Postnatal Development of the Cerebellar Cortex in the Rat IV. SPATIAL ORGANIZATION OF BIPOLAR CELLS, PARALLEL FIBERS
AND GLIAL PALISADES JOSEPH ALTMAN Laboratory of Developmental Neurobiology, D e p a r t m e n t of Biological Sciences, Purdue University, Lafayette, Indictnu 47907
ABSTRACT The ontogeny of the spatial organization of some components of the molecular layer was investigated in cerebella sectioned systematically in the sagittal, coronal and horizontal planes. There is no discernible organization in the distribution of cells of the proliferative zone of the external germinal layer (EGL) but from birth the differentiating bipolar cells of the subproliferative zone are aligned parallel to the surface and to the long axis of the folium. While they are still in or at the base of the EGL, the bipolar cells emit long processes, the future parallel fibers. The next step is the outgrowth of a vertical process which may reach the base of the molecular layer before the granule cell nucleus becomes translocated. The idea that the cell body truly migrates through the molecular layer is not supported by the observations. BergmaM glia cells are frequently seen in Golgi material in neonates but they are probably less numerous than i n older infants and their processes are not as well aligned. It is only gradually that the EGL is perforated by glial endfeet which in older infants are occasionally organized into longitudinal rows. I n mature cerebella the parallel fibers are separated by thin and relatively narrow, unstained spaces which are oriented in the longitudinal plane and can be traced from the pial surface to a zone just above the layer of Purkinje cells. It is postulated that these spaces are occupied by glial palisades formed by apposed thin vertical processes to which many Bergmann glia cells contribute. The alignment of these palisades is dependent on the orientation of parallel fibers. When the parallel fibers are reoriented by X-irradiation the glial palisades become correspondingly realigned. These observations indicate that the oriented growth of parallel fibers, which follows the polarization of bipolar cells, determines the spatial organization of the glial framework of the molecular layer. They also suggest that the glial palisades mediate functions that are not primarily developmental in nature.
The cells and cell processes of the cerebellar cortex form a regular three-dimensional lattice-work which is apparently uniform throughout and which, with minor modifications and few exceptions, is a characteristic feature of the cerebella of all higher vertebrates. The present paper and its companion (Altman and Winfree, in prep.) are concerned with aspects of the ontogeny of this geometric organization coupled with the attempt to identify some of the mechanisms involved. The geometric organization of the supraganglionic cerebellar cortex of the rat develops postnatally (Ramon y Cajal, '11, J.
COMP. NEUR.,163: 427-448.
'60). The perikarya of Purkinje cells disperse into a monolayer several days after birth and the development of their planar arbors in the molecular layer starts during the second week (Addison, '11; Altman, '72b). Cells multiplying in the external germinal layer differentiate as basket cells during the second week and most of the granule cells are formed during the second and third week (Altman, '69, '72a,b,c). These studies of normally developing cerebella and subsequent experimental work (Altman and Anderson, '72; Altman, '73a,b) have suggested that some properties of cells of the external germinal 427
428
JOSEPH ALTMAN
layer and their initial mode of morphogenic differentiation play a major role in the subsequent development of cortical geometry. While Purkinje cell perikarya display considerable autonomy and develop normally in size when the formation of granule and basket cells is prevented by X-irradiation (which kills selectively the multiplying germinal cells) the Purkinje cells do not disperse in the absence of parallel fibers, their primary and secondary dendrites become randomly oriented and they fail to develop tertiary and higher order spiny branchlets (Altman and Anderson, '72). If the orientation of parallel fibers is changed, which occurs after regeneration of the external germinal layer consequent to irradiation with certain schedules (Altman, '73b) the planar Purkinje cell dendritic plexus becomes rotated at a right angle to the pile of parallel fibers. We will attempt to show in this and the succeeding paper that the orientation of bipolar cells of the external germinal layer determines not only the orientation of parallel fibers but also the geometry of several other elements of the molecular layer. Special attention is paid in this paper to the radial glial processes that were postulated to guide the descending granule cells (Rakic, '71; Rakic and Sidman, '73). Our observations indicate that the differen tiating granule cells are translocated into the granular layer, following their descending vertical processes that find their way there. The organization of glial processes, which assemble into longitudinally aligned vertical palisades, occurs after, not before, the alignment of parallel fibers. In the subsequent paper the dispersion of Purkinje cells into a monolayer, the alignment of Purkinje cell dendrites in the transverse longitudinal plane and the alignment of the proximal and distal axons of basket cells are described as events that are secondary to the geometric organization of the parallel fiber system.
10, 12, 14, 16, 18 and 20 days were embedded in paraffin and cut serially at 10 mp in the coronal, sagittal and horizontal planes. The sections were stained in a systematic sequence with hematoxylin-eosin, cresyl violet, the Weigert-Loyez and Bodian's protargol-S methods. The cerebella of three adult cats were cut similarly in the three planes and every section was saved to allow three-dimensional reconstructions. Finally brains from rats aged 0, 1, 2, 3, 4, 7, 10, 15, 21 and 28 days, and from adults, were impregnated with the rapid Golgi and Golgi-Cox techniques. The irradiated cerebella came f?om the collection that was described earlier (Altman, '73a). RESULTS
1. T h e external germinal layer In the external germinal layer proliferative cells are localized in the superficial, or subpial zone. Some of these cells show mitotic activity (fig. 1A) and many of them are labelled when infant rats are injected with thymidine-H3 and are killed within 1-2 hours thereafter (Altman, '66, '72a). When the distribution of these cells is examined in sections cut parallel to the pial surface (fig. 1A) the tightly packed cells do not display any obvious geometric organization. In older infants circular cellfree regions may be seen among the cells; these are presumed to be the end-feet of Bergmann glia processes that reach the pia. Some of the cells appear to form rosettes around these processes but there is generally little change in the essentially random distribution of the tightly packed proliferative cells. The deeper zone of the external germinal layer is composed of nonproliferative cells which appear labelled only when at least six hours elapsed between injection of thymidine-Hs and the death of the animal (Altman, '72a). These bipolar cells form a thin zone in neonates, then increase rapidly to form a layer up to ten cells thick by about ten days, and decline MATERIALS A N D METHODS gradually thereafter. But from the very In addition to the extensive collection beginning these bipolar cells are oriented of normal developmental material that was regularly parallel to the long axis of the described in detail earlier (Altman, '72a) folium (fig. 1B). In the vermis they appear in this study use was made of some addi- accordingly as spindle-shaped cells in corotional material. The brains of male Pur- nal sections, and as small round or ovoid due-Wistar rats aged 0, l, 2, 3, 4, 5, 6, 8, cells in sagittal sections. This regular or-
DEVELOPING CEREBELLUM. I V
429
Fig. 1 Tangentially cut sections through the external germinal layer. X 288. Hematoxylineosin. A. Section through the proliferative zone in the vermis of a 2-day-old rat. Arrows point to some of the mitotic cells. B. Section through the subproliferative zone i n the vermis of a 3 - d a y d d rat. C. Similar to B from a n 8-day-old rat. Note the presumed perforating glial endfeet which are occasionally aligned into longitudinal rows (arrows).
ganization prevails as long as the external germinal layer is present (about 21 days of age) and the only developmental change that can be seen is the perforation of the layer by presumed processes of Bergmann’s glia cells (fig. 1C) which are occasionally contiguous with each other and form a row in the longitudinal plane. In some sagittal sections the transversely cut bipolar cells are aligned vertically. In many sagittal sections this is only vague-
ly recognizable or is not evident at all; it is assumed that the horizontal alignment is caused by the vertically oriented thin Bergmann glial processes and is seen only when the section is cut exactly in the same plane as the ascending glial processes. In well-stained sections the just-resolvable, transversely-cut parallel fibers of the molecular layer may be absent in vertical strips that are continuous with the cellfree vertical columns. Such a vertical align-
4 30
JOSEPH A1,TMAN
Fig. 2 Bergmann glia cells in 2-day-old ( B ) and 3-day-old (A and C ) rats; egl, external germinal layer. X 288. Golgi-Cox.
ment of cells is never seen in sections cut parallel to the long axis of the folium. Bergmann glia cells with ascending vertical processes are seen in neonates and are, indeed, the most frequently impregnated cellular elements in Golgi sections (fig. 2). Typically the cell body is located between or underneath Purkinje cells, its thin processes branch below the external germinal layer, traverse this region and terminate with end feet in the pia (where they may be impregnated jointly with the pia or blood vessels). Generally the branching of glial processes is richer in the sagittal plane (fig. 2B,C) than in the coronal plane (fig. 2A) but exceptions are seen.
nal plane. I t was not possible to establish with any certainty whether or not, or how often, the growing polar tips take the shape of growth cones.
