the fine structure of the lateral vestibular nucleus … · cytology of the nucleus, the fine...

THE FINE STRUCTURE OF THE LATERAL

VESTIBULAR NUCLEUS IN THE RAT

I. Neurons and Neuroglial Cells

CONSTANTINO SOTELO and SANFORD L. PALAY

From the Department of Anatomy, Harvard Medical School, Boston, Massachusetts 02115. Dr.Sotelo's present address is Laboratoire de Biologie Animale, Facult6 des Sciences de Paris,Paris, France

ABSTRACT

The lateral vestibular nucleus consists of multipolar isodendritic neurons of various sizesThe distal segments of some dendrites display broad expansions packed with slender mito-chondria and glycogen particles. These distinctive formations are interpreted as beinggrowing tips of dendrites, and the suggestion is advanced that they are manifestations ofarchitectonic plasticity in the mature central nervous system. Unlike large neurons else-where, the giant cells (Deiters) contain small Nissl bodies interconnected in a dense mesh-work. The Nissl substance is characterized by randomly arranged cisterns of the endoplasmicreticulum and by a high proportion of free ribosomes. Whether attached or free, ribosomesusually cluster in groups of four to six, and larger polysomal arrays are rare. Free ribosomalclusters also occur in the axon hillock and the initial segment. The neuronal perikaryacontain distinctive inclusions consisting of a ball of neurofilaments enveloped by a complexhoneycombed membrane. The failure of these fibrillary inclusions to stain with silver sug-gests that the putative argyrophilia of neurofilaments may reside in an inconstant matrixsurrounding them. Giant cells of Deiters are in intimate contact with two kinds of cellularelements-astroglial processes and synaptic terminals. Oligodendroglial cells are onlyrarely satellites of giant cells; in contrast, they are frequently satellites of small and medium-sized cells.

INTRODUCTION

The lateral vestibular nucleus has recently at-tracted a great deal of attention from investigatorsconcerned with integrative activity in the centralnervous system and with chemical changes asso-ciated with nervous function. In view of thisinterest, it is important for neurobiologists to havea clear idea of how the nucleus is organized: whatare the characteristics of its constituent neurons,what is the relationship between the neurons andthe supportive neuroglial cells, and what are thedistinguishing features of the various afferents

and their distribution? This nucleus receives

important afferent fibers from the vestibular

apparatus, the cerebellum, the reticular formation,and the spinal cord, and it gives rise to a massiveefferent bundle that ends in the spinal cord. Thenucleus thus contributes strategically to the controlof muscular coordination and the maintenanceof equilibrium. The variety of afferents in thenucleus offers the cytologist an opportunity tostudy the different characteristics of nerve ter-minals from many sources. As the nucleus contains

151

both extremely large nerve cells (the giant cells of

Deiters) and smaller cells, it also offers an op-

portunity to analyze the connections made withdifferent kinds of nerve cells.

Much is already known about the topographic

organization of the nucleus in terms of the distribu-

tion of afferent nerve fibers, but knowledge of the

interrelations between neurons and neuroglial

cells, or between these cells and the different kinds

of nerve terminals, is fragmentary. In the present

work, we have attempted to understand the

organization of the lateral vestibular nucleus at

the fine structural level. By means of optical and

electron microscopy as well as experimentalneuroanatomical techniques, we have made a

correlated analysis of the cells and nerve fibers.

Our results will be presented in a series of papers

of which this is the first. Here we present the

cytology of the nucleus, the fine structure of its

constituent neurons and neuroglial cells, and their

processes. In later communications, we shall

describe the morphology of the terminals in thenucleus and alterations in them after experimental

interruption of the afferent pathways.

MATERIALS AND METIIODS

The brains of 21 normal female rats,1 weighingfrom 150 to 300 g, were fixed by perfusion throughthe ascending aorta according to a method previouslydescribed (27). Each animal was anesthetized by anintraperitoneal injection of 35 mg of chloral hydrateper 100 g body weight. Four rats were perfused withwarm balanced salt solution (18) followed by 2,Os0 4 containing 0.5 g calcium chloride per 100 mlof solution and either 1% polyvinylpyrrolidone or6% dextran, buffered with 0.07 M acetate-veronal(pH 7.1-7.4). Thirteen rats were perfused first with100 150 ml of a warm solution containing 1% para-formaldehyde, 1.25%, glutaraldehyde, and 0.05c,calcium chloride, and buffered with 0.07 M sodiumcacodylate (pIt 7.2 7.4) [modified from Karnovsky(16)]. This solution was followed by 200 ml of 2~,Os04 containing 0.5/`6 calcium chloride and bufferedwith 0.1 M sodium cacodylate. The four remainingrats were perfused with about 500 ml of a warm solu-tion of 0.12 M phosphate buffer (pH 7.4) containingthe following compounds in each 100 ml: paraformal-dehyde, I g; 25%/, glutaraldehyde, 4 ml; calciumchloride, 2 mg. Fixation in this solution was pro-longed for several hours, and blocks taken from thesebrains were postfixed by immersion in osmium tetrox-ide as described below.

1 Obtained from Charles River Breeding Labora-tories, North Wilmington, Massachusetts.

The best results were obtained with the method ofdouble perfusion with aldehydes and osmium tetrox-ide. All the electron micrographs shown in this paperwere taken from material fixed in this way, exceptfor Fig. 22, which was derived from tissue perfusedwith aldehydes alone.

Thin slices of the brain stem were cut in the coronalplane and postfixed in fresh cold 2% osmium tetrox-ide solution for 2-3 hr. During this period, the lateralvestibular nuclei in the slices were identified underthe dissecting microscope and trimmed. After com-plete dehydration in ascending concentrations ofmethanol, tissue blocks were embedded in Epon 812or in Araldite. After polymerization, ultrathin sec-tions were made with a glass knife on a Porter-BlumMT-2 microtome. These sections were mounted oncopper grids, with or without a Formvar-carboncoat, and were double-stained with a saturatedaqueous solution of uranyl acetate followed by leadcitrate (40). Sections were examined in an RCA EMU3G electron microscope. Thicker sections (1-2 ,) forlight microscopy were also cut from the same blocksand were stained with toluidine blue.

Blocks of tissue obtained from the brain stems ofother rats were prepared for silver staining by Cajal'sneurofibrillary method (31). Additional silver prepa-rations were made according to a slight variation ofthe Golgi-Rio Hortega method (34) as follows. Severaladult rats were perfused with the modified Karnovskyaldehyde solution, followed by a freshly preparedaqueous solution of 6% potassium dichromate, 6ochloral hydrate, and 4% formaldehyde. After theperfusion, brain stem slices 3 mm thick and containingthe lateral vestibular nucleus were postfixed byimmersion in the same Golgi-Rio Hortega fixativefor 48 hr, with one change of fixative after the first24 hr. The slices were impregnated in 1.5% silvernitrate for 3 days. Sections 80-100 u thick were cutwith a freezing microtome and after dehydrationwere mounted in Permount under cover glasses.

OBSERVATIONS

Situated in the ventral lateral wall of the IVth

ventricle, the lateral vestibular nucleus occupies a

broad zone that is limited rostrally by the superior

vestibular nucleus, caudually by the descending

vestibular nucleus, laterally by the inferior cerebel-

lar peduncle, and medially by the medial vestib-

ular nucleus. The ventral boundary of this nucleus

is formed by the trigeminal complex and the

reticular formation, while dorsally the nucleus

extends along the medial margin of the inferior

cerebellar peduncle into the cerebellum. Although

the nucleus is readily demarcated in sections, the

constituent neuronal cell bodies are not compactly

aggregated, but are dispersed into small groups of

152 THE JOURNA. OF CELL BIOLOGY · VOLUME 36, 1968

three or more cells interlaced with neuropil and

numerous heavily myelinated fibers. Perikarya of

neuroglial cells, both astrocytes and oligodendro-

cytes, are relatively sparse.The neurons are usually multipolar. They can be

classified into three types: the giant cells of Deiters,

medium-sized neurons, and small neurons. In therat, as in the cat (3), the caudal part of the nucleus

contains more giant cells than does the rostralpart. In the nucleus as a whole, however, medium-sized and small neurons are more numerous than

the giant cells that are considered to be typical ofit.

