Brain Research, 105 (1976) 229-251

229

© Elsevier ScientificPublishing Company, Amsterdam - Printed in The Netherlands

ULTRASTRUCTURE MOUSE SI CORTEX

AND

SYNAPTIC

CONTACTS

IN

BARRELS

OF

EDWARD L. WHITE* lnstitut d'Anatomie Normale, Rue du Bugnon 9, Lausanne 1011 (Switzerland) (Accepted August 21st, 1975)

SUMMARY

Barrels in the posteromedial barrel subfield (PMBSF) in layer IV of mouse somatosensory cortex are consistently identifiable and somatotopically related to the large whiskers on the animal's snout32, 3s. This is a study of the ultrastructure - in particular the synaptic connections - - of mouse PMBSF barrels. A technique was developed which enabled the precise selection for electron microscopy of specific regions of barrels identified in the light microscope. Barrel 'sides', containing a high density of neuronal cell bodies and myelinated axons, contrasted with barrel 'hollows' in which these elements were relatively sparse. The types and frequencies of synapses were examined in series of thin sections through sides and/or hollows of barrels B2, C3, C7, D4, E4, E5 and Es from 6 mice. In all, 3042 synapses from 20 fields, each at least 10 #m × 10 #m × 4.8/~m in size, were classified. Four distinct kinds of presynaptic terminal were identified: (1) darkly, and (2) lightly stained, asymmetrically synapsing terminals (AD, AL respectively), and (3) darkly, and (4) lightly stained, symmetrically synapsing terminals (SD, SL respectively). Synapses of these terminals were distributed in roughly similar proportions throughout the neuropil of barrel sides and hollows. Thus in all barrel regions approximately 92 ~ of the synapses in neuropil were made by AD terminals, 1.0 ~ by AL, 2.4 ~ by SD and 4.0 ~ by SL terminals. Clusters of vertically oriented dendrites, many of which are probably apical dendrites of layer V pyramidal neurons, were preferentially located in sides and in areas ('septa') between barrels. The synaptic input to 54 segments (each ~ 5 ~m long) of 42 clustered dendrites in the sides and/or hollows of barrels Bz, C3, C7, D4, E5 and Es in 6 mice was analyzed. Almost all synapses onto clustered dendrite spines were from AD terminals; however, AD terminals formed only about two-thirds of the * Present address: Department of Anatomy, Boston University School of Medicine,Boston, Mass. 02118, U.S.A.

230 synapses onto clustered dendrite shafts. The remainder of the shaft synapses were from SD and SL terminals.

INTRODUCTION

Layer IV of mouse somatosensory cortex contains a field of discrete, multicellular units which, because of their three-dimensional form, were named 'barrels'. A detailed description of the cytoarchitectural organization of the mouse barrel field has been provided by Woolsey and Van der Loos 3s. These authors designated as the posteromedial barrel subfield (PMBSF) that region of the barrel field which contains 5 distinct, roughly parallel rows of large barrels. As Woolsey and Van der Loos pointed out, there is a remarkable constancy from mouse to mouse in number and arrangement of both the PMBSF barrels and the large mystacial vibrissae (whiskers) on the animal's snout. Welker 34 and Welker and Woolsey 35 studied barrels in rat somatosensory cortex and found a similar relationship between the large mystacial vibrissae and the PMBSF barrels in that species. Experiments 32,36 in which the removal of large vibrissae in mouse neonates resulted in the absence of the corresponding PMBSF barrels in adult cortex indicate that PMBSF barrels are indeed the somatotopically arranged cortical representation of the large mystacial vibrissae. The finding by Killackey and Leshin is, that specific thalamocortical input to rat barrels is in the form of discrete clusters which appear to fill the barrels but not regions between them, lends further support to the contention of Woolsey and Van der Loos 3s that PMBSF barrels are the morphological manifestation in layer IV of physiological columns such as described by Mountcastle ~4. This is a study of the ultrastructure and synaptic connections in normal barrels of the mouse PMBSF. The principal objective of this investigation was to provide a firm basis for future studies of the ultrastructural and functional organization of PMBSF barrels whose special properties (e.g., ease of localization in histological preparations of normal brains and their relation to peripheral receptors to which one has easy experimental access) well suit them to multidisciplinary approaches to the study of cortical function. MATERIALS AND METHODS

Male white mice of strain CD/1 *, 2-3 months of age, were perfused through the heart with aldehyde solutions 29. Whole brains were removed from the skull the next day and divided into halves by cutting along the mid-sagittal plane. A slice of about 2 mm thick was removed from the inferior surface of each half-brain by cutting horizontally at right angles to the mid-sagittal plane (see Fig. 7 of Woolsey and Van der Loos 3s for reference points used to determine the horizontal plane). Half-brains thus

* Charles River Breeding Labs., St. Aubin-les-Elbeuf, France.

231 ' s h a p e d ' were p l a c e d in buffered 2.0~/o OsO4 t o r 11-12 h p r i o r to d e h y d r a t i o n in g r a d e d m e t h a n o l s for the following t i m e s : 50 ~o - - 30 m i n ; 70 ~ , 80 ~o, a n d 95 ~o - 1 h each; a b s o l u t e m e t h a n o l - - 2 changes o f 1 h each. Tissue was s t a i n e d with 1.0~o u r a n y l acetate dissolved in the 70 ~ m e t h a n o l . H a l f - b r a i n s were r e m o v e d f r o m the a b s o l u t e m e t h a n o l a n d s u b m e r g e d in a well c o n t a i n i n g liquid paraffin (paraffin, melting p o i n t 42 °C, M e r c k , D a r m s t a d t , G . F . R . ) . T h e b o t t o m o f the well was the r o u g h e n e d t o p surface o f a w o o d e n cube. H a l f - b r a i n s were p l a c e d in the well so t h a t the inferior

A

/! II

(9

iI \

Fig. 1. Drawings showing the angles of inclination used to orient half-brains so that the PMBSF is approached during thick (40/~m) sectioning in a plane tangential to the pial surface. The drawing at the upper left shows a right half-brain as it rests, inferior surface down, on the top surface of a wooden cube. The medial surface (hatching) of the half-brain is parallel to plane A of the cube; a black dot labels the olfactory bulb. Not shown is the 'paraffin well' (see text) in which the half-brain sits. The cube is initially oriented so that the top surface is level. The drawings at the right and left bottom indicate how the cube is inclined prior to sectioning: 10° in plane A, and 30° in plane B. A knife which then passes through the half-brain in a plane parallel to the top surface of the cube before inclination will cut through the PMBSF in a plane which is tangential to the pial surface in the region of the barrel field.