3 . The translocating g r a n u l e cells Presumably after the horizontal portion of the future granule cell axon has reached a certain length, the formation of its vertical portion begins. It is generally thought that this is accomplished by migration of the granule cell through the molecular layer. The evidence provided by the Golgi technique is ambiguous. Often one sees leading perikarya with trailing processes, in other instances (fig. 4) shorter or longer descending processes precede the peri2. The transitional molecular layer karya. The outgrowing process (or procThe transitional molecular layer is the esses, figs. 4B,C) may initially have a zone composed of differentiating cells and somewhat uncertain orientation but as the their processes at the base of the external leading process becomes longer it unerrgerminal layer and the (temporary) ceil- ingly assumes a vertical orientation with ing of the molecular layer (Altman, ’72a). respect to the surface of the cortex. I t is We postulated that the spindle-shaped cells of some theoretical importance whether in this cortical sheet are not migrating the perikaryon or the vertical process is but rather have this shape because they in leading position; the former implies “extrude” cytoplasm at the two poles. cellular migration, and possibly mechaniThese become the horizontal or distal por- cal guidance (as by Bergman glial proctions of the future parallel fibers. That, esses; Rakic, ’71) while if the latter is the indeed, the horizontal fibers may acquire case nonmechanical guidance is a more considerable length before the differenti- parsimonious assumption and the nucleus ating granule cell descends through the may be thought of as translocating rather molecular layer is illustrated in Golgi-im- than migrating, as in the case of the interpregnated material in figure 3. A t this kinetic migration of neuroepithelial cells stage of their growth the horizontal fibers (Sauer, ’35). Examination of this problem tend to be somewhat kinky, more so when in favorably cut and strongly stained (heviewed in the horizontal than in the coro- matoxylin-eosin) sections at high magni-
431
DEVELOPING CEREBELLUM. IV
0
R
X 1
4
432
JOSEPH A L.T M AN
fication (figs. %13) indicate that the translocation of the differentiating granule cell from the transitional molecular layer into the granular layer is guided by the outgrowing vertical process. Cells are seen with pear-shaped perikarya a t the base of the external germinal layer whose single descending processes reach as far as the layer of Purkinje cells. In other instances oval or slender spindle shaped perikarya are “caught” a t different points in their descent through the molecular layer (figs. 11-13). This staining technique, unlike the Golgi procedure, tends to stain preferentially the leading portion of the vertical fiber which sometimes appear much thicker than the differentiated (trailing) parallel fiber. Perhaps the leading process is rich in cytoplasm, as i t was indicated earlier by electron microscopic observations (Altman, ’72c, fig. 4). 4. The glial palisades
We proposed earlier that the perpendicular columnar alignment of bipolar cells
seen in favorably cut transverse sections is produced by the ascending processes of Bergmann glia cells. In sections cut horizontally through the sheet formed by bipolar cells the perforating glial processes and their endfeet were occasionally lined up i n rows parallel to the long axis of the folium, or the orientation of bipolar cells (fig. 1C). This suggested that the vertical glial fibers are apposed and form longitudinally aligned vertical palisades. I n the cerebella of one week or older rats these vertical columns could sometimes be traced through the molecular layer as far as the layer of Purkinje cells; they were most apparent in brittle sagittal sections in which the tissue tended to fracture along these presumed columns. In mature cerebella these vertical columns were more obvious. The organization of these columns was therefore examined in cerebella from three adult rats that were cut serially in the horizontal, sagittal and coronal planes and were systematically stained with different techniques, as it was described ear-
Fig. 4 Transforming bipolar cells with vertical processes of different length, presumably displaying progressive phases of development. All from the same 10-day-old rat. X 720. Golgi-Cox.
DEVELOPING CEREBELLUM. I V
lier. The data described below were largely obtained by examining 10 p-thick sections prepared with Bodian’s protargol-S method. This procedure tends to tint lightly the parallel fibers. In sagittally cut sections of the vermis in which the parallel fibers appear as small dots (that is, are cut exactly at a right angle to the beam of parallel fibers) unstained “gaps” 2-4 mp wide extend from the pial surface to the proximity of the layer of Purkinje cells (fig. 14). Since these spaces are free of transversely cut parallel fibers (vertical fibers are sometimes seen coursing through them) it is concluded that they have the shape of vertical plates which are thin in the transverse plane of the folium and are at least 10 p wide in the longitudinal plane parallel to the alignment of the horizontal beams of parallel fibers. In coronal sections of the vermis, in which the parallel fibers are cut parallel or obliquely, these unstained vertical gaps are not seen (fig. 15). This is com-
433
patible with the idea that they are thin in this plane and would be obscured by parallel fibers situated below and/or above them. Finally in sections cut horizontally through the molecular layer the unstained spaces were identified as rows composed of a variable number of closely apposed circular or ovoid elements (fig. 5). These could be traced from the surface of the molecular layer to its base where basket cell axons impregnated by the Bodian technique are oriented at a right angle to these spaces and the beams of parallel fibers (figs. 1€L 18). Evidently these gaps are made up of long vertical fibers that form palisades in the longitudinal plane. It is postulated that the fibers are those of Bergmann glia cells. Bignami and Dahl (’74a) observed the alignment of Bergmann glial cells in rows in tangential sections (their fig. 6 ) . Since the fibers of a single Bergmann glia cell tend to fan out (fig. 2) it is assumed that the contiguous fibers originate fkom several cells. Moreover, considering that
Fig. 5 Molecular layer in horizontal section. Note the longitudinally oriented spaces which are apparently made up of many apposed circular and ovoid elements (arrows). Adult rat. X 1123.2. Bodian.
434
JOSEPH ALTMAN
the spread of glial fibers is more pronounced in the transverse (sagittal) than longitudinal plane, their longitudinal apposed orientation must be based on an organization illustrated in figure 6. Is there a direct relationship between the alignment of glial palisades and the spatial organization of other components of the molecular layer? Where such a relationship is demonstrable is there a developmental interdependence and possibly a causal one? These questions were examined with respect to two spatially organized systems: the alignment of the planar arbors of Purkinje cells and the Orientation of parallel fibers. 5. Relation of glial palisades to Purkinje cells The glial palisades and the arbors of Purkinje cells are both aligned at a right angle to the surface of the cortex. Moreover, the Bergmann glial fibers of single cells are apparently preferentially distributed in the sagittal plane, like the dendrites of Purkinje cells. But beyond these shared traits there is little or no structural relationship between the two. The oblique orientation of a large proportion of the secondary and tertiary dendritic branchlets in contrast to the vertical alignment of the palisades suggests that few of those processes of the Bergmann glia cells which constitute the palisades may be apposed to Purkinje cell dendrites. Indeed, in the lower aspect of the molecular layer, where the vertically oriented stem dendrites are located, the organization of glial palisades is least pronounced or may be altogether absent (fig. 14). Since the glial palisades are oriented in the longitudinal plane and since they are closely spaced in the same plane, several palisades must transect a single transversely oriented Purkinje cell arbor (fig. 7). Furthermore, a single palisade assembly, as a result of its length (estimated to reach up to 2 6 3 0 mp) may transect two or more Purkinje cell arbors. Thus it is difficult to conceive of a direct structural relationship between the geometry of glial palisades and Purkinje cell arbors and i t is tentatively concluded that the two are not correlated either developmentally or functionally.