The neurons of this nucleus belong to the

category designated as "isodendritic" by Ram6n-Moliner and Nauta (29), and all of the cells havea similar configuration regardless of size (Fig. 1).

They are elongated polyhedrons with long, fairlystraight dendrites. These give rise to relatively

few branches, which leave at acute angles to the

thick main stem. The dendrites extend for long

distances in all directions throughout the nucleus;for example, dendrites of neurons in the dorsalpart of the nucleus reach as far as the ventralpart and dendrites originating in the ventral partreach into the dorsal part. Dendritic thorns orspines are infrequent, even on the terminalbranches.

As the cytology of the neurons in the lateralvestibular nucleus resembles that of neuronsgenerally, the following description will emphasizeonly those features that seem to be peculiar tothis nucleus or that have not been described before,even though generally present in other neurons.

The Giant Cells of Deiters

In these cells, the nucleus is usually large,spherical, and excentrically placed. Its contour issmooth and free of the indentations and folds thatmark the nuclei of some large neurons in otherregions of the nervous system. The nuclear enve-lope has a gently undulating profile only rarely

FIGURE 1 Light micrograph of a Golgi-Rio Hortega preparation of the lateral vestibular nucleus in anadult rat. X 280.

C. SOTELO AND S. L. PALAY The Lateral Vestibular Nucleus. 1 153

FIGURE Electron micrograph of the cytoplasm of a giant cell of Deiters, showing the small masses ofNissl substance (N) interconnected to form a coarse network. In the remaining cytoplasm, containingmitochondria and the Golgi apparatus (G), neurofilaments and microtubules are disposed in arcs aroundthe Nissl bodies. X 10,000.

studded with attached ribosomes. The nuclearcontent consists of fine filaments, disposed singly orin pairs, and associated granules varying in sizefrom that of ribosomes to twice as large. Althoughmost of the nuclear chromatin is rather uniformlydispersed, small blocks of it occur in the form ofcoiled strands densely aggregated against thenuclear envelope. The nucleolus, as in mostneurons, forms a prominent globular body lyingoff-center in the nucleus. It consists of coarse,dense granules compacted into a thread that isloosely coiled upon itself. Open spaces between theturns of this nucleolonema can be large enough tobe discernible in the optical microscope as round,pale inclusions within the nucleolus.

In preparations stained for optical micropcsoy,the cytoplasm of the giant cells is characterized bylarge Nissl bodies similar to those in motoneurons.In electron micrographs of thin sections, however,the masses of Nissl substance are small. They areinterconnected by fine strands of the same compo-sition into a coarse network that is dispersedthroughout the cytoplasm except for a narrowband beneath the surface of the neuron. Thediscrepancy between these two appearances isprobably explained by the thickness of the sectionused for optical microscopy, which permits manynodal points in the network to overlap within theimage plane. The fluffy texture of the Nissl bodiesin the thick sections is probably the counterpartof the reticular dispersion of the Nissl substanceas seen in thin sections. The cisterns in the endo-plasmic reticulum of the Nissl substance are oftendisposed in parallel array (Fig. 6), but commonlythey are randomly arranged, with numerousbranchings or anastomoses between neighbors(Fig. 2). This arrangement conforms to the usualpattern in the small Nissl bodies of most neurons,and it is particularly well shown in the strands ofNissl substance interconnecting the larger masses.

In between the cisterns of the endoplasmicreticulum, large numbers of ribosomes are sus-pended in clusters of four to six granules. Only asmall proportion of the ribosomes are attached tothe outer surfaces of the cisterns, and most of thecisterns are free of ribosomes. Therefore, the endo-plasmic reticulum in these nerve cells can be con-sidered as generally smooth (or agranular), eventhough the surrounding matrix abounds in ribo-somes. The polysomal arrays attached to themembranes consist of straight chains or shortspirals of four to six ribosomes. The relatively

small number of ribosomes in the polysomes of thegiant cells in the rabbit has recently been pointedout by Ekholm and Hyd6n (7).

The cytoplasm between the Nissl bodies con-tains the usual organelles: Golgi apparatus,mitochondria, lysosomes, microtubules, neuro-filaments, pigment granules, and vesicles of varioussizes. The Golgi apparatus is clearly displayedsurrounding the nucleus and extending into thelargest dendrites. As can be seen in Figs. 3 and 8,the flattened cisterns of the Golgi apparatus areassociated with numerous vesicles and clumps ofmultivesicular bodies. Many of the vesicles in theGolgi region are "coated" or "alveolate," andmany of the smaller ones contain a dense sphericalinclusion about 500 A in diameter. The significanceof these inclusions is not clear.

The mitochondria are generally small, varyingfrom 0.1 to 1.5 in diameter, and are elongated inform. Their ultrastructure is similar to that ob-served in other neurons. Occasionally, a mito-chondrion in the peripheral cytoplasm of theneuron is associated with a subsurface cistern, ashas been reported by Rosenbluth (35) in spinalcord neurons and by Herndon (12) in Purkinjecells. Although this pairing of a mitochondrionwith a subsurface cistern often occurs beneath anaxon terminal attached to the cell membrane ofthe neuron (Fig. 5), this relationship is by nomeans exclusive. This association of organellesalso occurs beneath the neuroglial investment ofthe cell. It should be pointed out here that, sincethe giant cells of Deiters are covered only bysynaptic terminals and astroglial processes (seepage 166), the apparent association of thesestriking organelles with either axon terminals orneuroglia is merely a matter of chance.

The cytop]asmic matrix of the giant cells isoccupied by numerous microtubules and neuro-filaments that are disposed in arcs swirling andstreaming around the more massive Nissl bodiesand mitochondria. A certain amount of order isdiscernible in these zones. The filaments andtubules often run parallel with one another, form-ing arrays in which the strands are equidistant.Sometimes a short length of tubule or filamentlies athwart the main stream of parallel elements,indicating that some of them undergo gradualtwisting or spiraling. In the zones occupied by thesestructures there are also very fine granules, thinweblike strands, and pale filaments, which prob-ably are all images of the same structure captured


in different planes of section. The granules haveless than one-third the diameter of a ribosome andare much less dense. Their size, pallor, and fuzzyoutlines make them difficult to measure with anyprecision. Many granules and strands adhere tothe microtubules and neurofilaments (Fig. 20),producing a semblance of periodicity, but it isunclear whether these subsidiary structures shouldbe considered as intrinsic to the filaments andmicrotubules or as merely a kind of precipitateupon them.

The number of microtubules in a given field ofcytoplasm is inversely proportional to the numberof neurofilaments. In some neurons microtubulespredominate, whereas in others the neurofilamentspredominate. Although these differences betweenneurons can indicate different functional states, itmust also be noted that there are considerabledifferences from one field to another within thesame cell. For example, Fig. 3 shows the perinu-clear zone of a giant neuron with numerous neuro-filaments coursing among the Nissl bodies andGolgi apparatus. A similar region of a cell shownin Fig. 4 contains a centriole in addition to theother organelles, and in this micrograph there arenumerous microtubules concentrated at the rightwhere they radiate from a central point. This isprobably the edge of the centrosphere associatedwith the partner of the centriole in the middle ofthe picture. Thus, local variations in the distribu-tion of cell organelles can give rise to misleadingimpressions of the balance between neurofilamentsand microtubules in a given cell.

