pa, PML pp and the C2 zone in lobule V Figure 1 shows that some PML pa cases differed quite substantially in regard to both the number of sagittal levels with C-N labelling in NIP and the position of those levels in the sequence. In general, larger sites labelled larger C-N territories but positional differences did not appear to be systematically related to the position of the injected folium within PML pa nor to differences in the degree of involvement of buried cortex. However, their existence is consistent with the presence of some form of topographical localization, presumably organized on some basis other than a ªfolialº or ªdeep versus superficialº one. Figure 4, by comparison, provides little evidence for any form of topography within the PML pp projection because the labelling patterns are rather consistent from case to case. This may, however, be because the injection sites were less diverse both in size and in regard to the position of the main folia involved (cf. Table 1). If C-N topographical localizations are present the most likely hypothesis is that they involve subdivision of the C2 zone into narrower ªlongitudinalº subzones (see Introduction) and in fact our present cases taken alone are not well suited to revealing such a pattern because most injection sites spanned much or all of the zone width.

Table 2 Extent of the spatial overlaps between the overall corticonuclear (C-N) termination territories in nucleus interpositus posterior (NIP) for the PML pars anterior (pa) and PML pars posterior (pp) groups of cases and the lobule V C2 group of Trott et al. (1990). For each overlap the Table shows the percentage of NIP in which it occurred and the percentage of each of the two C-N territories that fell within the other territory. Note that variations in the density of labelled terminations have not been taken into account; this approach to defining the potential extent of convergence between projections is therefore only semi-quantitative C-N territories (T) in NIP

Extent of T1/T2 overlap as a %

T1

T2

of NIP

of T1

of T2

PML pa PML pa PML pp

PML pp V C2 V C2

19 25 21

43 58 64

59 72 58

However, on the basis of injections into the C1 zone in PML pa and pc, Trott et al. (1997) concluded that the relatively light and spatially restricted NIP labelling present in some cases was attributable primarily (probably entirely) to spread of injectate laterally into the C2 zone. Such spread would presumably involve the medial part of the C2 zone but central and lateral parts would be less often and/or less heavily involved. It seemed, therefore, worthwhile to compare the distributions of the NIP label in those ªspreadº cases with those found here after injections directly into the zone (ªdirectº cases) at a similar rostrocaudal level. This has been done for PML pa, PML pp and lobule V respectively in rows A, B and C of Fig. 6, in each of which the overall C-N territory revealed by injections directly into the C2 zone is shown by vertical hatching whereas the ªspreadº labelling after C1 zone injections is shown by horizontal hatching. In Fig. 6A the ªspreadº labelling (cf. Fig. 4A of Trott et al. 1997) at most of the medio-lateral levels shown was confined to dorsal parts of the overall territory and the same was true at most of the alternate levels omitted for reasons of space. In addition, the overall territory involved levels 6 to 19 while the ªspreadº labelling was confined to levels 9 to 19. Moreover, of the four individual ªspreadº cases of Trott et al. (1997) used to produce Fig. 6A only one showed label lateral to level 11 while such labelling occurred in four of the seven ªdirectº cases (cf. Fig. 1). These findings are compatible with a topographical localization such that the medial part of the C2 zone in PML pa projects to the dorso-medial portion of the overall territory. In relation to PML pp it is unfortunate that Trott et al. (1997) made no injections into its (usually narrow) C1 zone, so it is possible only to compare the present PML pp cases with theirs involving C1 injections into PML pc folia and this poses an interpretative difficulty (see Discussion). Nevertheless, if comparison is made, as in Fig. 6B, between the present PML pp cases and the PML pc cases of Trott et al. (1997; cf. their Fig. 4B), it emerges that NIP labelling occurred in the ªdirectº (i.e. the present) cases at levels 12 to 19 whereas in the ªspreadº cases it was

323 If NIP labelling after C1 zone injections is indeed attributable to injectate spread to the C2 zone it would be predicted it should lie within the C-N projection areas defined by injections directly into that zone. Overwhelmingly, this prediction was fulfilled.