6 . Glial palisades a n d parallel fibers The possible relationship between these two structural elements was examined from a developmental point of view and experimentally. We have seen that the proliferative zone of the external germinal layer is perforated by single glial endfeet. This must be either because the fibers composing the palisades dissociate before reaching this zone or because they tend to be formed developmentally after this zone begins to disappear. Longitudinal spaces are seen in the zone of bipolar cells (fig. 1B) but they may just reflect the natural alignment of spindle-shaped cells. In the older infants some of the identifiable perforating fibers are apparently assembled in longitudinal rows, others are not (fig. 1C). In general the presence of glial palisades is not easy to discern in the molecular layer of infant rats, although Bergmann glial fibers are obviously quite numerous. It is likely that their assembly into palisades is a relatively late developmental phenomenon. That the formation of glial palisades succeeds or may be secondary to the development of parallel fibers was indicated by reexamination of our X-irradiated cerebella. When the developing rat cerebellum is irradiated with a limited number of successive daily doses of X-ray, which subtotally eradicates the external germinal layer but perimits its regeneration within a few days (Altman et al., '69) the bipolar cells are often no longer aligned in the longitudinal plane and the orientation of parallel fibers is likewise altered (Altman, '73b). For instance in sagittal sections of the vermis one lobule has its parallel fibers normally oriented in the longitudinal plane while in an adjacent lobule they are aligned abnormally in the transverse plane; still in other lobules the molecular layer may be composed of two or more subdivisions with piles of parallel fibers oriented in different directions. As it is shown in figures 19-21, the alignment of parallel fibers (which is dictated by the alignment of bipolar cells) governs the presumed assembly of glial fibers into palisades. Palisades are formed always in register with the orientation of parallel fibers and change their orientation within the molecular layer as
DEVELOPING CEREBELLUM. IV
435
Fig. 6 Schematic illustration of the hypothetical alignment of vertical processes of several Bergmann glial cells to form longitudinally aligned palisades.
the alignment of parallel fibers changes. Often there are ectopic granule cells at the site where zones of parallel fibers with different orientations meet (fig. 20). Conceivably among these cells could be the somata of Bergmann glia cells that contribute fibers to the overlying palisades. But in other instances there is no cell accumulation at these interfaces (upper aspect of fig. 21) and is therefore assumed that the glial fibers that originate from
cells a t the base of the molecular layer become one or more times reassembled into palisades with different orientations. These observations indicate that the formation of glial palisades is a relatively late developmental event which follows rather than precedes the alignment of parallel fibers. If the glial palisades do not play a role in the spatial organization of parallel fibers one must assume that the interrelationship between them must be
436
JOSEPH ALTMAN
Fig. 7 Schematic illustration of the organization of glial palisades (shown as longitudinal vertical plates) and the spreading arbors of Purkinje cells. A single Purkinje cell arbor is transected by many palisades and one palisade is related to more than one Purkinje cell.
other than developmental. If one also considers the fact that the tall longitudinally oriented glial palisades are too short in the longitudinal plane to separate long vertical sheets of parallel fibers, the idea that they play a role in the structural organization of the molecular layer is furt her weakened. DISCUSSION
It is not possible to discern an organization in the spatial distribution of cells in the proliferative zone of the external germinal layer. But from the onset of their morphological differentiation, that is, when the precursors of granule cells assume a bipolar shape, they are aligned parallel to the surface of the cortex and to the long axis of the folium. This alignment is already apparent in neonates and it leads to a similar alignment of the parallel fibers in the horizontal and longitudinal planes. We suggested earlier (Altman, ’72a) that the parallel fibers grow, or are “extruded,” at the polar ends and that the future granule cells remain in the external germinal layer until the horizontal branches of the parallel fibers attain considerable length. This assumption is supported by observations described in this
study in material prepared with the Golgi technique. The next step in the differentiation of granule cells is usually described as migration through the molecular layer into the granular layer. However an examination of our Golgi and Nissl material showed that the vertical branch of the future granule cell axon grows ahead of the cell body. This is not a new observation and can be seen, for instance, in the illustrations of Athias (1897: fig. 12) and Ramon y Cajal (’11: figs. 61, 62) who spoke of the descent of a protoplasmic appendage which “drags the cell body including the nucleus with it towards the depths” (Ramon y Cajal, ’60: p. 291). Recent observers (e.g., Rakic, ’71) have repeatedly referred to the presence of such a descending process but ignored it in their conceptualization, explicitly referring to the event as migration. Whether or not differentiating granule cell perikarya migrate or are merely translocated within the growing cytoplasm (the future granule cell axon) is of some theoretical importance. Perikaryal migration is easy to conceive of in the presence of oriented surfaces that can support and guide locomotion. This kind of consideration prompted the idea that the migration of differentiating granule cells is aided by Bergmann glial fibers (Rakic, ’71). Such guiding elements could and probably do aid the growing granule cell processes, if for no other reason because they create vertical lines free of crisscrossing tissue elements that can divert or block the growing tips. But the hypothesis that the presence of aligned Bergmann fibers is essential for “migration” into the granular layer has been weakened by two sets of recent observations. According to Bignami and Dahl (’73), who used an immunofluorescent technique to visualize astrocytic elements, the developing Bergmann glial fibers are initially tortuous and do not become clearly aligned in the vertical direction until the end of the second week. But differentiating granule cells “migrate” through the granular layer in large numbers before these presumed guiding fibers are aligned. In another approach Rakic and Sidman (’73) studied the mechanism that may underlie the failure of many granule cells to reach the granular layer
DEVELOPING CEREBELLUM, IV
in the cerebellar cortex of heterozygous and homozygous weaver mice. They observed that the granule cell precursors of the mutants tend to die near the production site and that this pathology is coupled with a reduction in radially aligned glial fibers. They suggested that perhaps “the Bergmann glial disorder precedes the migration defect temporally and causally” (Rakic and Sidman, ’73: p. 129). But this hypothesis failed to gain direct support in a subsequent study by Bignami and Dahl (’74b) who found in heterozygous weaver and normal littermates that “the immunofluorescen t pattern of Bergmann fibers was identical throughout development (p. 219)’’ Bergmann fibers were also present in homozygous weavers in apparently unreduced numbers, though they were shorter and less regular in appearance than in normal littermates of comparable ages. Similar results were obtained by Sotello and Changeux (’74). These findings do not rule out the possibility that the glial fibers, though present, are abnormal and lack those surface properties that guide the descending granule cells (Bignami and Dahl, ’74: p. 228). But if our observation is correct that the translocation of a granule cell is preceded by the growth of its vertical fiber through the molecular layer then the event can be conceptualized as guided by some chemotactic mechanism that is relatively independent of the presence and orientation of glial fibers. Earlier we suggested (Altman, ’72a) the terms “proliferative zone” and “premigratory zone” for the two subdivisions of the external germinal layer. If the precursors of granule cells do not truly migrate it will no longer be meaningful to speak of a premigratory zone. We propose instead the term “subproliferative zone” for the latter region. The observation that the Bergmann glial fibers do not become aligned radially for several days after birth (Bignami and Dahl, ’73), that the majority of Bergmann glia cells are formed after day 9 (Das et al., ’74) and the results of the present study that the organization of these fibers into palisades occurs some time after granule cells have begun their translocation, seem to suggest that it is not likely that
437