232 surface was flush against the top surface of the wooden cube, whereas the mid-sagittal surface was parallel to one side of the cube (Fig. 1). Following hardening of the paraffin, the cube was fixed and oriented in a Reichert OmE sliding microtome whose specimen holder was modified so that the cutting plane of the knife could pass tangentially with respect to the pial surface in the region of the barrel field. Series of sections cut at 40 /~m were collected in absolute methanol and examined with the light microscope*. Drawings were made of PMBSF barrels and blood vessels (as reference points) with the aid of a microscope-mounted drawing tube. The PMBSF was reconstructed from the drawings, and individual barrels were identified using the terminology of Van der Loos and Woolsey3L Only those barrels which were clearly seen in 3 consecutive 40/zm thick sections were selected for electron microscopy. Thin sections were taken solely from the second of the three to insure that the selected barrel was present throughout the thickness of the 40 # m thick section chosen for ultramicrotomy. Thus, the uppermost and lowermost levels of layer IV were not examined with the electron microscope. Following light microscopic examination, chosen sections were embedded in Araldite (Ciba Co., Duxford, England). Blocks containing these sections were oriented in the ultramicrotome so that subsequent sectioning continued to be in a plane parallel to the pial surface. A 1 #m thick plastic section was taken from each block and stained with toluidine blue 30. The optical superposition, with the aid of the drawing tube, of a barrel/blood vessel drawing, first onto the 1 #m thick stained section in which barrel components (e.g. cell bodies, see Fig. 4) were clearly visible, and then onto the block of tissue (i.e. the 40 #m thick section) to be thin-sectioned, permitted the selection and precise trimming for ultramicrotomy of identified regions of selected barrels. The superposition of the drawing onto the block was required since barrels were no longer discernible in the 40 btm thick section once it was embedded in plastic. Blood vessels served as useful guides which enabled the accurate matching of the drawing and block so that the selected barrel could be !ocalized. During trimming, the plastic blocks were illuminated with transmitted light so that blood vessels and other reference points could be clearly seen. Unbroken series of silver/gold colored thin sections were cut on a Porter-Blum MT-2B ultramicrotome and picked up on formvar-carbon coated slotted grids (slots were 0.8 mm × 2 ram). Thin sections were then stained with lead citrate 33 and examined with a Philips 301 electron microscope. During electron microscopy, repeated reference was made to blood vessels and to clusters of large diameter dendrites which were also seen in thick (40 ¢tm and/or 1/~m) sections. In this way, specific regions of the selected barrel could be identified in thin sections. The types and frequencies of synapses were analyzed in series of thin sections averaging 60 sections in length. Fields were selected without regard to synaptic content, but rather in relation to a structure such as a blood vessel or large diameter dendrite seen in a 1 #m thick plastic section. Each field analyzed was at least 10/~m × I0/zm × 4.8/zm in size. In all, 20 fields from 7 barrels (Bz, C3, C7, D4, E4, E5 and * A similar technique for cutting blocks of non-embedded osmicated tissue was developed independently by Holl~inder9.

233 Es) o f 6 mice were a n a l y z e d for their s y n a p t i c c o n t e n t (Fig. 2). Synapses a n d o t h e r u l t r a s t r u c t u r a l details were e x a m i n e d in v a r i o u s regions o f the 7 barrels m e n t i o n e d a b o v e a n d in single a n d serial thin sections t h r o u g h b a r r e l s Ce a n d E7 of two a d d i t i o n a l mice; only d a t a f r o m the '20 fields' a r e i n c l u d e d in the q u a n t i t a t i v e analysis o f barrel synapses. RESULTS T h e P M B S F occupies the full thickness ot layer IV in m o u s e s o m a t o s e n s o r y cortex for a n a r e a a p p r o x i m a t e l y 1 m m square as. I n Nissl s t a i n e d sections cut tangentially with respect to the pial surface, each P M B S F b a r r e l consists o f a circular o r ellipsoidal ring o f densely p a c k e d cell bodies, the b a r r e l side, which encloses a region o f lesser cell b o d y density, the b a r r e l hollow. R e g i o n s with very few cell bodies, the septa, s e p a r a t e a d j a c e n t barrels. T w o sides with an intervening s e p t u m c o m p r i s e a barrel wall. Barrels are thus defined by an i r r e g u l a r d i s t r i b u t i o n o f cell bodies. T h e

m

a I p

@I

.

I

'

mm

Fig. 2. Camera lucida drawing made from a series of tangential sections through the mouse PMBSF (modified from Fig. la of Van der Loos and Woolsey~2) showing, in heavy outline, the 5 rows of PMBSF barrels. PMBSF barrel rows are labelled A through E posteroanteriorly, barrels within a row are numbered 1, 2, 3, etc. mediolaterally; barrels named a, fl, ~ and t~straddle the rows. Asterisks indicate locations of fields of neuropil in which the types and frequencies of synapses were analyzed in series of thin sections. Each small asterisk in barrels B~, Ca, C7, D4, E4, E5 and Es corresponds to one such field; each of the 3 large asterisks in barrel D4 corresponds to two contiguous fields, one located directly above, i.e., more superficially with respect to the pial surface than the other. Synapses and other fine structural details were examined in various regions of these 7 barrels and in barrels C~ and E7 which are indicated by black dots. The PMBSF from a left hemisphere is shown; for convenience, data from right hemispheres have been transposed and combined with data from left hemispheres.

"

234

Fig. 3. Light micrograph of a tangential section through layer IV of mouse somatosensory cortex showing portions of 3 of the 5 rows (C, D, E) of barrels in the PMBSF. Section is stained only with OsO4 and is unembedded. In osmicated preparations, barrels are seen as rings of darkly stained myelinated axons; the number, size and arrangement of these rings is consistent with the appearance and distribution of the rings of densely packed cell bodies which define barrel walls in Nissl stained preparations. Barrel E6 indicated by the open arrowhead is shown at higher magnification in Fig. 4. Note the relationship of this barrel to the large blood vessels indicated by closed arrowheads. Section thickness, 40 #m. Bar represents 200/~m; × 130. r e a d e r is referred to previous reports for a m o r e c o m p l e t e d e s c r i p t i o n o f the organiz a t i o n o f P M B S F barrels at the light m i c r o s c o p i c level 2~,~2,3s,zs. Cell bodies showed u p p o o r l y in the o s m i u m stained thick sections e m p l o y e d in this s t u d y ; however, large n u m b e r s o f d a r k l y stained m y e l i n a t e d axons p a c k e d in septa a n d in between cell bodies in b a r r e l sides facilitated the visualization o f barrels. In o s m i c a t e d p r e p a r a t i o n s (Fig. 3), barrel walls presented themselves as d a r k , myelin-rich rings (or polygons) enclosing m o r e lightly stained regions, the b a r r e l hollows, in which m y e l i n a t e d fibers were m o r e