Besides being dispersed throughout the cyto-plasmic matrix, the neurofilaments participate inthe formation of a peculiar inclusion that occurscommonly in the giant cells of Deiters and is heredescribed for the first time. When once discoveredin electron micrographs, this inclusion was readilyrecognized in 1-/g sections stained with toluidineblue and examined in the optical microscope. Inthe latter preparations (Fig. 12), it appears in thecytoplasm as a round or oval clear area sometimescontaining a dense central spot. As the inclusion

does not stain, it is easily overlooked or rejectedas an artifactitious vacuole. In electron micro-graphs of thin sections, it appears in its simplestform (Fig. 9) as a circular area filled with long,slender threads about 100 A in diameter, to whichdelicate, fluffy material adheres at regular inter-vals. The filaments run roughly parallel with oneanother and are set apart by equal distanceslike the neurofilaments in the rest of the cytoplasmand in axoplasm. In these inclusions, however, thefilaments are arranged in bundles gently coiled o,braided about one another (Figs. 9, 13, 14, and 17).In three dimensions the inclusions would appearas if filled with a knot of coarse twisted twine. Asthese inclusions contain a highly concentrated andalmost pure population of neurofilaments, theyoffer an unusual opportunity to test the argyro-philia of these filaments. In sections prepared ac-cording to Cajal's reduced silver method, however,the inclusions remained recognizable but un-stained, although neurofibrillae in the neurons andtheir processes were intensely impregnated(Fig. 10).

Some of the inclusions contain a dense core con-sisting of closely packed filaments (Figs. 11, 14,and 19). In these regions, the filaments are so ir-regularly interlaced that it is impossible to be surethat all of the punctate images represent transversesections of filaments rather than granules. Anotherkind of filamentous material occurs rarely in theinclusion. This consists (Fig. 14) of threads radiat-ing from a dense, finely granular central mass, andthe threads themselves appear as rows of alternat-ing densities, suggesting a helix. Not enough exam-ples of this material have been encountered topermit us to give a more precise description. Otherrare components of the inclusion are small numbersof clustered ribosomes (Figs. 14, 15, and 19) andsmall vesicles with either clear or dense centers(Figs. 9 and 14).

These fibrillary inclusions are delimited from thecytoplasmic matrix by a wall of varying com-plexity. In the simplest forms, the wall is a continu-ous single membrane about 70 A thick (Fig. 9), but

FIGURE 3 Electron micrograph of the perinuclear region of a giant cell of Deiters, showingthe Golgi apparatus (G). Granular vesicles and alveolate vesicles (arrows) are numerous inthis region X 15,000.

FIGURE 4 Electron micrograph of the cytoplasm of a giant cell showing the Golgi ap-paratus (G) and a centriole (c). At the right, the numerous microtubules are probably partof the centrosphere (s) associated with the partner of this centriole. X 13,600.

156 THE JOURNAL OF CELL BIOOGY VOLUME 36, 1968

C. SOTELO ANI) S. L. PAIAY The Lateral Vestibular Nucleus. 1 157

in many regions it consists of appressed tubularstructures that give it a multiloculated, honey-combed appearance (Figs. 11, 14, and 16-19).Some inclusions are surrounded completely by oneor more such honeycombed envelopes (Figs. 14, 16,and 18). The limiting membrane does not appearto be continuous with either the rough or smoothendoplasmic reticulum in the surrounding cyto-plasm.

The inclusions vary in diameter from 0.2 to 6 u.Several are frequently encountered in any one sec-tion through a cell, and they can be distributed asisolated individuals anywhere in the perikaryon.Usually, however, they tend to be clustered to-gether, interconnected through peduncles formedby their complex envelope. In such instances, theindividual inclusions often lose their sphericalshape, becoming elongated and irregular, as ifcompressed together (Fig. 11). Even isolated in-clusions have small satellites of the same composi-tion attached to, or included in, their envelopes(Fig. 17).

Neurons of Medium and Small Size

As these neurons resemble the giant cells ofDeiters in most particulars, the following remarkswill be devoted to observations of differences be-tween the two types of cells. One of the most im-pressive differences, one that is visible even in theoptical microscope, is that the nuclei of the smalland medium-sized cells occupy a much largerproportion of the cross-sectional area of the cellbody. The nuclear envelope is deeply invaginatedby cytoplasm filled with clustered ribosomes and afew vesicles (Fig. 23). The circular profiles of thefolds in transverse section show that they areusually long, finger-shaped indentations, ratherthan broad wrinkles in the surface of the nucleus.

Occasionally, three types of peculiar nuclearinclusions have been encountered: (1) bundles oflong, thin filaments, similar to those described byGray and Guillery (9) in ganglion cells of the leechand by Siegesmund et al. (37) in the granule cellsof mammalian olfactory bulb and cerebellum; (2)thin sheets of fine parallel filaments arrayed in alattice, similar to those reported by Chandler andWillis (4) in neurons of cerebral cortex, thalamus,and superior colliculus; and (3) bouquets of long,parallel microtubules about 270 A in diameter, asshown in Fig. 24. The first two types have also beenseen in the nuclei of cultured dorsal root ganglioncells subjected to X-irradiation (17). The thirdtype of inclusion has not been reported before. Asthe microtubules in it are associated with fine fila-ments, this inclusion may be related to the othertwo. It is not clear whether any of the inclusionscorrespond to Cajal's "batonnets intranucleaires"(30), as Siegesmund et al. have suggested. In elec-tron micrographs they appear to be very rare,whereas Cajal's comments indicate that the baton-nets are common. Only one example of these nu-clear inclusions has been observed in the giantcells-a bouquet of microtubules and fine filamentscorresponding to the third type. As the incidenceof the inclusions, even in the smaller cells, is notlarge, they may have been missed.

In the cytoplasm of the small and medium-sizedneurons, the Nissl bodies are small and widelydispersed throughout the perikaryon. In contrastto the characteristic pattern in the giant cells,Nissl bodies frequently appear in the mostperipheral cytoplasm (Fig. 23), where they under-lie either synaptic terminals or neuroglial processesadjacent to the surface of the neuron.

Although the association of subsurface cisternsand mitochondria is rarer in the smaller cells than

FmuIG E 5 Peripheral region of a giant cell, showing the association of a mitochondrionwith a subsurface cistern (arrow). X 25,000.

FIGUuRE 6 Nissl body of a giant cell with the endoplasmic reticulum disposed in parallelarray. Most of the ribosomes are free, arranged as small polysomal clusters. X 26,000.

FIGURE 7 Oblique section through a cilium (cil) of a medium-sized neuron. X 26,000.

FIGuRE 8 The most peripheral region of a giant cell is devoid of Nissl substance, beingoccupied by free ribosomes, mitochondria, and arrays of parallel neurofilaments and micro-tubules. The Golgi apparatus (G) is often associated with clusters of multivesicular bodies(my). X 13,000.

158 TIIE JOURNAL OF CELL BIOLOGY - VOLUME 36, 1968

C. SOTELO AND S. L. PALAY The Lateral Vestibular Nucleus. I 159

in the giant cells, stacks of several closely apposedcisterns beneath the surface are more frequent.These groups of cisterns, with the last memberstudded with ribosomes (Fig. 25), resemble thosedescribed by Rosenbluth (35) in the pyramidalcells of the cerebral cortex.

Centrioles and cilia (Fig. 7) are encounteredmore frequently in the smaller cells than in thegiant ones. In transverse section (Fig. 26), it can beseen that the cilium contains eight doublets in itsperiphery and one off-center, as described by Dahl(5). Isolated fibrillary inclusions, like those foundin the giant cells, also occur in the cytoplasm ofsmall and medium-sized neurons.

Axon Hillock

Fig. 21 shows a longitudinal section through theaxon hillock and initial segment of the axon be-longing to a giant cell of Deiters. The axon leavesthe perikaryon by a funnel-shaped process, nar-rows to a slender fiber about 2 A across and about27 u long, which is coated by thin neuroglial proc-esses, and then enters the first segment of its myelinsheath. The internal structure of this axon istypical of the initial segments of both giant cellsand smaller cells, and they conform to the patternseen in other multipolar neurons (28). In thisstudy, we have found the axon hillock of 30 differ-ent cells having the axon continuous with theperikaryon, and in two of these the entire lengthof the initial segment from hillock to myelin sheathwas included in the micrographs. In one case, wehave found the point of emergence of the axonfrom a large dendrite, as Cajal (30) has describedin Golgi preparations. We have not, however,

encountered initial segments in transverse section,a failure that is probably merely fortuitous, for wehave found them in oblique sections.

