The Number, Size and Spatial Distribution of Neurons in Lamina IV of the Mouse Sml Neocortex' JOSEPH F. PASTERNAK A N D THOMAS A . WOOLSEY2 D e p n r t m e n t of A x t r t o m y , Wtrshrngton U n i v e r s i t y School of Medicine, St. L oui s, Missoicri 631 10

ABSTRACT We located the corresponding barrel in Layer IV of the mouse SmI cortex in eleven cerebral hemispheres sectioned in a plane tangential to the pia overlying SmI and in one sectioned in a plane normal to the pia. All of the brains were serially sectioned and prepared by a combined Golgi-Nissl method. In the section in which barrel C - l could be optimally visualized each neuronal soma was outlined with a camera lucida and the cross-sectional area measured with the aid of a small computer. In all, nearly 7,000 neurons were measured. We estimate that on average barrel C - l contains about 2,000 neurons. The mean cross-sectional area of the perikarya is 62.51 l 2(S.D. * 14.51 p2) and the size distribution of the neurons is unimodal and positively skewed. There is no segregation of cells within the barrel on the basis of size. The spatial distribution of cells in the barrel is fairly constant, from specimen to specimen, and the characteristic cytoarchitectonic appearance of the barrel can be related to regional neuronal packing density since there are at least 1.6 a s many neurons in the sides of the barrel as the hollow. The constancy of the cellular composition of the barrels indicates that the mechanisms responsible for the development of the mouse SmI cortex are fairly rigidly determined, and that the barrel field should lend itself well to further quantitative, developmental and physiological analysis.

Shortly after histological techniques became available for the study of the central nervous system, neuroanatomists attempted to count and measure the structures they observed. Attention soon focused upon the cerebral cortex, and in 1872 Meynert estimated that there were 6.12 X 108 neurons in the cortex of one human cerebral hemisphere (Meynert, 1872). Betz (1874) measured the width and length of the cells in the precentral gyrus that now bear his name and from measurements of cortical neurons in different regions of the precentral gyrus, Campbell ('05) was able to correlate the sizes of the giant cells with the functional organization of the motor cortex. Summaries of these and other quantitative studies of the cerebral cortex have been presented elsewhere (Blinkov and Glezer, '68; Konigsmark, '70; Von Bonin, '73). Despite many studies, the few attempts to determine the variability of mammalian cerebral cortices have been restricted to the measurement of brain J.

COMP.

NEUR.,160: 291-306

weight, fissural pattern (Mall, '09), cortical surface area (Witelson and Pallie, '73) and neocortical volume (Wimer et al., '69). This is mainly because the number of cortical cells is difficult to precisely estimate. What we would really like to know is the variability in the neurons that comprise a well-defined portion of the cerebral cortex among different animals of the same species. This would provide a measure of the precision of the genetic and developmental events which lead to the normal adult cerebral cortex, as well a s giving a base from which a reliable determination of the effects of experimental and environmental factors on its morphology could be made. To our knowledge, attempts to enumerate an entire population of neocortical 1 Some of this work was presented a t the Third Ann u a l Meeting of the Society for Neuroscience, November, 1973. Address reprint requests to Dr. Woolsey. 2 T h i s study was supported by U.S.P.H.S. Grant NS 10244 from the National Institute of Neurological Diseases and Stroke.

29 1

292

JOSEPH F. PASTERNAK AND THOMAS A. WOOLSEY

neurons have been restricted to the Betz cells (although even Betz cells are identified somewhat arbitrarily). Lassek (’40) was able to count these neurons in both cerebral hemispheres from a single individual and found a 1o/o difference between the two sides. These large neurons can be counted because they are easily recognized and are confined to a well-defined cortical region. Nevertheless, the extent of the motor strip in man is considerable and the approximately 34,000 neurons per hemisphere that Lassek counted represent an absolutely large number (especially for laborious neuron counts) although in a relative sense (for the human cerebral cortex) the value is small. A more favorable site, in which to determine the variability of a discrete population of cortical neurons, is Lamina IV of the mouse SmI neocortex. This area consists of a distinctive cytoarchitectonic field consisting of multicellular units which Woolsey and Van der Loos (‘70) have named “barrels.” The larger barrels are remarkably constant in numbers ( 3 5 4 0 ) and arrangement (5 rows) from specimen to specimen. Together they comprise the posterior medial barrel subfield (PMBSF) and each barrel has been shown to be related to the large contralateral mystacial vibrissae (Van der Loos and Woolsey, ’73). The constancy of the barrels in the PMBSF permits the selection of homologous groups of cortical neurons in many different specimens. In this study we have undertaken a quantitative study of the neurons in an identified barrel in a number of specimens. In all, over 8,700 cortical neurons were counted and the cross-sectional areas of over 7,000 neurons measured. MATERIALS AND METHODS