235

Fig. 4. Light micrograph of a 1/~m thick plastic section through the barrel indicated by the open arrowhead in Fig. 3. Blood vessels indicated by closed arrowheads in Fig. 3 are here labeled b. Section is stained with toluidine blue which permits the visualization of cell bodies. A broken line ! passes through cell bodies which form the barrel 'side'. The broken line is doubled at the right to indicate that the side in this region is ill-defined in both this and in the 40/~m thick section (see Fig. 3). Some 'side neuropil' (see text) is enclosed by a black rectangle at the upper left of the figure. The area enclosed by the side is the 'hollow'; the region between two adjacent sides is the 'septum'. Part of a septum is indicated by the opposed arrowheads (upper left). Clusters of large diameter dendrites (small arrows) are more densely packed in barrel sides and in septa than in hollows. Bar represents 50/zm; x 630. sparsely distributed. The location, number, size and arrangement o f the myelin-rich rings was fully coincident with the barrel pattern as seen in Nissl stained preparations. The high concentration o f myelinated fibers in both sides and septa made it difficult to differentiate these regions in 40/~m thick sections. The b o u n d a r y between side and septum was better defined in toluidine blue stained 1 # m thick sections in which s o m a t a were m o r e clearly seen (Fig. 4). As pointed out by F e l d m a n and Peters ~ in their study o f rat barrels, 'strings o f three or m o r e apparently contiguous neurons' are characteristic o f barrel sides in 1/~m thick sections. Nevertheless, gaps in the line o f cell bodies forming the barrel side made it difficult to determine the location o f the side in some regions. I n addition, it could n o t always be determined if cell bodies and surr o u n d i n g neuropil at the very periphery o f barrel hollows would be judged, in thicker preparations, as belonging to the barrel side. Similar problems were encountered when attempts were made to precisely define the interfaces between hollows, sides and septa at the ultrastructural level. A l t h o u g h

236 sides were characterized by strings of closely spaced - - sometimes contiguous - - cell bodies, the existence of somata-free gaps complicated the identification of barrel regions and septa in the electron microscope. Neuropil in the gaps was usually filled with large numbers of myelinated and unmyelinated fibers cut in cross-section. These regions blended imperceptibly with adjacent septal regions which also contained a high concentration of myelinated and unmyelinated fibers. Similarly, no distinct line of demarcation separated neuropil in the gaps from hollow neuropil (Fig. 5). It was only by repeated reference to blood vessels, cell bodies, and clusters of large diameter dendrites visible in both thick (40 and/or 1 # m thick) and thin sections, that many areas in thin sections could be related to specific regions of identified barrels. In this report, 'side' neuropil refers to neuropil situated in gaps between lines of cell bodies clearly belonging to a barrel side,(e,g, see Fig. 5). Neuropil on the hollow side of cell bodies in barrel sides was judged as belonging to barrel hollows. Nearly all cell bodies were identified as neuronal in that they were postsynaptic (Figs. 6-9). A similar conclusion was reached by Feldman and Peters 5 in their light microscopic study of cell bodies in rat barrels. Most neuronal cell bodies were about 9/~m in diameter, spherical and contained only a thin rim of cytoplasm in which there was little Nissl substance (Fig. 6). Larger (up to 14 # m in diameter), more irregularly shaped neuronal somata contained larger amounts of Nissl substance and other

Fig. 5. Montage of electron micrographs of a single section showing neuropil in a 'gap' in the line of cell bodies (CB) forming the barrel side. Septum is to the left, hollow to the right. Axons cut in crosssection, both myelinated (A) and unmyelinated (large arrows), and clustered dendrites (De, small arrows), are more frequent in side and in septal than in hollow neuropil. Bar represents l/tin; × 12,000.

Fig. 6. Montage of electron micrographs of a single section showing, on the left, an example of the type of small neuronal soma most frequently found in barrel sides and hollows; on the right, an example of a larger diameter, less frequently encountered type of neuronal cell body. Rectangles numbered 7 and 8 enclose synaptic terminals shown in Figs. 7 and 8 respectively. The larger soma contains a greater amount of Nissl substance (ns) and other cytoplasmic organelles than does the smaller diameter soma. CB, cell body. Bar represents 3/~m; x 11,000. Figs. 7, 8 and 9. Each figure is an electron micrograph montage of serial thin sections. Dotted lines indicate boundary regions between micrographs of adjacent thin sections. Bars represent 0.5/~m. Fig. 7. Terminal (T), indicated by rectangle 7 of Fig. 6, synapses (arrows) onto a small dendrite (left) and onto a small cell body (right) ; both of which are shown in Fig. 6. x 37,000. Fig. 8. Terminal (T), enclosed by rectangle 8 of Fig. 6, synapses (arrows) onto both cell bodies shown in Fig. 6. x 35,000. Fig. 9. Two terminals (T) joined by a narrow axonal tube (b). Each terminal synapses (arrows) onto a cell body (CB). Portions of 3 consecutive thin sections are shown here: the synapse (a) of the bouton on the left is from one section, the connecting tube (b) is from the second, and the rest of the montage is from the third, x 25,000.

238 cytoplasmic organelles; invaginations of the nuclear membrane were characteristic of these larger somata. Reconstructions from series of thick and thin sections showed that dendrites of cell bodies in barrel sides were preferentially oriented towards the barrel hollow. This finding is consistent with the results of previous studies of mouse s, '2.1,37 and rat is barrels. There was no discernible difference between cell types in barrel hollows and those in barrel sides. Similarly, Pasternak and Woolsey ~5 report no significant difference in the mean sizes of cell bodies in the sides and hollows of mouse barrel C1. A detailed classification of cell types, and an in-depth analysis of their synaptic input, will not be considered in this report. Except for myelin sheaths, glial elements were uncommon in sides and hollows, and it is interesting to note that barrels are not circumscribed by astroglial processes (Elias Perentes, personal communication) as are synaptic glomeruli in other regions of the mouse brain 20. Synaptic contacts were identified on the basis of the usually accepted criteria for mammalian chemical synapses 26: each consisted of a cluster of vesicles and patches of electron-dense material at the presynaptic membrane, a synaptic cleft, and a postsynaptic membrane. Synapses were classified asymmetrical or symmetrical based on the amount of electron-dense material adjacent to the cytoplasmic surface of the postsynaptic membrane 4. Synaptic junctions which possessed a plate of postsynaptic dense material 40-50 nm thick were classified asymmetrical (e.g., Fig. 10); junctions which possessed a thin plate of postsynaptic dense material (less than 10 nm) were classified symmetrical (e.g., Fig. 14). A few synapses possessed intermediate amounts of postsynaptic dense material and these were not classified as either asymmetrical or symmetrical. Structures were identified as synapses only when the synaptic cleft was clearly visible in at least one thin section of a series, because we felt that we could not distinguish the chance collection of synaptic vesicles at obliquely sectioned membranes from an obliquely sectioned synaptic junction. In general, terminals which synapsed at symmetrical synaptic junctions possessed a higher proportion of flattened synaptic vesicles and somewhat narrower synaptic clefts than did neighboring, asymmetrically synapsing terminals. Differences in vesicle and cleft morphology were neither striking nor always consistent from one section to another in a series, and thus size and shape of synaptic vesicles and clefts were not used as criteria to differentiate synaptic types. The types and frequencies of barrel synapses were assessed in unbroken series of thin sections. Each field was at least 10/zm × 10 #m × 4.8/~m in size and contained approximately 60 sections. A total of 13 fields in the hollows of barrels B2, C3, D4, E4, E5 and E8, and 7 fields in the sides of barrels C7, D4, E5 and E8 were completely analyzed (see Fig. 2). Side fields always contained, or were in close proximity to, clusters of dend rites whose long axes were oriented perpendicularly with respect to the pial surface. Hollow fields were of two types: those containing clustered dendrites (hollow (-~-cd) fields), and those located at least 10/~m from those dendrites (hollow (--cd) fields). In general, proximal dendrites of neurons whose somata were situated in layer IV were more numerous in hollow (--cd) fields. The term, 'synaptic terminal' is applied to boutons en passant as well as to boutons terminaux; 'barrel neuropil' refers to neuropil in both sides and hollows.