the primary function of Bergmann glia cells is a developmental one. It appears rather that the organization of Bergmann fibers is linked with the presence of parallel fibers. This view is also supported by X-irradiation studies. The results described in this paper indicate that the directional alignment of glial palisades is dependent on the alignment of parallel fibers. Since the alignment of parallel fibers is the primary developmental event it is safe to say that when the glial palisades are realigned after recovery of the external germinal layer from X-irradiation that this is a consequence of the reorientation of parallel fibers, an event which is attributed to the realignment of bipolar cells in the external germinal layer (Altman, ’73b). More recent studies (Altman, in preparation) showed that if the cerebellum is irradiated with several successive doses of X-ray beginning on day 4 or day 8, which prevent cell regeneration, that glial palisades are not formed. However, if such an irradiation schedule is started after day 12 the formation of glial palisades is not interfered with. We can only speculate at present what the function of these glial elements might be. Since their maturation is coincident with the termination of parallel fiber morphogenesis and the onset of synaptogenesis in the molecular layer (Altman, ’72b) conceivably their role is not structural but rather functional, perhaps trophic in nature. This conclusion supports the earlier hypothesis (Altman, ’72b: figs. 13, 29, 31) that gliogenesis follows synaptogenesis in the maturing zones of the molecular layer. We pointed out earlier that the spatial organization of glial palisades suggests that they may be more closely related to parallel fibers than to Purkinje cells. But this view raises the question as to what cellular elements supply glial processes to Purkinje cells, especially to the dendrites and spiny branchlets situated between the palisades? Conceivably, the vertical gaps seen in Bodian preparation contain the thicker glial branches while the smaller glial processes, not resolved with this technique, are apposed to the smooth branches and spiny branchlets of Purkinje cells. Hopefully an answer to this question will be supplied by electron
438
JOSEPH ALTMAN
microscopic observations which should also shed light on the structural properties of the regular “gaps” that we postulated to represent glial palisades. ADKNOWLEDGMENT
I wish to thank Z.-Y. Chen for collecting some of the brains, Sharon Evander for the histology, Zeynep Kurgun- Chen for the photography and illustrations, and Dr. Shirley Bayer for discussions. This research is supported by the National Institute of Mental Health and the U. S. Energy Research and Development Administration. LITERATURE CITED Addison, W . H. F. 1911 The development of the Purkinje cells and of the cortical layers in the cerebellum of the albino rat. J. Comp. Neur., 21 : 4591187. Altman, J. 1966 Autoradiographic and histological studies of postnatal neurogenesis. 11. A longitudinal investigation of the kinetics, migration and transformation of cells incorporating tritiated thymidine in infant rats, with special reference to postnatal neurogenesis in some brain regions. J . Comp. Neur., 128: 431474. 1969 Autoradiographic and histological studies of postnatal neurogenesis. 111. Dating the time of production and onset of differentiation of cerebellar microneurons in rats. J. Comp. Neur., 1 3 6 : 269-294. 1972a Postnatal development of the cerebellar cortex in the rat. I. The external germinal layer and the transitional molecular layer. J . Comp. Neur., 1 4 5 : 35S398. 1972b Postnatal development of the cerebellar cortex in the rat. 11. Phases in the maturation of Purkinje cells and of the molecular layer. J. Comp. Neur., 1 4 5 : 39-64, 1972c Postnatal development of the cerebellar cortex in the rat. 111. Maturation of the components of the granular layer. J. Comp. Neur., 1 4 5 : 4 6 5 5 1 4 . 1973a Experimental reorganization of the cerebellar cortex. 111. Regeneration of the external germinal layer and granule cell ectopia. J. Comp. Neur., 1 4 9 : 153-180. 1973b Experimental reorganizationof the cerebellar cortex. IV. Parallel fiber reorientation following regeneration of the external germinal layer. J. Comp. Neur., 1 4 9 : 181-192.
Altman, J . , and W. J. Anderson 1972 Experimental reorganization of the cerebellar cortex. I. Morphological effects of elimination of all microneurons with prolonged X-irradiation started at birth. J. Comp. Neur., 1 4 6 : 355406. Altman, J., W. J. Anderson and K . A. Wright 1968 Differential radiosensitivity of stationary and migratory primitive cells in the brains ofinfant rats. Exp. Neur., 2 2 : 52-74. 1969 Early effects of X-irradiation of the cerebellum in infant rats; Decimation and reconstitution of the external granular layer. Exp. Neur., 2 4 : 196-216. Altman, J., and A. T. Winfree 1975 Postnatal development of the cerebellar cortex in the rat; V. Spatial organization of Purkinje cells and basket cells. In preparation. Athias, M. 1897 Recherches sur 1’histogC.nesede l’ecorce du cervelet. J. Anat. Physiol., 3 3 : 37% 404. Bignami, A , , and D. Dahl 1973 Differentiation of astrocytes in the cerebellar cortex and the pyramidal tracts of the newborn rat. An immunofluorescence study with antibodies to a protein specific to astrocytes. Brain Res., 4 9 : 393402. 1974a Astrocyte-specific protein and neuroglial differentiation. An immunofluorescence study with antibodies to the glial fibrillary acidic protein. J. Comp. Neur., 1 5 3 : 27-38. 1974b The development of Bergmann glia in mutant mice with cerebellar malformations: Reeler, staggerer and weaver. Immunofluorescence study with antibodies to the glial fibrillary acidic protein. J . Comp. Neur., 1 5 5 : 219-230. Das, G. D., G. L. Lammert and J . P. McAllister 1974 Contact guidance and migratory cells in the developing cerebellum. Brain Res., 6 9 : 13-29. Rakic, P. 1971 Neuron-glia relationship during granule cell migration in developing cerebellar cortex. A Golgi and electronmicroscopic study i n Macacus rhesus. J. Comp. Neur., 141 : 2 8 s 312. Rakic, P., and R. L. Sidman 1973 Sequence of developmental abnormalities leading to granule cell deficit in cerebellar cortex of weaver mutant mice. J . Comp. Neur., 1 5 2 : 103-132. Ramon y Cajal, S. 1911 Histologie du Systeme Nerveux de YHomme & des VertkbrCs. Two vols. Pans, Maloine. 1960 Studies on vertebrate neurogenesis. Translated by L. Guth. Springfield, Illinois, Thom as. Sauer, F. C. 1935 The cellular structure of the neural tube. J. Comp. Neur., 6 3 : 13-23. Sotello, C., and J. P. Changew 1974 Bergman fibers and granular cell migration in the cerebellum ofhomozygous weaver mutant mouse. Brain Res., 77: 4 8 4 4 9 1 .
PLATES
PLATE 1 EXPLANATION OF FIGURES
Photomicrographs illustrating the postulated two phases in the translocation of developing granule cells from the external germinal layer to the granular layer. From sagittal sections of the vermis of a single 10-day-old rat. Cresyl violet, oil immersion. X 900.
440
8-10
Descending single processes issuing from vertically oriented pearshaped cells at the base of the external germinal layer.
11-13
Spindle-shaped cells apparently translocating through the molecular laver.
DEVELOPING C E R E B E L L U M . I V Joseph Altman
PLATE 1
44 1
PLATE 2 E X P L A N A T I O N OF FIGURES
Appearance of the molecular layer in 10 mp-thick sections stained with Bodian’s protargol-S method which tends to tint the parallel fibers. Adult rats. x 312.
442
14
Section cut sagittally through the pyramis. Note the parallel fibers cut transversely ( h e dots) and the vertically oriented free spaces. The latter art?presumed to represent glial palisades.
15
Section cut coronally through the pyramis of another adult rat. The parallel fibers are cut parallel (or slightly obliquely) and there are no vertically oriented columns visible. Presumably the thin ( 2 4 m p ) columns are obscured by parallel fibers situated below and/or above them.
DEVELOPING CEREBELLUM, IV Joseph Altman
PLATE 2
443
PLATE 3 EXPLANATION O F FIGURES
Glial palisades at three levels. Selected levels from serial horizontal sections through a v e r n a l lobule in an adult rat. Bodian. X 312.
444
16
Interface of pia (note blood vessel and capillaries) and surface of molecular layer.
17
Middle portion of molecular layer.
18
Lower portion of molecular layer with impregnated basket cell axons oriented transversely to the glial palisades. Associated with an increase of cell bodies a t this level (basket and Bergmann glia cells?) the glial palisades become gradually less distinct at the base of the molecular layer.
DEVELOPING CEREBELLUM. I V Joseph Altmnn
PLATE 3
PLATE 4 E X P L A N A T I O N OF F I G U R E S
Realignment of glial palisades in sagittal sections of the vermis of X-irradiated rats i n association with the reorientation of parallel fibers. The cerebella were irradiated during infancy with several successive daily doses of 150 r X-ray which results in decimation and subsequent regeneration of the external germinal layer. Adult rats. Bodian. X 216.
446
19
Cerebellum irradiated with two successive doses on days 3 and 4. Note that except at the base of the molecular layer the alignment of parallel fibers a n d glial palisades is rssentially normal.
20,21
Cerebella received four successive doses on days 3, 4, 5 and 6. Note that the palisades are recognizable only in zones of the molecular layer in which the parallel fibers a r e cut transversely. Two zones may be identified in figure 20, three zones i n figure 21.