Finally, Trott et al. (1990) made a number (n=10) of WGA-HRP injections into the C3 zone in lobule V and four of these yielded ªsubsidiaryº C-N labelling in NIP. This occurred between levels 11 and 16 (i.e. at middle levels) where it heavily overlapped the lobule V C2 territory, especially its dorsal part. See Discussion for comment. Localization within the C-N projections from C1 zone to NIA

Fig. 6A±F Comparisons of C-N labelling in different groups of cases. A, B and C Overall labelling in NIP after tracer injections directly into the C2 zone (vertical hatching) and after injections into the C1 zone in which there was injectate spread into the C2, presumably mainly its medial part (horizontal hatching). In A and C both the ªdirectº and the ªspreadº cases involved respectively PML pa and lobule V. In B C2 injections in PML pp are compared with spread cases involving nearby folia of PML pc. D, E and F as A, B and C but showing label in NIA from direct injections into the C1 zone (vertical hatching) and injections in the C2 zone in which there was spread into the C1 zone, presumably mainly its lateral part (horizontal hatching). Data are from the present cases and from Trott et al. (1990, 1997)

confined to levels 15 to 19. Moreover, although at levels 18 and 19 both territories were complex and only partially overlapping, at levels 15, 16 and 17 ªspreadº case labelling was very largely restricted to the dorsal part of the territory for the ªdirectº cases. Therefore, overall, this comparison is also compatible with projection of the medial part of the C2 zone to the dorso-medial part of the overall NIP territory for the zone in PML pp. In light of the above findings Fig. 6C shows a parallel comparison made for the pooled lobule V cases of Trott et al. (1990) and it is clear that NIP labelling ascribable to ªspreadº in their C1 cases (n=3) extended much less far laterally and at levels 15 and 17 (and 16; not shown) it was also restricted more dorsally than in the ªdirectº cases involving C2 zone injections (n=5). Again, therefore, the findings are compatible with projection of the medial part of the C2 zone to the dorso-medial portion of the overall territory in NIP, for the zone in lobule V.

Some cases led to subsidiary C-N labelling in NIA (Figs. 1, 2 and 4) and it is highly probable (see Discussion) that most, perhaps all, reflected transport resulting from spread of injectate beyond the C2 zone. In Fig. 6D this NIA labelling for the pooled PML pa cases is represented by horizontal hatching while the vertical hatching shows for comparison the pooled findings from the direct C1 injections in PML pa reported by Trott et al. (1997). Clearly, much of the ªspreadº labelling was within the C1 zone C-N territory and confined to its caudo-lateral part. Figure 6E compares in the same way the present PML pp cases and the C1 injections into PML pc reported by Trott et al. (1997); the ªspreadº labelling was again within the C1 territory in NIA and again confined caudally. Finally, Fig. 6F relates to data from the lobule V case material of Trott et al. (1990) and compares the NIA labelling after injections directly into the C1 zone with that found at the same medio-lateral levels in those C2 cases in which spread occurred. Here, the ªspreadº label lay mainly in the dorso-lateral part of the C1 territory. For comment see Discussion. Nucleo-cortical projections As may be seen from Figs. 1, 2 and 4 retrogradely labelled cell bodies of N-C neurones were found ipsilateral to all 11 injection sites (though in case 21R only one cell was present). The number per case and their distribution between the different deep nuclei are given in Table 3. Their pooled distributions within the profiles of the different deep nuclei in sagittal sections are shown for the PML pa and pp groups of cases in Fig. 2A and B respectively and their medio-lateral distributions are represented graphically in Fig. 7A and B. A few cells were found in NIP in case 34R but artefactual clumps of reaction product prevented an accurate count (but did not obscure the C-N labelling). Table 3 shows that except in case 37R where the injection site was unusually small, the number of neurones was substantially greater in PML pa than in PML pp cases. However, in 37R there was minor accidental damage to

324 Table 3 Numbers of retrogradely labelled nucleo-cortical (N-C) neurones found ipsilateral to the injection site in nucleus interpositus posterior (NIP), nucleus interpositus anterior (NIA), nucleus lateralis (NL) and the region of NIA/NL fusion at medio-lateral levels 7 and 8. Separate totals and mean cells per case given for PML pa and PML pp groups of cases; note means are quoted to the nearest integer. In 34R cells were found in NIP but histochemical artefact prevented reliable counting Case

NIP

Pars anterior 39L 30L 23L 37R 55 34R 37L Total Mean

98 5 0 6 50 7 0 0 70 19 2 2 10 0 0 0 41 2 3 10 Present but counting not feasible 73 1 1 5 342 34 6 23 57 6 1 4