239

Fig. 10. Electron micrograph showing two examples of the most frequent axon terminal (AD, asymmetrical dark, see text) in all regions of barrel neuropil. One terminal synapses (arrows) onto two spines (s) - - one of which belongs to a large dendrite (D) - - while the other synapses (arrow) directly onto the dendrite shaft. The shaft synapse is shown in inset because the section in which it appeared best was located 20 thin sections away from the section containing the spine synapses. Bar for figure and inset represents 0.5 # m ; × 28,000. Figs. 11 and 12. Electron micrographs in which one may compare in a single section, the cytological features of A D and AL (asymmetrical light, see text) terminals. Synapses are indicated by arrows. Bars represent 0.5/~m; × 35,000.

240 Four distinct types of synaptic terminal were identified in barrel neuropil. Synaptic junction type (symmetrical or asymmetrical) and relative electron density of the presynaptic terminal (more electron-dense than the postsynaptic element -'dark', less or equally dense -- 'light') were used as criteria to differentiate types of synaptic terminal. Although all postsynaptic elements were not equally electron-dense, the variation in density was little and much less than the difference in density of 'dark' as compared to 'light' presynaptic elements.

Asymmetrical dark (AD) More than 90 ~ of the synaptic terminals in any region of barrel neuropil were presynaptic asymmetrically to spines and dendrites. Nearly all of these terminals were more darkly stained than their postsynaptic processes (Figs. 10-13); an appearance due primarily to a densely stained substance in the axoplasm, and partly to a high concentration of evenly dispersed synaptic vesicles. Occasionally the mitochondria of AD terminals were lightly stained and appeared swollen (Fig. 15). Many AD terminals were traced in serial sections to darkly stained unmyelinated axons about 0.3/~m in diameter. These fibers were scattered in bundles throughout the neuropil, but they were more highly concentrated in sides and in septa than in hollows.

Asymmetrical light (AL) The cytoplasm of a few asymmetrically synapsing terminals was of equal or lesser electron density than that of their postsynaptic elements (Figs. 11, 12). One of these terminals was traced to a myelinated axon about 1 #m in diameter. This fiber and the unmyelinated fibers, to which A L terminals were more often traced, were more lightly stained than fibers which gave rise to darkly stained terminals.

Symmetrieal dark ( SD) Some darkly stained terminals were presynaptic at symmetrical synaptic junctions; however, in other respects, such as electron density of axoplasm and packing of synaptic vesicles, these SD terminals were identical to the darkly stained, asymmetrically synapsing (AD) terminals described above (e.g., compare Figs. 13 and 14). Indeed, both types of terminal were traced to darkly stained unmyelinated fibers, but we emphasize that in no instance were different types of terminal traced to a single fiber.

Symmetrical light ( SL) Lightly stained, symmetrically synapsing terminals were the second most frequent synaptic type in barrel neuropil (Figs. 15, 17). Two SL boutons were often connected by a narrow axonal tube up to 4/~m in length; sometimes two interconnected SL terminals synapsed onto one postsynaptic element (see T, Fig. 9). Most of the synapses onto small ceil bodies and proximal portions (_< 20 #m in length) of their dendrites, and about one-half of the synapses onto larger somata, were from SL terminals. For instance, reconstructions of serial thin sections through the greater part of several small somata showed that they were postsynaptic exclusively at symmetrical synaptic

Fig. 13. Electron micrograph showing an A D terminal which synapses (arrow) onto a large dendrite (D). Bar represents 0.5 btm; x 45,000. Fig. 14. Electron micrograph showing an SD (symmetrical dark, see text) terminal which synapses (arrow) onto a large dendrite (D). Bar represents 0.5 # m ; × 45,000.

Fig. 15. Electron micrograph in which one may compare in a single section, the cytological characteristics of one SL (symmetrical light, see text) and 3 A D terminals. Inset: synapse onto spine (s) as it appears in an adjacent section. Synapses indicated by arrows. D, dendrite. Bar for figure and inset represents 0.5/~m; x 32,000.

242

Figs. 16 and 17. Electron micrographs showing in Fig. 16, a longitudinallysectioned myelinatedfiber (F) which, when followed in series, gives rise 16 sections later to the SL axon terminal shown in Fig. 17. Fig. 16, bar represents 1/~m; x 26,000. Fig. 17, bar represents 0.5 ktm; x 38,000.

junctions and that most of the presynaptic terminals were lightly stained. In 12 instances, SL terminals were traced to myelinated axons about I /~m in diameter (Figs. 16, 17). The proportions of the different types of synapses varied little from barrel to barrel and in different regions of the same barrel. Quantitative data concerning the 4 synaptic types described above are contained in Table I. Data from the 3 regions examined, i.e., sides, hollow ( + c d ) and hollow (--cd) fields, are listed separately. A few synaptic terminals were difficult to classify as light or dark. Their synapses and those of a few very electron-dense endings, which resembled degenerating terminals 14, are included in the category 'other synapses'. Also included in this category are those synapses which could not be classified asymmetrical or symmetrical. The results shown in Table I indicate that, for nearly equal volumes of neuropil, there are many more synapses in barrel hollows than in sides. This may be due to the large numbers of myelinated and unmyelinated fibers which pass through barrel sides, but which do not synapse at the levels of layer IV examined.