D E V E L 0 I’ I NG C: k: K E B E LL U M , I V Joseph h l t m a n
44 7
AND GLIAL PALISADES JOSEPH ALTMAN Laboratory of Developmental Neurobiology, D e p a r t m e n t of Biological Sciences, Purdue University, Lafayette, Indictnu 47907
ABSTRACT The ontogeny of the spatial organization of some components of the molecular layer was investigated in cerebella sectioned systematically in the sagittal, coronal and horizontal planes. There is no discernible organization in the distribution of cells of the proliferative zone of the external germinal layer (EGL) but from birth the differentiating bipolar cells of the subproliferative zone are aligned parallel to the surface and to the long axis of the folium. While they are still in or at the base of the EGL, the bipolar cells emit long processes, the future parallel fibers. The next step is the outgrowth of a vertical process which may reach the base of the molecular layer before the granule cell nucleus becomes translocated. The idea that the cell body truly migrates through the molecular layer is not supported by the observations. BergmaM glia cells are frequently seen in Golgi material in neonates but they are probably less numerous than i n older infants and their processes are not as well aligned. It is only gradually that the EGL is perforated by glial endfeet which in older infants are occasionally organized into longitudinal rows. I n mature cerebella the parallel fibers are separated by thin and relatively narrow, unstained spaces which are oriented in the longitudinal plane and can be traced from the pial surface to a zone just above the layer of Purkinje cells. It is postulated that these spaces are occupied by glial palisades formed by apposed thin vertical processes to which many Bergmann glia cells contribute. The alignment of these palisades is dependent on the orientation of parallel fibers. When the parallel fibers are reoriented by X-irradiation the glial palisades become correspondingly realigned. These observations indicate that the oriented growth of parallel fibers, which follows the polarization of bipolar cells, determines the spatial organization of the glial framework of the molecular layer. They also suggest that the glial palisades mediate functions that are not primarily developmental in nature.
The cells and cell processes of the cerebellar cortex form a regular three-dimensional lattice-work which is apparently uniform throughout and which, with minor modifications and few exceptions, is a characteristic feature of the cerebella of all higher vertebrates. The present paper and its companion (Altman and Winfree, in prep.) are concerned with aspects of the ontogeny of this geometric organization coupled with the attempt to identify some of the mechanisms involved. The geometric organization of the supraganglionic cerebellar cortex of the rat develops postnatally (Ramon y Cajal, '11, J.
COMP. NEUR.,163: 427-448.
'60). The perikarya of Purkinje cells disperse into a monolayer several days after birth and the development of their planar arbors in the molecular layer starts during the second week (Addison, '11; Altman, '72b). Cells multiplying in the external germinal layer differentiate as basket cells during the second week and most of the granule cells are formed during the second and third week (Altman, '69, '72a,b,c). These studies of normally developing cerebella and subsequent experimental work (Altman and Anderson, '72; Altman, '73a,b) have suggested that some properties of cells of the external germinal 427
428
JOSEPH ALTMAN
layer and their initial mode of morphogenic differentiation play a major role in the subsequent development of cortical geometry. While Purkinje cell perikarya display considerable autonomy and develop normally in size when the formation of granule and basket cells is prevented by X-irradiation (which kills selectively the multiplying germinal cells) the Purkinje cells do not disperse in the absence of parallel fibers, their primary and secondary dendrites become randomly oriented and they fail to develop tertiary and higher order spiny branchlets (Altman and Anderson, '72). If the orientation of parallel fibers is changed, which occurs after regeneration of the external germinal layer consequent to irradiation with certain schedules (Altman, '73b) the planar Purkinje cell dendritic plexus becomes rotated at a right angle to the pile of parallel fibers. We will attempt to show in this and the succeeding paper that the orientation of bipolar cells of the external germinal layer determines not only the orientation of parallel fibers but also the geometry of several other elements of the molecular layer. Special attention is paid in this paper to the radial glial processes that were postulated to guide the descending granule cells (Rakic, '71; Rakic and Sidman, '73). Our observations indicate that the differen tiating granule cells are translocated into the granular layer, following their descending vertical processes that find their way there. The organization of glial processes, which assemble into longitudinally aligned vertical palisades, occurs after, not before, the alignment of parallel fibers. In the subsequent paper the dispersion of Purkinje cells into a monolayer, the alignment of Purkinje cell dendrites in the transverse longitudinal plane and the alignment of the proximal and distal axons of basket cells are described as events that are secondary to the geometric organization of the parallel fiber system.
10, 12, 14, 16, 18 and 20 days were embedded in paraffin and cut serially at 10 mp in the coronal, sagittal and horizontal planes. The sections were stained in a systematic sequence with hematoxylin-eosin, cresyl violet, the Weigert-Loyez and Bodian's protargol-S methods. The cerebella of three adult cats were cut similarly in the three planes and every section was saved to allow three-dimensional reconstructions. Finally brains from rats aged 0, 1, 2, 3, 4, 7, 10, 15, 21 and 28 days, and from adults, were impregnated with the rapid Golgi and Golgi-Cox techniques. The irradiated cerebella came f?om the collection that was described earlier (Altman, '73a). RESULTS
1. T h e external germinal layer In the external germinal layer proliferative cells are localized in the superficial, or subpial zone. Some of these cells show mitotic activity (fig. 1A) and many of them are labelled when infant rats are injected with thymidine-H3 and are killed within 1-2 hours thereafter (Altman, '66, '72a). When the distribution of these cells is examined in sections cut parallel to the pial surface (fig. 1A) the tightly packed cells do not display any obvious geometric organization. In older infants circular cellfree regions may be seen among the cells; these are presumed to be the end-feet of Bergmann glia processes that reach the pia. Some of the cells appear to form rosettes around these processes but there is generally little change in the essentially random distribution of the tightly packed proliferative cells. The deeper zone of the external germinal layer is composed of nonproliferative cells which appear labelled only when at least six hours elapsed between injection of thymidine-Hs and the death of the animal (Altman, '72a). These bipolar cells form a thin zone in neonates, then increase rapidly to form a layer up to ten cells thick by about ten days, and decline MATERIALS A N D METHODS gradually thereafter. But from the very In addition to the extensive collection beginning these bipolar cells are oriented of normal developmental material that was regularly parallel to the long axis of the described in detail earlier (Altman, '72a) folium (fig. 1B). In the vermis they appear in this study use was made of some addi- accordingly as spindle-shaped cells in corotional material. The brains of male Pur- nal sections, and as small round or ovoid due-Wistar rats aged 0, l, 2, 3, 4, 5, 6, 8, cells in sagittal sections. This regular or-
DEVELOPING CEREBELLUM. I V
429
Fig. 1 Tangentially cut sections through the external germinal layer. X 288. Hematoxylineosin. A. Section through the proliferative zone in the vermis of a 2-day-old rat. Arrows point to some of the mitotic cells. B. Section through the subproliferative zone i n the vermis of a 3 - d a y d d rat. C. Similar to B from a n 8-day-old rat. Note the presumed perforating glial endfeet which are occasionally aligned into longitudinal rows (arrows).
ganization prevails as long as the external germinal layer is present (about 21 days of age) and the only developmental change that can be seen is the perforation of the layer by presumed processes of Bergmann’s glia cells (fig. 1C) which are occasionally contiguous with each other and form a row in the longitudinal plane. In some sagittal sections the transversely cut bipolar cells are aligned vertically. In many sagittal sections this is only vague-
ly recognizable or is not evident at all; it is assumed that the horizontal alignment is caused by the vertically oriented thin Bergmann glial processes and is seen only when the section is cut exactly in the same plane as the ascending glial processes. In well-stained sections the just-resolvable, transversely-cut parallel fibers of the molecular layer may be absent in vertical strips that are continuous with the cellfree vertical columns. Such a vertical align-
4 30
JOSEPH A1,TMAN
Fig. 2 Bergmann glia cells in 2-day-old ( B ) and 3-day-old (A and C ) rats; egl, external germinal layer. X 288. Golgi-Cox.
ment of cells is never seen in sections cut parallel to the long axis of the folium. Bergmann glia cells with ascending vertical processes are seen in neonates and are, indeed, the most frequently impregnated cellular elements in Golgi sections (fig. 2). Typically the cell body is located between or underneath Purkinje cells, its thin processes branch below the external germinal layer, traverse this region and terminate with end feet in the pia (where they may be impregnated jointly with the pia or blood vessels). Generally the branching of glial processes is richer in the sagittal plane (fig. 2B,C) than in the coronal plane (fig. 2A) but exceptions are seen.
nal plane. I t was not possible to establish with any certainty whether or not, or how often, the growing polar tips take the shape of growth cones.