The axon hillock and initial segment character-istically contain sheaves of parallel microtubuleswhich collect in the broad portion of the hillock(Fig. 22) and extend into the axon until the begin-ning of the myelin sheath. Here they either termi-nate abruptly or disperse. The microtubules ap-pear to be bound together by dense materialdistributed like bridges at irregular intervals alongtheir length. The close association of these tubulesinto bundles or ribbons produces, in longitudinalsections, an appearance of extreme density dueto the fact that two or three microtubules overlapwithin the thickness of the section. The bundles,consisting of three to six microtubules, are dis-tributed across the axon without any definite pat-tern. Although individual microtubules could befollowed for 2-3 p along the length of the axon, wefound none that shifted from one bundle to an-other. An extensive series of sections in the trans-verse plane would be necessary for one to ascertainwhether a particular microtubule always remainswith its primary bundle.

Aside from the microtubules, the initial segmentcontains many elongated mitochondria generallyarranged with their long axes parallel to the axonand especially congregated in the axon hillockregion (Fig. 22). There are also many neurofila-ments and tubules of endoplasmic reticulum,which are drawn out longitudinally, parallel withthe other structures.

Contrary to what might be expected from prepa-rations stained with basic dyes, the axon hillock is

FIGmUE 9 Simplest form of the fibrillary inclusions in the neurons of the lateral vestibularnucleus. The arrow points to a part of their wall formed by only a single membrane. Acluster of ribosomes (r) is found in one of these bodies. X 13,000.

FIGURE 10 Light micrograph of a giant cell of Deiters stained according to the Cajalneurofibrillary method. Neurofibrils in the perikaryon and dendrites are well impregnated.The arrow points to the negative image of a fibrillary body, which has not been impreg-nated. X 900.

FIGURE 11 Large aggregate of fibrillary inclusions in a giant cell. The inclusions have losttheir spherical shape, but retain their individuality, each one being surrounded by a wall.Some of the inclusions contain a dense core (D) formed by closely packed filaments. X33,000.

FIGURE 12 Light micrograph of a giant cell stained with toluidine blue. Three inclusions(arrows) appear as unstained vacuoles; two of them contain a dense central spot. X 1,500.

160 THE JOURNAL OF CELL BIOLOGY · VOLUaE 36, 1968

C. SOTE.LO AND S. L. PALAY The Lateral Vestibular Nucleus. I 161

FIGURE 13 Peripheral region of a Deiters cell showing two fibrillary bodies (F) and one axon terminaldeeply invaginated into the cytoplasm and suggesting an intracellular axon terminal. X 19,000.

not devoid of ribosomes. It is true that large aggre-gates and complete Nissl bodies do not encroachupon the hillock (Fig. 22), yielding a relativelyclear zone (occupied by mitochondria, neurofila-ments, and microtubules) from which the axonarises. But small clusters of ribosomes, usually inpolysomal array, do extend into the hillock and theinitial segment, becoming more and more sparseuntil the region of the myelin sheath is reachedwhere they disappear altogether (28). The ribo-somes are not usually attached to the membranesof the endoplasmic reticulum.

The surface of the axon in its initial segment isspecialized by the presence of an undercoating offine granules. The coating first appears at thepoint where the axon hillock narrows to form thebeginning of the initial segment and disappearswhen the axon enters the myelin sheath. Occa-sionally, alveolate vesicles or pits occur in thesurface of the initial segment, as well as smallthornlike projections. Small presynaptic terminal

boutons are attached to the initial segment, gen-erally near its origin from the cell body. Theundercoating of punctate axoplasmic material isabsent under the part of the membrane that liesopposite these synaptic terminals.

The Neuropil

The greater part of the nucleus is occupied bymyelinated fibers, some of which approach closeto the cell bodies and give off slender preterminalunmyelinated branches or end abruptly as largebulbous terminals. Many of the myelinated fibers.however, are fibers of passage on their way to thecerebellum and to the other vestibular nuclei.Among these fibers and neuronal perikarya aresmall islands of neuropil consisting of dendrites,unmyelinated axons with their associated synapticterminals, and neuroglial cells. As in other regionsof the mammalian central nervous system, the ex-tracellular space is represented by an interval ofabout 150 A between the limiting membranes

162 THE JOURNAL OF CELL BIOLOGY · VOLUME 36, 1968

FIGURE 14 Complex fibrillary bodies displaying several kinds of filamentary arrangments. The arrowpoints to filamentous material forming threads radiating from a dense central mass. In one of these in-clusions, clusters of ribosomes (r) are present. This micrograph also shows the tubular structure of the wallof the fibrillary inclusions. X 7,000.

of the cellular elements that compose the neuro-pil.

As the synapses on the surfaces of the perikarya

and dendrites vary considerably in their form andcomplexity, the full description of them will be re-

served for the second paper of this series. In thepresent paper, we shall confine our attention to thestructure of dendrites and axons coursing throughthe neuropil. Fig. 20 shows a longitudinal sectionof a dendrite in which microtubules, neurofila-ments, and agranular endoplasmic reticulum canbe distinguished. The endoplasmic reticulum oc-curs in a tubular form whose irregular swellingsdeceptively appear to be continuous with some ofthe microtubules. This appearance is fairly fre-quent in longitudinal sections of dendrites and hasgiven rise to some confusion (24). Two differentlongitudinal and tubular structures are involvedhere. When they overlap in the thickness of a sec-

tion, they can give the impression of continuity.Fig. 20 shows several points of overlap, and thearrow indicates one in which the boundary sepa-rating the microtubule from the endoplasmicreticulum is clear. This interpretation has alreadybeen advanced by Metuzals (19) in a study ofaxons in the diencephalon of the frog.

The dendrites of the lateral vestibular nucleusdisplay certain peculiar features of general interest.Longitudinal sections (Fig. 32) show that some ofthe dendrites suddenly enlarge to form varicositiesfilled with mitochondria. The swellings vary from1 to 8 / in diameter, and some are even larger. Innearly every instance, their dendritic nature isconfirmed by the presence of axonal terminalssynapsing upon them (Figs. 28, 30, and 31). Inaddition, a few leaflike or club-shaped processesare given off from these varicosities to make synap-tic contact with axonal terminals in the surround-

C. SOTELO AND S. L. PALAY The Lateral, Vestibular Nucleus. 1 163

ing neuropil (Figs. 31 and 32). The mitochondriain these dendritic varicosities are long and slender(0.2-0.25 /j in diameter) and generally containlongitudinally oriented cristae. They either lieparallel to the long axis of the dendrite or form agently swirling figure around the long axis (Fig.28). Usually, transverse sections show that themitochondria are remarkably uniform and each isin almost perfect alignment with its neighbors(Figs. 29 and 30). The number of mitochondriathat can be packed into such profiles is astonishing.In one instance, 134 mitochondria were countedinside a dendrite of 12.5 2 cross-sectional area, andin another instance 297 mitochondria werecrowded inside a dendrite of 26 u2.

Despite this crowding, there is room in the den-drite for other organelles: microtubules runninglongitudinally, agranular endoplasmic reticulumin the form of meandering tubules and small cis-terns, multivesicular bodies, and various types ofvesicles. Aside from the mitochondria, the moststriking inclusions in these dendritic profiles areglycogen particles, which vary from single, irregu-lar granules ( particles), 200-300 A in diameter,to rough clumps (a particles), 800-1000 A in diam-eter and consisting of four to eight individual 3particles (Fig. 29). In most instances, glycogenparticles are dispersed irregularly among the mito-chondria, but sometimes glycogen fills so much ofthe profile that the mitochondria are displaced tothe periphery, where they remain curiouslycrowded together (Fig. 31).