243 TABLE I FREQUENCY OF SYNAPSES IN BARREL NEUROPIL

(+cd): fields containing clusters of dendrites whose long axes were oriented perpendicularly with respect to the pial surface. (--cd): fields whose limits were located at least 10 #m from clusters of vertically oriented dendrites. Please note: the results in this table are derived from a serial thin section analysis of neuropil in 20 fields (each at least 10/~m × 10/~m × 4.8/~m in size) from 7 identified barrels of 6 animals. The proportion of the different types of synapses varies little from one region of the barrel to another (i.e. in side, hollow (÷cd) and hollow (--cd). Barrel region

Side Hollow (÷cd) Hollow (--cd)

V o l u m e Number of % Asymmetrical (in cu.btm) * synapses* * Dark Light

% Symmetrical

5090 5091 1738

1.2

2.9

3.1

1.0 1.0

2.4 1.8

2.6 6.4

900 1473 669

92.1 93.7 90.1

Dark

Light

% Other synapses

0.7 0.3 0.6

* Volumes of cell bodies and blood vessels excluded. ** Synapses onto cell bodies excluded. Clusters of medium to large dendrites cut in cross-section were found throughout the PMBSF; however, there were more of these dendrites per unit area in barrel sides and septa than in barrel hollows (Fig. 4). Reconstructions of clustered dendrites from serial thin sections showed that their shafts were roughly cylindrical with occasional spines; shaft cytoplasm contained portions of smooth endoplasmic reticulum, small diameter (0.3/*m) mitochondria which often exceeded 4/~m in length, and large numbers of fairly regularly disposed microtubules. Mitochondria and microtubules were typically cut in cross-section as were the clustered dendrite shafts themselves (Fig. 5). Since, in this study, sections were cut exclusively in a plane tangential to the pial surface, it was concluded that the long axes of clustered dendrites, and of their mitochondria and microtubules, were oriented perpendicularly with respect to the pial surface*. The synaptic input to portions of these vertically oriented, clustered dendrites was assessed in reconstructions of serial thin sections. In this report, 'dendritic segment' refers to that portion of a clustered dendrite contained within a series of thin sections. The approximate length of dendritic segments was 5 #m, and they averaged about 2 p m in diameter. The synaptic input to 41 segments of 32 clustered dendrites passing through the hollows of barrels B2, C3, D4, E5 and Es (segments of 9 dendrites were contained in two successive series of thin sections) and to 13 segments of 10 clustered dendrites located in the sides of barrels C7, D4, E5 and E8 was analyzed. Seven dendritic segments received no synapses, 7 others were postsynaptic only onto their shafts, 8 received synapses only onto their spines, while the remaining 32 dendritic segments were postsynaptic at both spines and shafts. In general, the larger the diameter of the dendritic segment, the greater the

* For brevity, 'vertically oriented' will be used to denote structures whose long axes were oriented perpendicularly with respect to the pial surface in the region of the PMBSF.

244 TABLE II DISTRIBUTION OF SYNAPSES IN LAYER I V IN FIELDS CONTAINING VERTICALLY ORIENTED CLUSTERED DENDRITES*

Number of synapses in 15 fields** Percentages of total number of synapses (N 2182)

Other

Asymmetrical

Symmetrical

Dark

Light

Dark

Light

2036 93.3

25 1.2

55 2.5

59 2.5

7 0.3

52 118

0 6

19 2

5 0

1 0

0 24

34.5 3.6

8.5 0

Number o f synapses onto dendritic segment

Shafts Spines Proportion o f synapses onto dendritic segment* * *

Shafts Spines

2.6 5.8

14.3 0

* Data derived from serial section reconstructions of 54 dendritic segments, each ~ 5 pm long, of 42 clustered dendrites from the sides and hollows of 6 barrels in 6 animals. ** Includes all hollow (+cd) fields and all but one side field described in Table I. Minimum field size 10/~m × 10 #m × 4.8 pro. n um ber of dendritic segment shaft (spine) synapses of a particular type ** * Each number is a ratio = total number of synapses of that type in 15 fields

n u m b e r o f spines, a n d spine a n d shaft synapses it possessed. The 54 dendritic segments received a total o f 77 s y n a p t i c contacts o n t o their shafts a n d 126 synaptic c o n t a c t s o n t o 133 spines - - 7 spines were o p p o s e d to p o r t i o n s o f d a r k l y stained terminals c o n t a i n i n g clusters o f vesicles, b u t in these 7 instances, the o p p o s e d cell m e m b r a n e s were obliquely sectioned. N o synaptic cleft was discerned in the resultant m e m b r a n e s m e a r a n d the s m e a r was thus n o t considered to be a synapse. A consequence o f the rejection o f this a n d o t h e r m e m b r a n e smears as synapses is t h a t if one assumes t h a t all spines receive synapses, then in this report, a b o u t 5 ~o (7/133) o f a x o s p i n o u s synapses were n o t identified as such because their synaptic j u n c t i o n s were obliquely sectioned. One w o u l d presume t h a t a s o m e w h a t lower percentage o f a x o d e n d r i t i c and a x o s o m a t i c synapses were rejected for a similar r e a s o n because m o s t dendrites and cell bodies h a d m u c h larger d i a m e t e r s than spines a n d were thus likely to possess p r o p o r t i o n a t e l y less obliquely cut m e m b r a n e t h a n spines. A n analysis o f the types a n d frequencies o f synapses o n t o the 54 dendritic segments showed a d i s t r i b u t i o n t h a t was different f r o m the d i s t r i b u t i o n o f the t o t a l n u m b e r o f synapses in the analyzed fields. F o r example, 38.1 ~ o f the synapses m a d e by d a r k l y stained, s y m m e t r i c a l l y synapsing (SD) terminals, in fields c o n t a i n i n g clustered dendrites, synapsed o n t o these dendrites, whereas these terminals f o r m e d only 2.5 )o~ o f the total n u m b e r o f synapses in the same fields. These d a t a are c o n t a i n e d in T a b l e II which shows the p r o p o r t i o n s o f the t o t a l n u m b e r s o f each type o f synapse which were p r e s y n a p t i c to shafts a n d spines o f dendritic segments. The types and fre-

245 TABLE III FREQUENCY OF SYNAPSES IN LAYER IV ONTO SHAFTS AND SPINES OF VERTICALLY ORIENTED CLUSTERED DENDRITES*

Percentage o f shaft synapses (N = 77)** Percentage of spine synapses (N = 126)**

Asymmetrical

Symmetrical

Dark

Light

Dark

Light

67.5 93.7

0 4.8

24.7 1.6

6.5 0

Other

1.3 0

* See * of Table II. ** N u m b e r s of each type of synapse are expressed as percentages of the total n u m b e r of synapses onto dendritic segment spines or shafts.

quencies of synapses onto dendritic segments in sides and hollows was similar and thus data from these two regions are combined in Tables II and III. In Table III, the synaptic input to dendritic segments by each synaptic type is expressed as a percentage of the total number of synapses onto shafts or spines of dendritic segments. Thus, 67.5 ~ of the synapses onto clustered dendrite shafts and 93.7 ~o of the synapses onto their spines were made by darkly stained, asymmetrically synapsiog (AD) terminals. Reference to Table II shows us that these synapses represented only 8.4 ~ of all the synapses made by AD terminals in fields containing clustered dendrites. DISCUSSION