3 . The translocating g r a n u l e cells Presumably after the horizontal portion of the future granule cell axon has reached a certain length, the formation of its vertical portion begins. It is generally thought that this is accomplished by migration of the granule cell through the molecular layer. The evidence provided by the Golgi technique is ambiguous. Often one sees leading perikarya with trailing processes, in other instances (fig. 4) shorter or longer descending processes precede the peri2. The transitional molecular layer karya. The outgrowing process (or procThe transitional molecular layer is the esses, figs. 4B,C) may initially have a zone composed of differentiating cells and somewhat uncertain orientation but as the their processes at the base of the external leading process becomes longer it unerrgerminal layer and the (temporary) ceil- ingly assumes a vertical orientation with ing of the molecular layer (Altman, ’72a). respect to the surface of the cortex. I t is We postulated that the spindle-shaped cells of some theoretical importance whether in this cortical sheet are not migrating the perikaryon or the vertical process is but rather have this shape because they in leading position; the former implies “extrude” cytoplasm at the two poles. cellular migration, and possibly mechaniThese become the horizontal or distal por- cal guidance (as by Bergman glial proctions of the future parallel fibers. That, esses; Rakic, ’71) while if the latter is the indeed, the horizontal fibers may acquire case nonmechanical guidance is a more considerable length before the differenti- parsimonious assumption and the nucleus ating granule cell descends through the may be thought of as translocating rather molecular layer is illustrated in Golgi-im- than migrating, as in the case of the interpregnated material in figure 3. A t this kinetic migration of neuroepithelial cells stage of their growth the horizontal fibers (Sauer, ’35). Examination of this problem tend to be somewhat kinky, more so when in favorably cut and strongly stained (heviewed in the horizontal than in the coro- matoxylin-eosin) sections at high magni-
431
DEVELOPING CEREBELLUM. IV
0
R
X 1
4
432
JOSEPH A L.T M AN
fication (figs. %13) indicate that the translocation of the differentiating granule cell from the transitional molecular layer into the granular layer is guided by the outgrowing vertical process. Cells are seen with pear-shaped perikarya a t the base of the external germinal layer whose single descending processes reach as far as the layer of Purkinje cells. In other instances oval or slender spindle shaped perikarya are “caught” a t different points in their descent through the molecular layer (figs. 11-13). This staining technique, unlike the Golgi procedure, tends to stain preferentially the leading portion of the vertical fiber which sometimes appear much thicker than the differentiated (trailing) parallel fiber. Perhaps the leading process is rich in cytoplasm, as i t was indicated earlier by electron microscopic observations (Altman, ’72c, fig. 4). 4. The glial palisades
We proposed earlier that the perpendicular columnar alignment of bipolar cells
seen in favorably cut transverse sections is produced by the ascending processes of Bergmann glia cells. In sections cut horizontally through the sheet formed by bipolar cells the perforating glial processes and their endfeet were occasionally lined up i n rows parallel to the long axis of the folium, or the orientation of bipolar cells (fig. 1C). This suggested that the vertical glial fibers are apposed and form longitudinally aligned vertical palisades. I n the cerebella of one week or older rats these vertical columns could sometimes be traced through the molecular layer as far as the layer of Purkinje cells; they were most apparent in brittle sagittal sections in which the tissue tended to fracture along these presumed columns. In mature cerebella these vertical columns were more obvious. The organization of these columns was therefore examined in cerebella from three adult rats that were cut serially in the horizontal, sagittal and coronal planes and were systematically stained with different techniques, as it was described ear-
Fig. 4 Transforming bipolar cells with vertical processes of different length, presumably displaying progressive phases of development. All from the same 10-day-old rat. X 720. Golgi-Cox.
DEVELOPING CEREBELLUM. I V
lier. The data described below were largely obtained by examining 10 p-thick sections prepared with Bodian’s protargol-S method. This procedure tends to tint lightly the parallel fibers. In sagittally cut sections of the vermis in which the parallel fibers appear as small dots (that is, are cut exactly at a right angle to the beam of parallel fibers) unstained “gaps” 2-4 mp wide extend from the pial surface to the proximity of the layer of Purkinje cells (fig. 14). Since these spaces are free of transversely cut parallel fibers (vertical fibers are sometimes seen coursing through them) it is concluded that they have the shape of vertical plates which are thin in the transverse plane of the folium and are at least 10 p wide in the longitudinal plane parallel to the alignment of the horizontal beams of parallel fibers. In coronal sections of the vermis, in which the parallel fibers are cut parallel or obliquely, these unstained vertical gaps are not seen (fig. 15). This is com-
433
patible with the idea that they are thin in this plane and would be obscured by parallel fibers situated below and/or above them. Finally in sections cut horizontally through the molecular layer the unstained spaces were identified as rows composed of a variable number of closely apposed circular or ovoid elements (fig. 5). These could be traced from the surface of the molecular layer to its base where basket cell axons impregnated by the Bodian technique are oriented at a right angle to these spaces and the beams of parallel fibers (figs. 1€L 18). Evidently these gaps are made up of long vertical fibers that form palisades in the longitudinal plane. It is postulated that the fibers are those of Bergmann glia cells. Bignami and Dahl (’74a) observed the alignment of Bergmann glial cells in rows in tangential sections (their fig. 6 ) . Since the fibers of a single Bergmann glia cell tend to fan out (fig. 2) it is assumed that the contiguous fibers originate fkom several cells. Moreover, considering that
Fig. 5 Molecular layer in horizontal section. Note the longitudinally oriented spaces which are apparently made up of many apposed circular and ovoid elements (arrows). Adult rat. X 1123.2. Bodian.
434
JOSEPH ALTMAN
the spread of glial fibers is more pronounced in the transverse (sagittal) than longitudinal plane, their longitudinal apposed orientation must be based on an organization illustrated in figure 6. Is there a direct relationship between the alignment of glial palisades and the spatial organization of other components of the molecular layer? Where such a relationship is demonstrable is there a developmental interdependence and possibly a causal one? These questions were examined with respect to two spatially organized systems: the alignment of the planar arbors of Purkinje cells and the Orientation of parallel fibers. 5. Relation of glial palisades to Purkinje cells The glial palisades and the arbors of Purkinje cells are both aligned at a right angle to the surface of the cortex. Moreover, the Bergmann glial fibers of single cells are apparently preferentially distributed in the sagittal plane, like the dendrites of Purkinje cells. But beyond these shared traits there is little or no structural relationship between the two. The oblique orientation of a large proportion of the secondary and tertiary dendritic branchlets in contrast to the vertical alignment of the palisades suggests that few of those processes of the Bergmann glia cells which constitute the palisades may be apposed to Purkinje cell dendrites. Indeed, in the lower aspect of the molecular layer, where the vertically oriented stem dendrites are located, the organization of glial palisades is least pronounced or may be altogether absent (fig. 14). Since the glial palisades are oriented in the longitudinal plane and since they are closely spaced in the same plane, several palisades must transect a single transversely oriented Purkinje cell arbor (fig. 7). Furthermore, a single palisade assembly, as a result of its length (estimated to reach up to 2 6 3 0 mp) may transect two or more Purkinje cell arbors. Thus it is difficult to conceive of a direct structural relationship between the geometry of glial palisades and Purkinje cell arbors and i t is tentatively concluded that the two are not correlated either developmentally or functionally.