Frequently, mitochondria sequester glycogen

particles within deep, cup-shaped depressions inthe surface of the organelles. Three such distortedmitochondria are shown in Figs. 29 and 30, wherepossible stages in their formation can be seen. InFig. 29, mitochondrion I displays an armlike ex-tension partially surrounding a few glycogen par-ticles; mitochondrion 2 completely encloses acomparable cluster. These extensions appear toconsist of the inner and outer membranes of themitochondrial envelope and some included matrix,but no cristae. In Fig. 30, a field of glycogen par-ticles is surrounded by an incomplete ring formedby the mitochondrial envelope and the enclosedmatrix. These figures may be precursors of otherforms, more difficult to interpret with certainty, inwhich double-walled vacuoles containing mem-branous laminae resembling mitochondrial cristaeappear with or without enclosed glycogen particles.All of these figures can be distinguished from arti-factitious swellings or explosions of mitochondria,such as are associated with glutaraldehyde fixation.In the latter forms, the extensions from the mito-chondrial envelope involve only its outer mem-brane.

Neuroglia

Two types of neuroglial cells, astrocytes andoligodendrocytes, occur in the normal lateral ves-tibular nucleus. Since theultrastructure of these cellsis not unusual (22, 26, and 27), we shall devote ourattention to the relations between neurons andneuroglial cells.

A large proportion of the oligodendrocytes are

FIGURE 15 Clusters of ribosomes (r) in a fibrillary body. X 16,000.

FIGURE 16 Tangential section through the wall of the fibrillary inclusions showing itshoneycombed appearance. X 31,000.

FIGURE 17 Isolated fibrillary body with a small satellite included in its envelope. X19,000.

FIGURE 18 Multiloculated appearance of the envelope of several small fibrillary inclu-sions. X 25,000.

FIGURE 19 Large complex fibrillary bodies, one with a dense core (D) of closely packedfilaments, and another with clusters of ribosomes (r). X 25,000.

FIGnRE 20 Longitudinal section of a dendrite showing neurofilaments (nf), micro-tubules (mt), and agranular endoplasmic reticulum in tubular form (er). Some points ofoverlap between microtubules and endoplasmic reticulum can be seen; in one of them(arrow) the boundary separating the two structures is evident. X 40,000.

164 THE JOURNAL OF CELL BIOLOGY VOLUME 36, 1968

found in those areas of the nucleus that are oc-cupied almost exclusively by myelinated fibers. Inthe neuropil, extensions of the oligodendrocytesare rarely encountered. Furthermore, these neuro-glial cells are not usually found in immediate ap-position to a giant neuron in the lateral vestibularnucleus. In the rare instances in which an oligo-dendrocyte appears to be a satellite of a giant, orDeiters, cell, there is almost always a thin layer ofastrocytic cytoplasm interposed between the oligo-dendrocyte and the neuronal cell body (see Fig.22). Most of the surface of the giant cell is coveredby synaptic boutons, and the remainder is coveredby slender astrocytic expansions and myelinatedfibers. Generally, the oligodendrocytes in theneighborhood of a giant cell are located at a littledistance from the neuron and are separated from itby a narrow zone containing synaptic boutonsenveloped by lamellae of astrocytic cytoplasm.

Most of the perineuronal oligodendrocytes aresatellites of the small or medium-sized neurons.One of these is shown in Fig. 23, where an oligo-dendrocyte can be seen in close association withpart of the neuronal surface. The rest of the surfaceis covered by myelinated nerve fibers or lamellarexpansions of astrocytes.

The common neuroglial cell of the neuropil isthe protoplasmic astrocyte (Fig. 27), which sendsout numerous fine lamellar processes extendingbetween the other cellular elements. The lamellaeenfold dendrites and axons and insinuate them-selves into the crevices around the neuronal peri-karya, leaving small apertures for the entrance ofthe preterminal axons that synapse upon them. Asa result, nearly all neuronal surfaces are coated byone or more layers of astrocytic cytoplasm, exceptwhere synaptic contacts are made. Sometimes theastrocytic lamellae form spirally wound figures

(Fig. 27) that resemble immature myelin. Al-though the neuronal perikarya are usually sepa-rated by zones of neuropil, or at least by a thinlamella of astrocytic cytoplasm, there are occa-sional instances in which one neuron abuts directlyupon another with only the usual interstice of150 A between them. In these regions of immediatecontact, which can extend over several micra, theadjacent surface membranes display no specializedjunctional characteristics.

DISCUSSION

The advantage of an electron microscopic analysisof a small region in the brain is that it reveals de-tails of structure and intercellular relations thatare not discernible in optical microscopic studies.Electron microscopy does not replace opticalmicroscopy and its time-honored techniques, butextends the range and precision of the observer andlimits the number of possible interpretations. Itmay be anticipated that an electron microscopicstudy of a specific region will disclose structuralfeatures that, upon further study, prove to be gen-erally applicable to many other regions. Otherfeatures, however, may prove to be restricted tothe region or cell under study and may, therefore,provide valuable clues to its special function ororganization. Because of these encouraging possi-bilities, neuroanatomists are stimulated to continuetheir cytological exploration of the central nervoussystem at the fine structural level until the generalpattern is well filled out.

In the present paper, we have reported manysimilarities between the cells of the lateral vestibu-lar nucleus and those of other parts of the centralnervous system. It was not to be expected thatthese cells should differ in their essential structure

FIGURE 21 Origin of the axon of a giant cell. The initial segment, about 27 long and 2 across, extends from the axon hillock on the left to the beginning of the myelin sheath onthe right. The sheaves of parallel microtubules within the initial segment disappear whenthe axon enters the first segment of its myelin sheath. X 6,000.

FIGURE 22 Axon hillock of a giant cell of Deiters. Clusters of free ribosomes (r) arenumerous in this region, but Nissl bodies (N) are absent. The consequent pallor of thetransitional zone between the perikaryon and the initial segment of the axon is evident inthis electron micrograph as well as in photomicrographs. The arrow points to a thin layerof astrocytic cytoplasm located between the neuron and a neighboring oligodendrocyte (0).X 10,000.

166 THE JOURNAL OF CELL BIOLOGY - VOLUME 36, 1968

FIGURE 23 Medium-sized neuron. The nucleus shows some finger-shaped indentations. One oligodendro-cyte (0) is in a satellite position to this neuron. X 14,000.

168

FIGURE 24 Nuclear inclusion, in a medium-sized neuron, formed by a bundle of microtubules associatedwith fine filaments. X 26,000.

FIGURE 25 The arrow points to a stack of three closely apposed cisterns beneath the surface of a smallneuron. X 20,000.

from the general picture of neurons and neurogliaalready well established by earlier studies. Never-theless, the present report discloses new features ofnerve cells and their processes and, interestingly,these features, although not previously described,turn out to be present in other regions of the ner-vous system as well.

The Fibrillary Inclusions

One of these structures is the fibrillary inclusionbody found in both giant and smaller neurons.Although the nature of this inclusion is obscure,certain suggestions about its significance can beexcluded by our observations. The form of thesimplest inclusions and their impressive content ofneurofilaments immediately suggested that the in-clusions were axons that penetrated deeply into theperikaryon. An occasional micrograph like Fig. 13,

with a row of two or three inclusions leading to anapparently intracellular axonal terminal, rein-forced this suggestion. But further study revealedthe more complex profiles (Figs. 14, 16, 19), whichare very difficult to reconcile with the form of anaxon. Furthermore, examination of serial sectionsin the light microscope disclosed that the inclusionsare spherioidal or lobulated, are never attached tothe surface of the cell, and always begin and endwithin the neuron. Occasional images like Fig. 13must be interpreted as accidental coincidences of asynaptic terminal located in a depression in theneuronal surface and neighboring fibrillary inclu-sions. In addition to the neurofilaments, the inclu-sions also contain ribosomes (Figs. 14, 15, and 19)and dense aggregates of finer, perhaps helical fila-ments, both of which have not been found interminal parts of axons. Moreover, if these were


FIGURE 26 Parts of two adjacent medium-sized neurons. In the neuropil the arrow points to a transversesection of a cilium containing eight doublets in its periphery and one off-center. X 17,000.

axons, their profiles should be surrounded by twosurface membranes, the plasmalemma of the axonand the plasmalemma of the perikaryon. But thelimiting envelope of the inclusions does not re-semble the plasmalemma. It varies from a singlemembrane to a complicated honeycomb structure.All of these features make it clear that they are notaxons or parts of other cells invaginating the peri-karyon, but are truly intracellular inclusions.