The ultrastructure, with emphasis on synaptic connections of mouse PMBSF barrels, has been examined using a technique which permits the precise selection for electron microscopy of specific regions of barrels identified in the light microscope. Barrel sides, containing densely packed cell bodies, and high concentrations of myelinated axons and large diameter dendrites, contrast with barrel hollows in which these elements are more sparsely distributed. The synapses of 4 distinct types of synaptic terminal are distributed in roughly similar proportions throughout the neuropil of sides and hollows. Approximately 9 0 ~ of the synapses in all regions of barrel neuropil are made by densely stained axon terminals which synapse at asymmetrical synaptic junctions. Axon terminals of this type are a frequent component of the neuropil in layer IV of monkey SI cortexla. An examination of this region in preparations containing lesioninduced degenerating axons indicates that some of these terminals arise from cell bodies located in the ventral posterior thalamic nucleus14,15. Light microscopic studies have shown that layer IV in mouse'l, 2~ and in rat 17 somatosensory cortex receives a large thalamic projection which is especially dense in the hollows of rat PMBSF barrels is. The cell bodies of origin of axon terminals could not be identified in this study of mouse PMBSF barrels; however, preliminary results of an examination of mouse barrels following thalamic lesions suggest that here, as in the monkey SI cortex, thalamocortical axon terminals are darkly stained and synapse at asym-

246 metrical synaptic junctions (E.L.W. with M. T. Shipley, work in progress). In general, the proportion of the different types of synapses varies little from barrel to barrel and in different regions of the same barrel. An exception to the similar distribution of synaptic types in the 3 different barrel regions (side, hollow ( + c d ) and hollow (--cd)), is seen in the frequencies of symmetrical synapses in fields located at least l0 #m from the closest dendrite cluster (i.e., hollow (--cd) fields): SL synapses are more than twice as frequent, while SD synapses are about three-fourths as frequent in hollow (--cd) fields than in those fields associated with clustered dendrites, namely, hollow (d-cd) fields and all but one side field. The high proportions of SI, synapses in hollow (--cd) fields might be due to the presence in these fields of numerous proximal dendrites of layer IV neurons; these dendrites were postsynaptic nearly exclusively to SL terminals, and each dendrite received several of this type of ending. The low proportion of SD synapses in hollow (--cd) fields might be explained by the fact that hollow (--cd) fields lack a principal postsynaptic site for SD terminals, namely, the clustered dendrites. The fairly regular distribution of synaptic types might mean that the various inputs to barrels from other regions of the brain are evenly distributed across barrel neuropil. Interesting in this regard is the finding of Sloper 31 that degenerating thalamocortical axon terminals in primate motor cortex are localized in proximity to clusters of pyramidal celt apical dendrites. Whether a similar distribution of thalamocortical axon terminals exists in mouse barrels remains to be determined. The precise distribution of identified barrel inputs could be determined by quantitative analyses of barrel synapses in preparations in which axon terminals from known regions of the brain have been somehow marked for identification. One result of such studies would be the knowledge of how many axon terminals of different regions of origin have been categorized in the present study as one synaptic type, and conversely, whether two or more different types of terminal arise from a single region. Synaptic terminals were often traced into continuity with others of the same type; but in no instance were two different types of terminal traced into continuity with each other. This suggests that the 4 synaptic types arise from at least 4 different kinds of neuron, but the possibility of a common cellular origin for two synaptic types cannot be excluded. For example, Kane 16 concluded that single axonal fibers in the cat cochlear nucleus give rise to both asymmetrically and to symmetrically synapsing terminals. We cannot exclude that the preterminal axons of asymmetrically and symmetrically synapsing terminals (or of 'light' and 'dark' terminals) are confluent in regions beyond the limits of our reconstructions. There is also the possibility that two different synaptic types possess a common cellular origin but are in different functional states at the time of fixation. For instance, electrical stimulation of peripheral nerve terminals induces mitochondrial swelling and a redistribution of cellular membranes 7, lJ. Similarly, it is possible that synaptic junction type, or the electron density - - light or darkness - - of barrel axon terminals is linked to the functional state of the terminals. Clusters of large diameter, vertically oriented, dendrites present throughout the PMBSF are more numerous per unit area in barrel sides and in septa than in hollows. It is likely that many of these dendrites are the apical dendrites of pyramidal

247 neurons whose somata are located in layer V. This conclusion is based on findings in various regions of rat 5,~7, cat 5,6, rabbit~3, monkey2,~ and human 2,5 neocortex, that apical dendrites of layer V pyramidal neurons are grouped into clusters as they ascend through the cortex before ramifying in more superficial cortical layers. Furthermore, the ultrastructure of clustered dendrites in the mouse PMBSF is fully consistent with Peters and Walsh's 27 description of clustered pyramidal cell apical dendrites in layer IV of rat somatosensory cortex (e.g., compare clustered dendrites of Fig. 5 with appearance of clustered dendrites in Fig. 2 of ref. 27). It cannot be excluded, however, that some dendrites in mouse PMBSF clusters arise from cell bodies situated in cortical layers other than layer V. For instance, cell bodies in barrel sides have dendrites whose proximal portions are vertically oriented for some distance before the dendrites enter the barrel hollow 21,37, and these proximal dendrites might be confused in tangential sections with the apical dendrites of layer V somata. Thus it could be that the high concentration of vertically oriented clustered dendrites in barrel sides is partly related to the dense packing of layer IV somata in these regions. The finding in the mouse of a preferential distribution of vertically oriented clustered dendrites in barrel sides and in septa differs from the conclusion of Feldman and Peters 5 that the distribution of dendritic clusters bears no spatial relationship to barrels in the rat. These divergent results might represent an interspecies difference in the organization of barrel components. For instance, Welker and Woolsey 35 have shown that hollows of rat PMBSF barrels contain a much higher density of cell bodies with respect to barrel sides than is found in PMBSF barrels of the mouse. Alternatively, the random distribution of clustered dendrites reported by Feldman and Peters 5 for the rat may be a consequence of their examination of the smaller, more rostrally located, non-PMBSF barrels which perhaps do not possess a spatial relationship with clustered dendrites. These authors observed that in the rat, cell bodies are 1.5-2.0 times more concentrated in barrel sides than in hollows - - a finding which supports our interpretation that non-PMBSF barrels were examined. It is known that the distribution of cell bodies in non-PMBSF barrels in the rat resembles that in PMBSF barrels of the mouse 35, and that mouse PMBSF barrels possess a side/hollow cell body ratio of 1.6 (ref. 25). By contrast, the packing density of cell bodies in rat PMBSF barrels does not differ so markedly from side to hollow 35. Thus it would seem that Feldman and Peters examined non-PMBSF barrels. An examination of the distribution of dendrite clusters in both kinds of barrel in each species is needed to enable a choice between the above alternatives. Myelinated and unmyelinated axons whose long axes in layer IV are oriented perpendicularly to the pial surface are much more densely packed in the sides and septa of mouse PMBSF barrels than in their hollows. Most of these fibers do not synapse at those levels* of layer IV examined in this study. Thus sides and septa contain a high concentration of axons which probably pass through layer IV, without synapsing, on their way to or from other layers of cortex. This might account for the

* See Materials and Methods.