6 . Glial palisades a n d parallel fibers The possible relationship between these two structural elements was examined from a developmental point of view and experimentally. We have seen that the proliferative zone of the external germinal layer is perforated by single glial endfeet. This must be either because the fibers composing the palisades dissociate before reaching this zone or because they tend to be formed developmentally after this zone begins to disappear. Longitudinal spaces are seen in the zone of bipolar cells (fig. 1B) but they may just reflect the natural alignment of spindle-shaped cells. In the older infants some of the identifiable perforating fibers are apparently assembled in longitudinal rows, others are not (fig. 1C). In general the presence of glial palisades is not easy to discern in the molecular layer of infant rats, although Bergmann glial fibers are obviously quite numerous. It is likely that their assembly into palisades is a relatively late developmental phenomenon. That the formation of glial palisades succeeds or may be secondary to the development of parallel fibers was indicated by reexamination of our X-irradiated cerebella. When the developing rat cerebellum is irradiated with a limited number of successive daily doses of X-ray, which subtotally eradicates the external germinal layer but perimits its regeneration within a few days (Altman et al., '69) the bipolar cells are often no longer aligned in the longitudinal plane and the orientation of parallel fibers is likewise altered (Altman, '73b). For instance in sagittal sections of the vermis one lobule has its parallel fibers normally oriented in the longitudinal plane while in an adjacent lobule they are aligned abnormally in the transverse plane; still in other lobules the molecular layer may be composed of two or more subdivisions with piles of parallel fibers oriented in different directions. As it is shown in figures 19-21, the alignment of parallel fibers (which is dictated by the alignment of bipolar cells) governs the presumed assembly of glial fibers into palisades. Palisades are formed always in register with the orientation of parallel fibers and change their orientation within the molecular layer as
DEVELOPING CEREBELLUM. IV
435
Fig. 6 Schematic illustration of the hypothetical alignment of vertical processes of several Bergmann glial cells to form longitudinally aligned palisades.
the alignment of parallel fibers changes. Often there are ectopic granule cells at the site where zones of parallel fibers with different orientations meet (fig. 20). Conceivably among these cells could be the somata of Bergmann glia cells that contribute fibers to the overlying palisades. But in other instances there is no cell accumulation at these interfaces (upper aspect of fig. 21) and is therefore assumed that the glial fibers that originate from
cells a t the base of the molecular layer become one or more times reassembled into palisades with different orientations. These observations indicate that the formation of glial palisades is a relatively late developmental event which follows rather than precedes the alignment of parallel fibers. If the glial palisades do not play a role in the spatial organization of parallel fibers one must assume that the interrelationship between them must be
436
JOSEPH ALTMAN
Fig. 7 Schematic illustration of the organization of glial palisades (shown as longitudinal vertical plates) and the spreading arbors of Purkinje cells. A single Purkinje cell arbor is transected by many palisades and one palisade is related to more than one Purkinje cell.
other than developmental. If one also considers the fact that the tall longitudinally oriented glial palisades are too short in the longitudinal plane to separate long vertical sheets of parallel fibers, the idea that they play a role in the structural organization of the molecular layer is furt her weakened. DISCUSSION
It is not possible to discern an organization in the spatial distribution of cells in the proliferative zone of the external germinal layer. But from the onset of their morphological differentiation, that is, when the precursors of granule cells assume a bipolar shape, they are aligned parallel to the surface of the cortex and to the long axis of the folium. This alignment is already apparent in neonates and it leads to a similar alignment of the parallel fibers in the horizontal and longitudinal planes. We suggested earlier (Altman, ’72a) that the parallel fibers grow, or are “extruded,” at the polar ends and that the future granule cells remain in the external germinal layer until the horizontal branches of the parallel fibers attain considerable length. This assumption is supported by observations described in this
study in material prepared with the Golgi technique. The next step in the differentiation of granule cells is usually described as migration through the molecular layer into the granular layer. However an examination of our Golgi and Nissl material showed that the vertical branch of the future granule cell axon grows ahead of the cell body. This is not a new observation and can be seen, for instance, in the illustrations of Athias (1897: fig. 12) and Ramon y Cajal (’11: figs. 61, 62) who spoke of the descent of a protoplasmic appendage which “drags the cell body including the nucleus with it towards the depths” (Ramon y Cajal, ’60: p. 291). Recent observers (e.g., Rakic, ’71) have repeatedly referred to the presence of such a descending process but ignored it in their conceptualization, explicitly referring to the event as migration. Whether or not differentiating granule cell perikarya migrate or are merely translocated within the growing cytoplasm (the future granule cell axon) is of some theoretical importance. Perikaryal migration is easy to conceive of in the presence of oriented surfaces that can support and guide locomotion. This kind of consideration prompted the idea that the migration of differentiating granule cells is aided by Bergmann glial fibers (Rakic, ’71). Such guiding elements could and probably do aid the growing granule cell processes, if for no other reason because they create vertical lines free of crisscrossing tissue elements that can divert or block the growing tips. But the hypothesis that the presence of aligned Bergmann fibers is essential for “migration” into the granular layer has been weakened by two sets of recent observations. According to Bignami and Dahl (’73), who used an immunofluorescent technique to visualize astrocytic elements, the developing Bergmann glial fibers are initially tortuous and do not become clearly aligned in the vertical direction until the end of the second week. But differentiating granule cells “migrate” through the granular layer in large numbers before these presumed guiding fibers are aligned. In another approach Rakic and Sidman (’73) studied the mechanism that may underlie the failure of many granule cells to reach the granular layer
DEVELOPING CEREBELLUM, IV
in the cerebellar cortex of heterozygous and homozygous weaver mice. They observed that the granule cell precursors of the mutants tend to die near the production site and that this pathology is coupled with a reduction in radially aligned glial fibers. They suggested that perhaps “the Bergmann glial disorder precedes the migration defect temporally and causally” (Rakic and Sidman, ’73: p. 129). But this hypothesis failed to gain direct support in a subsequent study by Bignami and Dahl (’74b) who found in heterozygous weaver and normal littermates that “the immunofluorescen t pattern of Bergmann fibers was identical throughout development (p. 219)’’ Bergmann fibers were also present in homozygous weavers in apparently unreduced numbers, though they were shorter and less regular in appearance than in normal littermates of comparable ages. Similar results were obtained by Sotello and Changeux (’74). These findings do not rule out the possibility that the glial fibers, though present, are abnormal and lack those surface properties that guide the descending granule cells (Bignami and Dahl, ’74: p. 228). But if our observation is correct that the translocation of a granule cell is preceded by the growth of its vertical fiber through the molecular layer then the event can be conceptualized as guided by some chemotactic mechanism that is relatively independent of the presence and orientation of glial fibers. Earlier we suggested (Altman, ’72a) the terms “proliferative zone” and “premigratory zone” for the two subdivisions of the external germinal layer. If the precursors of granule cells do not truly migrate it will no longer be meaningful to speak of a premigratory zone. We propose instead the term “subproliferative zone” for the latter region. The observation that the Bergmann glial fibers do not become aligned radially for several days after birth (Bignami and Dahl, ’73), that the majority of Bergmann glia cells are formed after day 9 (Das et al., ’74) and the results of the present study that the organization of these fibers into palisades occurs some time after granule cells have begun their translocation, seem to suggest that it is not likely that
437