These inclusions are not specific to the neuronsof the lateral vestibular nucleus. They have beenseen in the ventral cochlear nucleus, in the spinalcord2 , and in the posterior colliculus3. So far, they

have not been found in any species other than therat. Like many other neuronal inclusions that havebeen described (20, 21, and 39), the fibrillarybodies are mysterious. We now possess a muchlarger catalogue of cytoplasmic organelles than we

2 R. B. Wuerker. Personal communication.3 D. K. Morest. Personal communication.

have functions to be ascribed to them. The role ofthese structures awaits a more detailed under-standing of the integrated function of the nerve cellthan is now available.

According to investigations by Gray, Guillery,and Boycott (2, 8), the neurofibrillae shown byoptical microscopy in nerve cells, processes, andendings correspond to bundles of the neurofila-ments shown by electron microscopy. Theseauthors have presented strong evidence for theconclusion that the neurofilaments are responsiblefor the argyrophilia of neural tissue. It is, therefore,of some interest to examine the fibrillary inclusionsdescribed in the present paper in the light of theconclusions offered by Gray and his coworkers.Although filled with filaments morphologicallyidentical to neurofilaments elsewhere, these fibril-lary bodies failed to stain with Cajal's neurofibril-lary silver stain. If, on the one hand, the filamentsin these bodies are chemically identical with neuro-filaments elsewhere, then we must conclude that


FIGURE 27 Electron micrograph of an astrocyte containing bundles of fine filaments (f) and giving offseven thin processes within the field. At the left of center, a spiral lamellar arrangment of thin astroplasmicsheaths resembles immature myelin. X 26,000.

argyrophilia is not inherent in the filaments them-selves, but in a matrix that surrounds them. In thatcase, the matrix in the inclusions must differ fromthat around neurofilaments elsewhere. If, on theother hand, these filaments are chemically differentfrom other neurofilaments as shown by their lackof argyrophilia, then we must conclude that thereare different types of neurofilaments and that theirmorphology is not sufficient to establish either theirsimilarities or chemical differences. If the latterconclusion is correct, we can no longer assume thatall filaments found in nerve cells and having themorphological characteristics of neurofilaments(25) are chemically identical. Furthermore, thechemical characterization of the filamentous pro-tein extracted from squid axons (6, 36) can nolonger be generalized to other species or even toother nerve cells. This line of argument seems tolead in a pessimistic direction. Of course, it is possi-ble that neurofilaments may all be constructed ofthe same protein, but that in different situations itssurface may have different distributions of chargesor exposed radicals so that its reaction to the pro-cedures of silver staining would be different.Against this possibility is the great dependence ofthe conformation of proteins upon these precisedistributions of charges or exposed radicals. Themorphology of the filament is probably congruentwith its chemical constitution.

We are thus led back to the first alternative: thatthe argyrophilia is a property of the matrix sur-rounding the neurofilaments. In fixed and sec-tioned material, most neurofilaments (and micro-tubules) are surrounded by fuzzy weblike strandsand granules that may represent part of the matrix.But it is unlikely that this is the argyrophilicmatrix, for it coats the filaments in the fibrillarybodies as well as those elsewhere in the neuron. Theargyrophilic material may be located in the wall

of the filament itself and may not be resolvable byour methods, or it may not have been preserved.

The Terminal Segments of Dendrites

One of the impressive findings in this study of thelateral vestibular nucleus is the peculiar dendriticprofiles loaded with mitochondria and glycogen.These profiles are most abundant in the dorsal partof the nucleus, but are not restricted to it. Theyhave been seen in other regions of the nervous sys-tem, for example, in the cerebellar cortex and thehypothalamus4 and in the superior cervical gan-glion.' They have not, however, been observed sofrequently in these regions as in the lateral vestibu-lar nucleus. It is impossible to decide on the basisof currently available information whether thisdifference in frequency is due merely to the smallsample examined by electron microscopy or to lackof attention on the part of observers. Within thelateral vestibular nucleus, however, there is a cleardifference in the frequency of these profilesbetween the dorsal and ventral parts of the nu-cleus. Therefore, they must belong to the processesof only certain types of neurons and not to all. Fur-Lher work is necessary in order to identify the celltype that gives rise to these dendrites.

That these profiles represent sections throughdendrites is proved by images like Fig. 32 showingthat they are continuous with typical dendrites.The location of axonal terminals in presynapticrelation to their surface (Figs. 28 and 30) is furtherevidence in favor of this interpretation. It is lesseasy to ascertain which part of a dendrite is repre-sented by these profiles. They often appear as adilation or varicosity in the course of a dendritesuddenly crowded with mitochondria that are

4 S. L. Palay. Unpublished observations.T. H. Williams. Personal communication.

FIGUtRE 28 Dendritic expansion filled with mitochondria. The arrow points to an axonterminal synapsing on this dendrite. X 17,000.

FIGURE 29 Dendritic profile containing cup-shaped mitochondria with enclosed glycogenparticles. Mitochondrion 1 partially surrounds a few glycogen particles; mitochondrion 2completely encloses a similar cluster. X 25,000.

FIGURE 30 Part of a dendritic profile containing a distorted mitochondrion forming thewall of a vesicle with enclosed glycogen particles (arrow). An axon terminal (A) synapses onthis dendrite. X 31,000.

172 TaE JOURNAL OF CELL BIOLOGY VOLUME 36, 1968

FIGURE 31 Two dendritic profiles with the mitochondria displaced peripherally by large masses ofglycogen, which occupy most of the dendritic area. Fine finger-like processes (arrows) project from the sur-face of the lower dendrite and an axon terminal (A) synapses with it on the right. X 3,000.

FIGURE 32 Longitudinal section of a dendritic tip. The dendrite (Den) suddenly enlarges, forming avaricosity filled with mitochondria. Two fingerlike(arrows). X 29,000.

nicely aligned parallel to the longitudinal axis. The

dilated region is, however, so large that the areaincluded in a thin electron microscopic section

does not usually contain a further stretch of thedendrite supposedly narrowing as it leaves the vari-cosity. This circumstance suggests that many of

these profiles are actually ends of dendrites, a sug-gestion that is supported by the appearance ofsmall finger- or leaflike projections from their sur-

faces (Fig. 32). If not the actual terminations, theseprofiles must represent at least the distal segmentsof dendrites. Sections passing longitudinallythrough the origin of dendrites from the perikaryonnever display the dense aggregations of mitochon-dria and glycogen seen in these profiles. Further-more, the part of the dendrite continuous withthese profiles is generally narrow and free ofribosomes, characteristics of the more distal den-dritic branches.

It would be interesting to compare these electron

projection extend from the distal parts of the tip

micrographic appearances with the distal dendritic

branches seen in the optical microscope. Althoughthe high concentration of mitochondria in them

should make them rather easy to identify in thicksections, we have thus far been unable to distin-guish them, even at magnifications of 1500 times,from collections of presynaptic terminals andglomeruli. Descriptions of the forms of dendritic

tips in Golgi preparations are strangely elusive.Dendrites are described as branching repeatedly

and then fading from sight. Either the tips are notreally recognizable in the optical microscope orthey break up into short, thin strands. This vaguepicture could be explained by the content of these

peculiar profiles, if it is remembered that in the

Golgi preparations the precipitate of silver salts is

deposited in the cytoplasm of the nerve cell andits processes but does not impregnate the mito-chondria (1, 38). Thin lamellae of silver saltsaround or between the mitochondria could


produce the appearance of fraying out at the endof the dendrite. 6

The spectacular aggregation of mitochondriaand glycogen in these dendritic profiles poses thequestion of their function. What is the significanceof the high concentration of mitochondria in justthis part of the neuron? Nothing that we knowabout the physiology of dendrites points to anunusually high requirement for energy or otherproducts of oxidative phosphorylation. The asso-ciation of aggregates of mitochondria and abun-dant glycogen in these profiles is reminiscent of thecytoplasm of brown fat cells (23), where the mito-chondrial oxidative activity is uncoupled fromphosphorylation. In this example, the role of themitochondrion in heat production seems evident,but no such requirement is yet known in the ner-vous system. It is possible that the mitochondriaare actually inactive and have merely collected atthe tips of the dendrites as a result of protoplasmicstreaming in the dendrites. The reciprocal relationbetween quantities of glycogen and number ofmitochondria, as shown in Figs. 28 and 31, mayindicate different states of dendritic metabolism.These peculiar dendritic profiles introduce apuzzling problem that is difficult to investigatebecause of their small size and diffuse distributionin a complex tissue.