248 approximately 4 0 ~ lower concentration of synaptic contacts in side, as compared with hollow, neuropil. That the neocortex contains radially oriented bundles of fibers has long been known 1, but this 'vertical' arrangement of fibers has not as yet been shown to be related to other elements contributing to cortical architecture - - in this instance, a cortical barrel. Clusters of large diameter dendrites constitute another vertically oriented neuronal element characteristic of barrel sides, although the presence of synapses onto clustered dendrites indicates that these dendrites are not simply 'conduits' passing through layer IV. Further support for the hypothesis that side neuropil is primarily concerned with the inter-layer transmission of signals is provided by the finding that myelinated fibers which are evenly distributed in the deeper levels of layer V, become grouped into bundles which pass around hollows - - and thus into barrel sides and septa - - as they approach layer IV (as observed in coronally and tangentially sectioned myelin stained preparations). However, it cannot be excluded that some proportion of these fibers synapses onto barrel components near the boundaries of layers 11I and IV, and IV and V since these regions were not examined in the present study. This is a likely possibility, since the most frequent type of synapse onto cell bodies in barrel sides is from terminals which have on occasion been traced to myelinated fibers. Furthermore, it is possible that the axons of celt bodies in layer IV become myelinated and contribute to the vertically oriented bundles of myelinated fibers in barrel sides. Thus it might be that the high concentration of myelinated fibers in side neuropil is due not only to fibers passing through layer IV, but is in part related to the high concentration of cell bodies in barrel sides. It would be interesting to examine the relationship of myelinated fibers to barrel sides in the r a t PMBSF since, here, cell bodies are nearly as densely packed in barrel hollows as they are in barrel sides 35. An even distribution of myelinated fibers and cell bodies throughout the PMBSF in this species would be consistent with a close relationship between these two neuronal elements. Suggestive evidence for such a relationship is provided by the finding that very few bundles of myelinated fibers are found in the lower levels of layer III above barrel sides; but we cannot determine from our data what proportion of the myelinated fibers in layer IV originate or terminate in this layer, or pass, perhaps unmyelinated, to other layers of cortex. Similarly, the site of termination of many of the vertically oriented unmyelinated axons in barrel sides is unknown. More detailed information on the origin and distribution of axons in barrel sides might be obtained by an examination of Golgi-Nissl preparations in which both somata and axons may be seen. Another approach would be to examine coronally sectioned material with the electron microscope, since vertically oriented fibers can be followed for greater distances in coronal than in tangential sections. Postsynaptic elements could not always be identified as belonging to pyramidal or non-pyramidal neurons ('non-pyramidal' includes the classical 'stellate' cells of Cajal 3 and Lorente de N621,22). An examination of barrels in coronal thin sections, in which the apical dendrites of layer IV and V pyramidal cells are demonstrated, should facilitate the identification of somata and processes as pyramidal or nonpyramidal. Nevertheless, the results of this study point to a possible inadequacy in the ultrastructural criteria currently employed to differentiate even the somata of

249 non-pyramidal from pyramidal neurons. Observations made on Golgi21,22 and Nissl a stained preparations indicate that small diameter stellate cells constitute by far the largest proportion of the cell bodies in layer IV of rodent parietal cortex. Nearly all cell bodies seen in thin sections through barrels are of small diameter and are postsynaptic only to symmetrically synapsing terminals. Based on their abundance, it seems reasonable to conclude that these small diameter somata correspond to the small stellate cells seen with the light microscope. This conclusion is, however, inconsistent with several electron microscopic studies of cerebral cortex in which somata postsynaptic only at symmetrical synaptic junctions have been identified as pyramidal cells (e.g. ref. 12). Another finding consistent with our conclusion that perikarya postsynaptic exclusively to symmetrically synapsing terminals do not have to be pyramidal is found in LeVay's19 electron microscopic study of Golgi impregnated neurons in visual cortex. LeVay concluded that 'spiny stellate' as well as pyramidal somata were postsynaptic only at symmetrical synaptic junctions. Thus it should not be assumed, in the absence of supporting data (e.g., presence of an apical dendrite), that a cell body which is postsynaptic only at symmetrical synaptic junctions is pyramidal. Evidence that mouse PMBSF barrels are somatotopically related to the large mystacial vibrissae on the animal's snout has been provided by studies which show that removal of vibrissae in neonatal mice results in the absence of the corresponding barrels in adult cortexa2,a6. In rats also, the cortical representation of mystacial vibrissae is in the barrel region and is somatotopically organizeda4. Killackey and Leshinis have demonstrated that the specific thalamocortical projection to rat PMBSF barrels is in the form of discrete clusters which appear to fill the barrel hollows. This finding is reminiscent of Lorente de N6's 21 description of the branching pattern of specific thalamocortical axons in mouse parietal cortex as 'glomeruli', and further supports the concept, initially proposed by Woolsey and Van der Loos38, that each PMBSF barrel is the morphological manifestation of a physiological column of the type present in, for example, monkey2S,39 and cat 1°. A study of tangential sections through several mammalian cortices, including that of cat and monkey5, demonstrated only isolated, scattered rings of neurons in layers IV and II. Thus, consistently identifiable units such as the barrels which constitute the PMBSF of mice and rats are probably not a general feature of mammalian brains. The presence of barrels in the cortices of these two 'whisking' rodents might, as Woolsey and Van der Loos suggested, be related to the punctate nature of the sensory surface formed by the vibrissae; the prominence of barrels might be related to the great importance of vibrissae in the behavior of mice and rats 38. Although perhaps limited to a few species, the likely equivalence of barrels with physiological columns suggests that studies of the structure and function of barrels will elucidate principles of cortical organization which could be generalized to other cortical systems.

250 ACKNOWLEDGEMENTS The author thanks Drs. Hendrik Van der Loos, Konrad Kaufmann, and Frank L. R i c e f o r useful, i n f o r m a t i v e discussions,

Mesdames

Marie-Christine

C r u z et

R o s a l y n M. W h i t e q u i p a r leurs expertises m ' o n t b e a u c o u p aid6, ainsi q u e M m e . Henriette Hammond

p o u r la d a c t y l o g r a p h i e d u m a n u s c r i t et p o u r sa gentillesse.

Je r e m e r c i e i n f i n i m e n t le F o n d s N a t i o n a l Suisse de la R e c h e r c h e S c i e n t i f i q u e p o u r le subside N o . 3.1350.73 S R a t t r i b u 6 fi H. V a n d e r L o o s .