Pars posterior 35R 21R 36R 23R Total Mean

14 1 15 11 41 10

NIA

0 0 0 0 0 0

NIA/NL

0 0 0 0 0 0

NL

0 0 1 0 1 0

Total 109 57 93 10 56 80 405 68 14 1 16 11 42 10

the folial surface near the injection site and in view of the restricted C-N labelling it is possible that orthograde and retrograde axonal transport were both somewhat impaired. Pars anterior cases In view of our previous demonstration (Trott et al. 1990) that the part of the C2 zone in lobule V of the anterior lobe receives a N-C projection from NIP we were not surprised to find labelled cells in NIP. The average number per case (n=57; range 10±98) was in fact even higher than for lobule V (where n=35). Figure 7A shows that cells were present across most of the width of NIP but they were most numerous at middle levels so that no less than 83% were at levels 9 to 14. Figures 1 and 2A show that the majority of cells were in the dorsal half of NIP and also that in each individual case most lay within the C-N projection territory labelled in that case. Overall, no less than 74% were so located though that includes 11% that were at the edge of the territory. When the distributions were compared with the CN territory defined for PML pa by pooling the cases the proportion within that territory was even larger (92%). The presence of labelled N-C neurones in NIA was more surprising as such neurones were non-existent in lobule V cases (Trott et al. 1990). They were found in all cases except 37R and the average number per case was 6 (range 0±19). They were in the lateral third of NIA (Fig. 7), both dorsally and ventrally in its caudal half. They were fewer than were found after C1 injections in PML pa (cf. Trott et al. 1997) and the issue as to whether

Fig. 7A, B Medio-lateral distributions of N-C neurones retrogradely labelled after injections in the C2 cerebellar cortical zone. A The distributions in NIP (open circles), NIA (filled circles) and NL (crosses) of 405 neurones labelled after injections in PML pa plus the injection in case 37L. Note that the medio-lateral distribution of each symbol indicates the medio-lateral extent of the corresponding deep nucleus; also that levels 7 and 8 in NIA represent the NL/NIA transition region. B The medio-lateral distribution in NIP of cells labelled in the PML pp cases of Figs. 4 and 2B (open circles) and of cells labelled after C2 zone injections in lobule V (filled circles; derived from data of Trott et al. 1990)

their presence resulted from transport of injectate that had spread to the C1 zone is dealt with in Discussion. An average of five cells per case was also found in NL and the junctional region where NIA and NL are fused, mostly near the caudal border of the neuropil. Pars posterior cases Figures 2B and 4 and Table 3 show that with a single exception (in NL) all labelled cells in the PML pp cases were in NIP. Like the NIP cells in pa cases, most were in the dorsal half of the nucleus, but although the distributions overlapped medio-laterally, Fig. 7B (open circles) shows that they were restricted to fewer sagittal levels and were most numerous at a level (16) close to the medial edge of the PML pa population (cf. open circles in Fig. 7A). Of these cells, 83% were within the C-N territo-

325

ry labelled in NIP in the same case and 88% were within that derived by pooling the cases. In Fig. 7B the filled circles show the distribution of the numerous NC cells labelled in NIP after C2 zone injections in lobule V (data of Trott et al. 1990). Like those from the present PML pa cases they were present (mainly in the dorsal half of the nucleus) at all except the most lateral levels; however, the peak of the distribution is somewhat more medially placed.

Distribution of retrogradely labelled olivary neurones The estimates made of site width and zone width in each case provide an indication of the extent to which the sites are likely to be confined to the C2 zone but further evidence was provided by the pattern of retrograde labelling in the contralateral inferior olive. This pattern has been illustrated in detail for cases 39L and 37R by Trott and Apps (1993) and a more compressed presentation is provided for all 11 cases by Table 4. In five cases label was confined to the part of rMAO which projects to the C2 zone. Among the six remaining cases between 20% (case 21R) and 54% (case 39L) of the olivary labelling was outside rMAO. In four of these it was confined or nearly confined to rDAO, indicating spread of tracer to the C1 and/or the C3 zone. In two cases (39L and 55) in addition to labelling in rMAO and rDAO, there was substantial labelling (respectively 30% and 39% of the total) in PO which innervates the D zones. For further comment see Discussion. In lobule V of the anterior lobe the C1 zone is divided into a medial and a lateral portion; both project to NIA (Trott and Armstrong 1987) but the ªlateral C1º zone (the CX zone of Campbell and Armstrong 1985) receives its climbing fibres from a ªcolumnº of olive cells medial in MAO between stereotaxic levels P 10.5 and 12 (Trott and Apps 1991). However, Trott and Apps (1993) showed that in rostral PML the whole width of the C1 zone receives its climbing fibres from rDAO (implying absence of a ªlateral C1º zone). In all the present cases all the labelled cells in MAO were in the more lateral parts of rMAO that innervate the C2 zone and none was in the ªlateral C1º-related region.