the primary function of Bergmann glia cells is a developmental one. It appears rather that the organization of Bergmann fibers is linked with the presence of parallel fibers. This view is also supported by X-irradiation studies. The results described in this paper indicate that the directional alignment of glial palisades is dependent on the alignment of parallel fibers. Since the alignment of parallel fibers is the primary developmental event it is safe to say that when the glial palisades are realigned after recovery of the external germinal layer from X-irradiation that this is a consequence of the reorientation of parallel fibers, an event which is attributed to the realignment of bipolar cells in the external germinal layer (Altman, ’73b). More recent studies (Altman, in preparation) showed that if the cerebellum is irradiated with several successive doses of X-ray beginning on day 4 or day 8, which prevent cell regeneration, that glial palisades are not formed. However, if such an irradiation schedule is started after day 12 the formation of glial palisades is not interfered with. We can only speculate at present what the function of these glial elements might be. Since their maturation is coincident with the termination of parallel fiber morphogenesis and the onset of synaptogenesis in the molecular layer (Altman, ’72b) conceivably their role is not structural but rather functional, perhaps trophic in nature. This conclusion supports the earlier hypothesis (Altman, ’72b: figs. 13, 29, 31) that gliogenesis follows synaptogenesis in the maturing zones of the molecular layer. We pointed out earlier that the spatial organization of glial palisades suggests that they may be more closely related to parallel fibers than to Purkinje cells. But this view raises the question as to what cellular elements supply glial processes to Purkinje cells, especially to the dendrites and spiny branchlets situated between the palisades? Conceivably, the vertical gaps seen in Bodian preparation contain the thicker glial branches while the smaller glial processes, not resolved with this technique, are apposed to the smooth branches and spiny branchlets of Purkinje cells. Hopefully an answer to this question will be supplied by electron
438
JOSEPH ALTMAN
microscopic observations which should also shed light on the structural properties of the regular “gaps” that we postulated to represent glial palisades. ADKNOWLEDGMENT
I wish to thank Z.-Y. Chen for collecting some of the brains, Sharon Evander for the histology, Zeynep Kurgun- Chen for the photography and illustrations, and Dr. Shirley Bayer for discussions. This research is supported by the National Institute of Mental Health and the U. S. Energy Research and Development Administration. LITERATURE CITED Addison, W . H. F. 1911 The development of the Purkinje cells and of the cortical layers in the cerebellum of the albino rat. J. Comp. Neur., 21 : 4591187. Altman, J. 1966 Autoradiographic and histological studies of postnatal neurogenesis. 11. A longitudinal investigation of the kinetics, migration and transformation of cells incorporating tritiated thymidine in infant rats, with special reference to postnatal neurogenesis in some brain regions. J . Comp. Neur., 128: 431474. 1969 Autoradiographic and histological studies of postnatal neurogenesis. 111. Dating the time of production and onset of differentiation of cerebellar microneurons in rats. J. Comp. Neur., 1 3 6 : 269-294. 1972a Postnatal development of the cerebellar cortex in the rat. I. The external germinal layer and the transitional molecular layer. J . Comp. Neur., 1 4 5 : 35S398. 1972b Postnatal development of the cerebellar cortex in the rat. 11. Phases in the maturation of Purkinje cells and of the molecular layer. J. Comp. Neur., 1 4 5 : 39-64, 1972c Postnatal development of the cerebellar cortex in the rat. 111. Maturation of the components of the granular layer. J. Comp. Neur., 1 4 5 : 4 6 5 5 1 4 . 1973a Experimental reorganization of the cerebellar cortex. 111. Regeneration of the external germinal layer and granule cell ectopia. J. Comp. Neur., 1 4 9 : 153-180. 1973b Experimental reorganizationof the cerebellar cortex. IV. Parallel fiber reorientation following regeneration of the external germinal layer. J. Comp. Neur., 1 4 9 : 181-192.
Altman, J . , and W. J. Anderson 1972 Experimental reorganization of the cerebellar cortex. I. Morphological effects of elimination of all microneurons with prolonged X-irradiation started at birth. J. Comp. Neur., 1 4 6 : 355406. Altman, J., W. J. Anderson and K . A. Wright 1968 Differential radiosensitivity of stationary and migratory primitive cells in the brains ofinfant rats. Exp. Neur., 2 2 : 52-74. 1969 Early effects of X-irradiation of the cerebellum in infant rats; Decimation and reconstitution of the external granular layer. Exp. Neur., 2 4 : 196-216. Altman, J., and A. T. Winfree 1975 Postnatal development of the cerebellar cortex in the rat; V. Spatial organization of Purkinje cells and basket cells. In preparation. Athias, M. 1897 Recherches sur 1’histogC.nesede l’ecorce du cervelet. J. Anat. Physiol., 3 3 : 37% 404. Bignami, A , , and D. Dahl 1973 Differentiation of astrocytes in the cerebellar cortex and the pyramidal tracts of the newborn rat. An immunofluorescence study with antibodies to a protein specific to astrocytes. Brain Res., 4 9 : 393402. 1974a Astrocyte-specific protein and neuroglial differentiation. An immunofluorescence study with antibodies to the glial fibrillary acidic protein. J. Comp. Neur., 1 5 3 : 27-38. 1974b The development of Bergmann glia in mutant mice with cerebellar malformations: Reeler, staggerer and weaver. Immunofluorescence study with antibodies to the glial fibrillary acidic protein. J . Comp. Neur., 1 5 5 : 219-230. Das, G. D., G. L. Lammert and J . P. McAllister 1974 Contact guidance and migratory cells in the developing cerebellum. Brain Res., 6 9 : 13-29. Rakic, P. 1971 Neuron-glia relationship during granule cell migration in developing cerebellar cortex. A Golgi and electronmicroscopic study i n Macacus rhesus. J. Comp. Neur., 141 : 2 8 s 312. Rakic, P., and R. L. Sidman 1973 Sequence of developmental abnormalities leading to granule cell deficit in cerebellar cortex of weaver mutant mice. J . Comp. Neur., 1 5 2 : 103-132. Ramon y Cajal, S. 1911 Histologie du Systeme Nerveux de YHomme & des VertkbrCs. Two vols. Pans, Maloine. 1960 Studies on vertebrate neurogenesis. Translated by L. Guth. Springfield, Illinois, Thom as. Sauer, F. C. 1935 The cellular structure of the neural tube. J. Comp. Neur., 6 3 : 13-23. Sotello, C., and J. P. Changew 1974 Bergman fibers and granular cell migration in the cerebellum ofhomozygous weaver mutant mouse. Brain Res., 77: 4 8 4 4 9 1 .
PLATES
PLATE 1 EXPLANATION OF FIGURES
Photomicrographs illustrating the postulated two phases in the translocation of developing granule cells from the external germinal layer to the granular layer. From sagittal sections of the vermis of a single 10-day-old rat. Cresyl violet, oil immersion. X 900.
440
8-10
Descending single processes issuing from vertically oriented pearshaped cells at the base of the external germinal layer.
11-13
Spindle-shaped cells apparently translocating through the molecular laver.
DEVELOPING C E R E B E L L U M . I V Joseph Altman
PLATE 1
44 1
PLATE 2 E X P L A N A T I O N OF FIGURES
Appearance of the molecular layer in 10 mp-thick sections stained with Bodian’s protargol-S method which tends to tint the parallel fibers. Adult rats. x 312.
442
14
Section cut sagittally through the pyramis. Note the parallel fibers cut transversely ( h e dots) and the vertically oriented free spaces. The latter art?presumed to represent glial palisades.
15
Section cut coronally through the pyramis of another adult rat. The parallel fibers are cut parallel (or slightly obliquely) and there are no vertically oriented columns visible. Presumably the thin ( 2 4 m p ) columns are obscured by parallel fibers situated below and/or above them.
DEVELOPING CEREBELLUM, IV Joseph Altman
PLATE 2
443
PLATE 3 EXPLANATION O F FIGURES
Glial palisades at three levels. Selected levels from serial horizontal sections through a v e r n a l lobule in an adult rat. Bodian. X 312.
444
16
Interface of pia (note blood vessel and capillaries) and surface of molecular layer.
17
Middle portion of molecular layer.
18
Lower portion of molecular layer with impregnated basket cell axons oriented transversely to the glial palisades. Associated with an increase of cell bodies a t this level (basket and Bergmann glia cells?) the glial palisades become gradually less distinct at the base of the molecular layer.
DEVELOPING CEREBELLUM. I V Joseph Altmnn
PLATE 3
PLATE 4 E X P L A N A T I O N OF F I G U R E S
Realignment of glial palisades in sagittal sections of the vermis of X-irradiated rats i n association with the reorientation of parallel fibers. The cerebella were irradiated during infancy with several successive daily doses of 150 r X-ray which results in decimation and subsequent regeneration of the external germinal layer. Adult rats. Bodian. X 216.
446
19
Cerebellum irradiated with two successive doses on days 3 and 4. Note that except at the base of the molecular layer the alignment of parallel fibers a n d glial palisades is rssentially normal.
20,21
Cerebella received four successive doses on days 3, 4, 5 and 6. Note that the palisades are recognizable only in zones of the molecular layer in which the parallel fibers a r e cut transversely. Two zones may be identified in figure 20, three zones i n figure 21.
D E V E L 0 I’ I NG C: k: K E B E LL U M , I V Joseph h l t m a n
44 7