Another possibility is that this part of the den-drite is a growing or expanding tip of a growing orregenerating nerve cell. If this possibility is ad-mitted for consideration, then dendrites becomemore plastic and variable than they have beenthought to be. The presence of axonal terminalssynapsing upon the surfaces of these dendrites doesnot argue against the possibility, for growth of thedendrites may be a method of establishing newsynaptic contacts (see Footnote 6). Peculiar axonalterminals have also been seen in the lateral vestibu-lar nucleus, as well as in many other parts of thecentral nervous system (11), and it has been sug-gested that these may represent stages in a cycle of

6Through the courtesy of Dr. Kent Morest, ofthis Department, we have recently had the privilegeof examining dendrites in Golgi preparations of thelateral vestibular and the lateral geniculate nucleiof the cat. In young, growing kittens, the distalsegments of dendrites display irregular weblikeexpansions and varicosities from which thin terminalprocesses extend. In 6-wk-old animals such formationshave not been seen as yet, but only the lateral genic-ulate nucleus has been examined at this time.

degeneration and regeneration of nerve endings inthe normal animal. Both axons and dendritesmight engage in such a process, reflecting the con-tinued experience of the animal, making and dis-carding interneuronal connections by the activityof both pre- and postsynaptic elements. Althoughthis hypothetical interpretation is tempting to in-vestigators interested in finding a morphologicalsubstrate for the process of learning, it would bevery difficult to test experimentally.

A last possibility is that these profiles are path-ological. This seems to be unlikely because theirincluded mitochondria appear completely normal,and lysosomes are lacking. Furthermore, no ab-normal neurons or neuroglial cells have ever beenobserved in these specimens. Finally, the dendriticprofiles have been seen in many different parts ofthe brain and in a large number of animals fromdifferent sources.

Ribosomal Patterns in the Giant Cells

Ribosomes in the giant cells of Deiters displaytwo characteristics that are typical of these cells.First, they are more rarely attached to the mem-branes of the endoplasmic reticulum than in otherlarge neurons, for example, the Purkinje cell or themotor neuron. Second, the ribosomes are arrayedin a distinctive polysomal pattern consisting ofshort rows of granules attached to the endoplasmicreticulum or small clusters suspended in the matrixbetween cisterns. Although large polysomes arepresent, usually each polysome is composed of onlyfour to six ribosomes, considerably fewer than thereare in the polysomes of such neurons as thePurkinje cell. These results on sectioned materialagree with the findings of Ekholm and Hyd6n (7),who examined intact polysomes in fragments ofisolated cells prepared by microdissection andnegative staining.

Rich et al. (33) have clearly shown that the sizeof the polysome is related to the complexity of theprotein molecule synthesized. For example, thereticulocyte, which synthesizes hemoglobin, hassmall polysomes composed of five to six ribosomes.The messenger RNA that codes the polypeptidechain of hemoglobin requires only 450 nucleotides.In contrast, in the HeLa cell infected with polio-myelitis virus, where the synthesized protein isvery complex, the relevant messenger RNArequires 6000 nucleotides, and the polysomes inthis instance are composed of 60 ribosomes. If,then, the polysomal patterns of different cells can


be considered as reflecting the kinds and com-plexities of the proteins synthesized in these cells,it is interesting to notice the variety of polysomalarrays in different types of neurons. These patternsare not consistent with the size or volume of thecell, as the pattern of the giant Deiters cell is similarto that of cerebellar granule cells. The variety ofpatterns suggests that neurons may produce dis-tinctive proteins, specific to the cell type, or differ-ent proportions of similar proteins of various com-plexities.

Relation Between Neurons and

Neuroglial Cells

In recent years, Hyden and his collaborators(14, 15, 10) have carried out intensive neurochemi-cal investigations on the lateral vestibular nucleus,particularly on the giant cells of Deiters and theirassociated neuroglial cells. Hyden uses a delicatemethod of microdissection that permits him toisolate the giant neurons from their surroundingneuroglial cells and thus to analyze each independ-ently. In the first studies (14, 15), a giant cell wasthought to be enclosed by a neuroglial envelopeformed of about 35-40 oligodendrocytes, the pro-portion of astrocytes not exceeding 10%. Morerecently, Hamberger (10), using the same method,found that the specimen considered as the peri-neuronal neuroglia of the Deiters cells containedonly seven or eight neuroglial nuclei, but he con-firmed the proportions of cell types given by Hyd6nand Pigon (15), i.e., 90% oligodendrocytes and10% astrocytes. Hyden and his collaborators haveused these samples to analyze the content of oxida-tive enzymes and the base composition of nucleicacids in the neuroglial cells, compared with thegiant neurons, under normal and experimentalconditions, such as forced activity and learning.For example, they found that, in the normal ani-mal, cytochrome oxidase and succinoxidase ac-tivities are twice as intense in the satellite neurogliaas in an equal mass of Deiters neuron. Conversely,the neuroglia contain only about 0.1 of the RNAfound in the neuron. Under certain types of chemi-cal stimulation and forced activity, the base com-position of RNA in the neurons and neuroglialcells were reported to change in opposite direc-tions. On the basis of these and many other inge-nious investigations, these authors have proposed ageneral theory of a symbiotic relationship betweenneurons and satellite neuroglial cells in which the

two cell types are linked together in the productionof energy, proteins, and bioelectric potentials.Although this theory is an elaboration of specula-tions expressed earlier by Holmgren (13) andCajal (32), it has received a great deal of attentionfrom modern neurobiologists because of the elegantmicrodissections and microchemical analyses thathave been used to support it.

The electron microscopic observations reportedin this paper have an important bearing on theapplicability of cytochemical data to the supportof this interesting speculation. Generally, the giantcell of Deiters is immediately surrounded by twotypes of cellular component: (1) axon terminalssynapsing on its sufrace, and (2) delicate sheets ofastrocytic cytoplasm. Oligodendrocytes rarely ap-proach the surface of the giant cell, and when theydo come close they are always separated from itssurface by at least one thin astrocytic sheet. Thus,the true satellite of the giant cell is the astrocyteand not the oligodendrocyte. This is in contrastto the situation in the spinal cord, where both typesof neuroglial cells can be in immediate contactwith neuronal perikarya. The identification of thesatellite is perhaps not critical for the success of thesymbiotic theory, but the morphology of the satel-lite relationship is essential for interpreting thecytochemical data used to support it. The astro-cytic neuroglial processes envelop not only theperikaryon and the synaptic endings on its surface,but also a wide variety of structures includingmyelinated nerve fibers, unmyelinated preterminalfibers, and dendrites of the same and otherneurons. Although the cytochemical sample of theperikaryon can be relatively clean, the neuroglialsample can only be heterogeneous. Thus, the re-sults of these cytochemical analyses are difficult tu

interpret, because the localization of enzynmatlcactivities and other functions in such samplesremains ambiguous.

The investigations on which this report is based weresupported, in part, by Public Health Service Re-search Grant No. B-3659 from the National Instituteof Neurological Diseases and Blindness, Bethesda,Maryland. Part of this work was carried out whileDr. Sotelo was the recipient of a Bourse de Recherchefrom the North Atlantic Treaty Organization. Dr.Sotelo's present position is Attache de Recherche(C.N.R.S.) at the Laboratoire de Biologie Animale,Faculty des Sciences de Paris.Received for publication 25 July 1967.