REFERENCES 1 BERLIN,R., Beitrag zur Structurlehre der Grosshirnwindungen, lnauguralabhandlung, A. E. Junge' schen Universitatsbuchdrukerei, Erlangen, 1858. 2 BONIN, G. VON, AND MEHLER, W. R., On columnar arrangement of nerve cells in cerebral cortex, Brain Research, 27 (1971) 1-10. 3 CAJAL, S. RAM6N Y, Histologie du Systdme Nerveux, Vol. 11, Maloine, Paris, 1908-1911. (Transl. by AZOULAY,L., Consejo Superior de Investigaciones Cientfficas, Madrid, 1955.) 4 COLONNIER,M., Synaptic patterns on different cell types in the different laminae of the cat visual cortex. An electron microscopic study, Brain Research, 9 (1968) 268-287. 5 FELDMAN, M. L., AND PETERS, m., A study of barrels and pyramidal dendritic clusters in the cerebral cortex, Brain Research, 77 (1974) 55-76. 6 FLEISCHHAUER,K., PETSCHE, H., AND WITTKOWSKI, W., Vertical bundles of dendrites in the neocortex, Z. Anat. Entwickk-Gesch., 136 (1972) 213-223. 7 HEUSER, J. E., AND REESE, T. S. R., Evidence for recycling of synaptic vesicle membrane during transmitter release at the frog neuromuscular junction, J. Cell Biol., 57 (1973) 315-344. 8 HINRICHSEN,C. F. L., AND STEVENS, G. E., Preliminary observations on the structure of cortical barrels in the mouse, J. Anat. (Lond.), 118 (1974) 391. 9 HOLL,~NDER,H., The section embedding (SE) technique. A new method for the combined light microscopic and electron microscopic examination of central nervous tissue, Brain Research, 20 (1970) 39~47. 10 HUBEL, n . H., AND WIESEL, T. N., Shape and arrangement of columns in cat's striate cortex, J. Physiol. (Lond.), 165 (1963) 559-568. 11 JONES, S. F., AND KWANBUMBUMPEN,S., The effects of nerve stimulation on synaptic vesicles at the mammalian neuromuscular junction, J. Physiol. (Lond.), 207 (1970) 31-50. 12 JONES, E. G., AND POWELL,T. e. S., Electron microscopy of the somatic sensory cortex of the cat. 1. Cell types and synaptic organization, Phil. Trans. B, 257 (1970) I 11. 13 JONES,E. G., AND POWELL, T. P. S., Electron microscopy of the somatic sensory cortex of the cat. Ill. The fine structure of layers II1-VI, Phil. Trans. B, 257 (1970) 23-28. 14 JONES, E. G., AND POWELL, T. P. S., An electron microscopic study of terminal degeneration in the neocortex of the cat, Phil. Trans. B, 257 (1970) 29-43. 15 JONES, E. G., ANn POWELL, T. P. S., An electron microscopic study of the laminar pattern and mode of termination of afferent fibre pathways in the somatic sensory cortex of the cat, Phil. Trans. B, 257 (1970) 45-62. 16 KANE, E. C., Octopus cells in the cochlear nucleus of the cat: heterotypic synapses upon homotypic neurons, Int. J. Neurosci., 5 (1973) 251-279. 17 KILLACKEY, H. P., Anatomical evidence for cortical subdivisions based on vertically discrete thalamic projections from the ventral posterior nucleus to cortical barrels in the rat, Brain Research, 51 (1973) 326-331. 18 KILLACKEY,H. P., AND LESHIN, S., The organization of specific thalamocortical projections to the posteromedial barrel subfield of the rat somatic sensory cortex, Brain Research, 86 (1975) 469-472. 19 LEVAY, S., Synaptic patterns in the visual cortex of the cat and monkey. Electron microscopy of Golgi preparations, J. comp Neurol., 150 (1973) 53-86.

251 20 LIEBERMAN, A. R., Comments on the fine structural organization of the dorsal lateral geniculate nucleus of the mouse, Z. Anat. Entwickl.-Gesch., 145 (1974) 261-267. 21 LORENTE DE N6, R., La corteza del raton, Trab. Lab. Invest. biol. (Madr.), 20 (1922) 41-78. 22 LORENTE DE N6, R., Architectonics and structure of the cerebral cortex. In J. F. FULTON (Ed.), Physiology of the Nervous System, Oxford University Press, London, 1938, pp. 291-330. 23 MASSING,W., AND FLEISCHHAUER, K., Further observations on vertical bundles of dendrites in the cerebral cortex of the rabbit, Z. Anat. Entwickl.-Gesch., 141 (1973) 115-123. 24 MOUNTCASTLE, V.B., Modality and topographic properties of single neurons of cat's somatic sensory cortex, J. Neurophysiol., 20 (1957) 408-434. 25 PASTERNAK, J. F., AND WOOLSEY, T. A., The number, size and spatial distribution of neurons in lamina IV of the mouse SmI neocortex, J. comp. NeuroL, 160 (1975) 291-306. 26 PETERS, A., PALAY, S. L., AND WEBSTER, H., The Fine Structure of the Nervous System, Hoeber, New York, 1970. 27 PETERS, A., AND WALSH, T. M., A study of the organization of apical dendrites in the somatic sensory cortex of the rat, J. comp. Neurol., 144 (1972) 253-268. 28 POWELL, T. P. S., AND MOUNTCASTLE, V. B., Some aspects of the functional organization of the cortex of the postcentral gyrus of the monkey: a correlation of findings obtained in a single unit analysis with cytoarchitecture, Bull. Johns Hopk. Hosp., 105 (1959) 133-162. 29 REESE, T. S., AND KARNOVSKY, M.J., Fine structural localization of blood-brain barrier to exogenous peroxidase, J. Cell BioL, 34 (1967) 207-217. 30 RICHARDSON, K. C., JARETT, L., AND FINKE, E., Embedding in epoxy resins for ultrathin sectioning in electron microscopy, Stain Technol., 35 (1960) 313-323. 31 SLOPER, J. J., An electron microscopic study of the termination of afferent connections to the primate motor cortex, J. NeurocytoL, 2 (1973) 361-368. 32 VAN DER LOOS, H., AND WOOLSEY, T. A., Somatosensory cortex: structural alterations following early injury to sense organs, Science, 179 (1973) 395-398. 33 VENABLE,J. H., AND COGGESHALL, R., A simplified lead citrate stain for use in electron microscopy, J. Cell BioL, 25 (1965) 407. 34 WELKER, C., Microelectrode delineation of fine grain somatotopic organization of SmI cerebral neocortex in albino rat, Brain Research, 26 (1971) 259-275. 35 WELKER, C., AND WOOLSEY, T. A., Structure of layer IV in the somatosensory neocortex of the rat: description and comparison with the mouse, J. comp. NeuroL, 158 (1974) 437-454. 36 WELLER, W. L., AND JOHNSON, J. I., Barrels in cerebral cortex altered by receptor disruption in newborn, but not in five-day-old mice (Cricetidae and Muridae), Brain Research, 83 (1975) 503-508. 37 WOOLSEY, T. A., DIERKER, M. L., AND WANN, O. F., Mouse SmI cortex: qualitative and quantitative classification of Golgi-impregnated barrel neurons, Proc. nat. Acad. Sci. (Wash.), 72 (1975) 2165-2169. 38 WOOLSEY, T. A., AND VAN DER LOOS, n . , The structural organization of layer IV in the somatosensory region (SI) of mouse cerebral cortex, Brain Research, 17 (1970) 205-242. 39 WHITSEL, B. L., PETRUCELLI, L. U., AND WERNER, G., Symmetry and connectivity in the map of the body surface in somatosensory area II of primates, J. Neurophysiol., 32 0969) 170-183.