Discussion Cortico-nuclear projections NIP as the deep nuclear target of the C2 zone Table 4 shows that in five cases olivary retrograde labelling was confined to the part of rMAO than projects to the C2 zone and in all five C-N labelling was entirely restricted to NIP. In three of these cases the injection site was narrower than the zone (cases 34R, 35R) or little wider (by 0.1 mm; case 36R). In case 37R zone width could not be determined but the site width was 0.9 mm and only in one other case was the zone narrower than this (by 0.1 mm in case 55; see Table 1); therefore, in this case also it is highly probable that site and zone were the same or almost the same in width. Thus, only in case 37L was the site appreciably wider than the zone (by 0.6 mm).

Table 4 Injection site width relative to zone width and size of C-N termination territory in NIP and of territory in the rostral part of the medial accessory olive (rMAO) containing retrogradely labelled olive cells, expressed as a percentage of the total C-N and olivary territories in the same case. Note negative widths indicate the medio-lateral extent of the site was narrower than the zone. Note more details of olivary labelling in cases 39L and 37R are available in Trott and Apps (1993) Case

Width: siteŸzone (mm)

C-N in NIP (% of total)

Olive label in rMAO (% of total)

Pars anterior 39L 30L 23L 37R 55 34R 37L

2.1 2.7 0.4 ?0 0.9 Ÿ0.7 0.6

90 84 70 100 66 100 100

46 51 55 100 47 100 100

Pars copularis 35R 21R 36R 23R

Ÿ0.2 1.5 0.1 0.2

100 80 100 95

100 80 100 73

By contrast, in the remaining six cases subsidiary C-N labelling was present outside NIP in amounts ranging from 5% to 34% of the total (Table 4) and it is necessary to consider whether this constitutes evidence for additional C-N projections from the C2 zone. However, in all these cases there was olivary labelling outside rMAO in amounts ranging from 20% to 54% and site width exceeded zone width by from 0.2 to 2.7 mm. Except in two cases (39L and 55) all or virtually all the additional olivary labelling was in rDAO, consistent with tracer spread to the C1 and/or the C3 zone, and the subsidiary C-N labelling was in NIA. In cases 39L and 55 respectively 30% and 39% of olivary labelling was in PO and 2% and 17% of C-N labelling was in NL. These two cases were in PML pa where the C3 zone is absent or narrow (cf. for example Materials and methods; Trott and Apps 1993) and the injection sites were substantially wider than the C2 zone (by 2.1 and 0.9 mm respectively); the apparent involvement of the D zones in the injection sites is therefore not surprising. Finally, Fig. 8 provides, for the present cases taken as a whole and for the cases of Trott et al. (1997), strong evidence linking NIP and the C2 zone in PML: clearly the nearer an injection site came to being confined to the zone the more closely the C-N labelling approached being restricted to NIP. Overall, therefore, we conclude that some of our cases show unequivocally that some (quite extensive) parts of the C2 zone project only to NIP and that our body of cases is entirely consistent with the hypothesis that NIP is the sole target for the C2 zone in PML pa and pp. This conclusion is of course in line with the pattern of the O-C-N compartments as deduced by Voogd (Voogd 1964, 1969; Voogd and BigarØ 1980) and others, including ourselves (e.g. Trott and Armstrong 1987; Trott et al. 1990). However, ours is the first study to benefit from

326

Fig. 8 Relationship between the proportion of C-N labelling that was in NIP and the proportion of olivary labelling that was in the rostral part of the medial accessory olive (rMAO) after WGA-HRP injections in PML. Filled circles represent the present 11 cases; open circles represent 12 cases of Trott et al. (1997) involving injections into the C1 zone in PML pa or pars copularis (pc). Note that in nine of the latter there was some olivary labelling in rMAO (indicative of injectate spread to the C2 zone) and in six of those there was some C-N label in NIP. Filled circle at top right (100,100) represents five cases. Broken line is diagonal line of equality. Unbroken line shows result of a least-squares linear regression. A close correlation is evident (r2=0.79). An F test yields P