REFERENCES

1. BLACKSTAD, T. W. 1965. Mapping of experi-mental axon degeneration by electron micros-copy of Golgi preparations. Z. Zellforsch. mikr.Anat. 67:819.

2. BOYCOTT, B. B., E. G. GRAY, and R. W. GUIL-LERY. 1961. Synaptic structure and its altera-tion with environmental temperature: a lightand electron microscope study of the centralnervous system of lizards. Proc. Roy. Soc.(London), Ser. B. 154:151.

3. BRODAL, A. 1964. Anatomical organization andfiber connections of the vestibular nuclei. InNeurological Aspects of Auditory and Vestib-ular Disorders. W. S. Fields and B. R. Alford,editors. Charles C Thomas, Springfield, Illinois.107.

4. CHANDLER, R. L., and R. WILLIS. 1966. Anintranuclear fibrillar lattice in neurons. J.Cell Sci. 1:283.

5. DAHL, H. A. 1963. Fine structure of cilia in ratcerebral cortex. Z. Zellforsch. mikr. Anat. 60:369.

6. DAVISON, P. F., and E. W. TAYLOR. 1960.Physical-chemical studies of proteins of squidnerve axoplasm, with special reference to theaxon fibrous protein. J. Gen. Physiol. 43:801.

7. EKHOLM, R., and H. HYDCN. 1965. Polysomesfrom microdissected fresh neurons. J. Ultra-struct. Res. 13:269.

8. GRAY, E. G., and R. W. GUILLERY. 1961. Thebasis for silver staining of synapses of themammalian spinal cord: a light and electronmicroscope study. J. Physiol. (London). 157:581.

9. GRAY, E. G., and R. W. GUILLERY. 1963. Onnuclear structure in the ventral nerve cord ofthe leech Hirudo medicinalis. Z. Zell/orsch. mikr.Anat. 59:738.

10. HAMBERGER, A. 1963. Difference between isolatedneuronal and vascular glia with respect torespiratory activity. Acta Physiol. Scand. Suppl.203. 58:5.

11. HASHIMOTO, P. H., and S. L. PALAY. 1965.Peculiar axons with enlarged endings in thenucleus gracilis. Anat. Record. 151:454.

12. HERNDON, R. M. 1963. The fine structure of thePurkinje cell. J. Cell Biol. 18:167.

13. HOLMGREN, E. 1900. Weitere Mitteilungen fiberdie "Saftkan/lchen" der Nervenzellen. Anat.Anz. 18:290.

14. HYDEN, H. 1959. Biochemical changes in glialcells and nerve cells at varying activity. InProc. Intern. Congr. Biochem., 4th Vienna, 1958,Symposium III.

15. HYDEkN, H., and A. PIGON. 1960. A cytophysio-logical study of the functional relationship

between oligodendroglial cells and nervecells of Deiters' nucleus. J. Neurochem. 6:57.

16. KARNOVSKY, M. J. 1965. A formaldehyde-glutaraldehyde fixative of high osmolality foruse in electron microscopy. J. Cell Biol. 27:137A. (Abstr.)

17. MASUROVSKY, E. B., M. B. BUNGE, and R. P.BUNGE. 1967. Cytological studies of organo-typic cultures of rat dorsal root ganglia follow-ing X-irradiation in vitro. I. Changes inneurons and satellite cells. J. Cell Biol. 32:467.

18. McEwEN, L. M. 1956. The effect on the isolatedrabbit heart of vagal stimulation and itsmodification by cocaine, hexamethonium andouabain. J. Physiol. (London). 131:678.

19. METUZALS, J. 1963. Ultrastructure of myelinatednerve fibers in the central nervous system ofthe frog. J. Ultrastruct. Res. 8:30.

20. MORALES, R., and D. DUNCAN. 1966. Multi-laminated bodies and other unusual configura-tions of endoplasmic reticulum in the cerebel-lum of the cat. An electron microscopic study.J. Ultrastruct. Res. 15:480.

21. MORALES, R., D. DUNCAN, and R. REHMET. 1964.A distinctive laminated cytoplasmic body inthe lateral geniculate body neurons of thecat. J. Ultrastruct. Res. 10:116.

22. MUGNAINI, E., and F. WALBERG. 1964. Ultra-structure of neuroglia. Ergebn. Anat. Entwickl.Ges. 37:194.

23. NAPOLITANO, L., and D. FAWCETT. 1958. Thefine structure of brown adipose tissue in thenewborn mouse and rat. J. Biophys. Biochem.Cytol. 4:685.

24. PALAY, S. L. 1958. The morphology of synapsesin the central nervous system. Exptl. Cell Res.Suppl. 5: 275.

25. PALAY, S. L. 1964. The structural basis for neuralaction. In Brain Function: Vol. II: RNA andBrain Function; Memory and Learning. M.A. B. Brazier, editor. University of CaliforniaPress, Los Angeles. 69.

26. PALAY, S. L. 1966. The role of neuroglia in theorganization of the central nervous system.In Nerve as a Tissue. K. Rodahl, editor.Harper and Row, New York. 3.

27. PALAY, S. L., S. M. McGEE-RuSSELL, S. GORDON,and M. A. GRILLO. 1962. Fixation of neuraltissues for electron microscopy by perfusionwith solutions of osmium tetroxide. J. CellBiol. 12:385.

28. PALAY, S. L., C. SOTELO, A. PETERS, and P.ORKAND. 1968. The axon hillock and theinitial segment. In preparation.

29. RAM6N-MOLINER, E., and W. J. H. NAUTA. 1966.


The isodendritic core of the brain stem. J.

Comp. Neurol. 126:311.30. RAM6N Y CAJAL, S. 1909. Histologie du Systbme

Nerveux de 'Homme et des Vertebr6s.

Maloine, Paris. 1.31. RAM6N Y CAJAL, S. 1910. Las f6rmulas del

proceder del nitrato de plata reducido y susefectos sobre los factores integrantes de lasneuronas. Trab. Lab. Invest. Biol. 8:1.

32. RAM6N Y CAJAL, S. 1959. Degeneration and

Regeneration of the Nervous System. R. M.May, translator and editor. Hafner PublishingCo., New York. 2:459. Reprint.

33. RICH, A., S. PENMAN, Y. BECKER, J. DARNELL,

and C. HALL. 1963. Polyribosomes: Size innormal and polio-infected HeLa cells. Science.142:1658.

34. Rio-HoRrEGA, P. del. 1928. Tercera aportaci6n alconocimiento morfol6gico e interpretaci6nfuncional de la oligodendroglia. Mems. R. Soc.esp. Hist. nat. 14:5.

35. ROSENBLUTH, J. 1962. Subsurface cisterns andtheir relationship to the neuronal plasmamembrane. J. Cell Biol. 13:405.

36. SCHMITT, F. O., and P. F. DAVISON. 1961.

Biologie moleculaire des neurofilaments. InActualites Neurophysiologiques, 3rd series.A. M. Monnier, editor. Masson et Cie, Paris.355.

37. SIEGESMUND, K. A., C. R. DUTTA, and C. A. Fox.

1964. The ultrastructure of the intranuclearrodlet in certain nerve cells. J. Anat. (London).98:93.

38. STELL, W. K. 1965. Correlation of retinal cyto-architecture and ultrastructure in Golgi prepa-rations. Anat. Record. 153:389.

39. TAKAHASHI, K., and K. HAMA. 1965. Some obser-

vations on the fine structure of nerve cell bodiesand their satellite cells in the ciliary ganglion ofthe chick. Z. Zellforsch. mikr. Anat. 67:835.

40. VENABLE,J. H., and R. COGGESHALL. 1965. A sim-

plified lead citrate stain for use in electronmicroscopy. J. Cell Biol. 25:407.


the fine structure of the lateral vestibular nucleus … · cytology of the nucleus, the fine...

Documents