Animals. The data are taken from nine Swiss mice of both sexes. The animals were from three different litters, weighed between 20 and 26 gm, and ranged in age from 89-120 days. Tissue preparation. Following anesthesia with a n intraperitoneal injection of 3.5% chloral hydrate (dose: 1 cc/100 gm body wt), the animals were decapitated and the heads rapidly immersed in the Cox fixative after the parietal bones were separated along the sagittal suture to allow

more direct fixation of the cerebral cortex. Following two day’s fixation, the brains were removed from the skull and placed in fresh Cox solution. After fixation for 22 to 25 additional days, the brains were divided in the midline and each hemisphere was embedded in celloidin (Ceducol). Serial sections were cut at thicknesses varying from 40-56 p in a plane parallel to the pia overlying barrel C-1 (Woolsey and Van der Loos, ’70). One hemisphere was c u t at 42 p in a plane normal to the pia and oblique to the coronal in order to parallel the major axis of barrel C-1. The sections were collected in 70% alcohol, and processed by the method of Van der Loos (’56). The Golgi precipitate was either removed or developed prior to Nissl staining. To improve the Nissl counterstain, we modified the Van der Loos combined procedure by substituting 75 minutes in distilled water with repeated agitation prior to staining for 10 minutes in 1 % thionin at 56°C for Van der Loos’ 90 minutes in 0.2% methylene blue-C1. This modification improved the quality of the counterstain in the combined Golgi-Nissl preparations without appreciable leaching of the Golgi and provided better control for the differentiation. The sections were mounted in Ceadex and covered with No. 0 coverslips. Datu collection. The entire PMBSF was reconstructed from serial tangential sections, by tracing the individual barrel outlines with a camera lucida using identified blood vessels for orientation. From these drawings barrel C-1 could always be located (fig. 1). Usually barrel C-1 appeared in two or three adjacent sections; the section in which the septum surrounding C-1 was most distinct was chosen for further study. The boundary of C - l was taken to be the interface between the barrel side and the septum. Photomicrographs of the optimal section and the sections immediately deep and superficial to the optimal one were taken and used to identify this boundary. In addition, these photomicrographs were useful for determining the position of individual neurons within the specimen as they were being drawn. Barrel C - l was found in the specimen cut normal to the pia by projecting the section outlines onto paper and carefully noting the position of barrel elements (sides, hol-

293

MEASUREMENTS OF MOUSE BARREL NEURONS

lows and septa). (See Woolsey and Van der Loos, '70 for the definition of these terms.) The barrel field was then reconstructed in a manner similar to that used by Hubel and Wiesel ('72) for mapping terminal degeneration in the monkey striate cortex and a section through the long axis of barrel C-1 was studied. We traced all neurons within the barrel boundaries, using 100 X objective and the Zeiss drawing apparatus (final magnification X 1,500). The main criteria for identifying a cell as a neuron were a large central nucleolus and visible cytoplasm. Astrocytes, oligodendrocytes, capillary endothelial cells, and cells which appeared

to be neurons but lacked a n identifiable nucleolus were not traced. The neuronal boundaries were outlined in the plane of focus of the nucleolus (fig. 4). Cross-sectional areas of both barrel C-1 and the individual neurons were determined with the aid of a small computer (Cowan and Wann, '73). In order to assess regional differences in neuronal sizes and packing density the traced barrels were arbitrarily divided into a maximum of five concentric rings 20 p in width (Rl-Rz); the outer ring includes the whole of the barrel side and is designated Rl (fig. 5). To determine neuronal packing density and to estimate the number of neurons

I

1mm

Fig. 1 To show the posterior medial barrel subfield (PMBSF) from six of the mouse cerebral hemispheres and the nomenclature of the PMBSF. On the right of the figure is the drawing of PMBSF from SmI of the left cerebral hemisphere of the mouse. Orientation with respect to the brain is given above: m, medial; 1, lateral; p. posterior; a, anterior. Note the barrels are arranged in five distinct rows labeled A through E, and that each row has several barrels labeled 1 through 4, for row A. Four large barrels straddle the five rows and are labeled alpha through delta. Barrel C-1,which was chosen for this study, 1s indicated in black. On the left of the figure the position of barrel C-l is similarly indicated in PMBSF camera lucida reconstructions from six hemispheres used in this study.

2 94

JOSEPH F. PASTERNAK AND THOMAS A . WOOLSEY

per barrel, we measured the thickness of each section using the 100 X objective and the calibration on the fine focus adjustment of the microscope always moving in an antigravity direction to reduce mechanical backlash. RESULTS

Qualitatiue obseniations Six reconstructed PMBSF's are shown in figure 1 along with the key diagram indicating the nomenclature of the barrel field which is taken from the paper by Van der Loos and Woolsey ('73). Barrel C - l occupies a position in the PMBSF which is particularly easy to identify and for this reason, as well as its relatively larger size, was studied. The reconstructed

PMBSF's, which are taken from three pairs of hemispheres, show the remarkable constancy of this cytoarchitectonic formation in Layer IV; specifically, barrel C - l could always he located. Some of the variation in these reconstructed views is in part due to variation in the tangential plane of section of each of the brains (Woolsey and Van der Loos, '70). A low power view of a single section through the PMBSF of one of the specimens is shown in figure 2. The sides, hollows and intervening septa can be recognized and using these features it is possible to recognize portions of barrels from each of the five rows in figure 2. In figure 3 the side and hollow and septum surrounding barrel C-1 can be clearly seen.

Fig. 2 A low-power view of single tangential section through the right hemisphere of mouse 32R to show the cytoarchitectonic appearance of barrels in combined Golgi-Nissl material. Anterior, to the right; posterior, to the left; medial is up; lateral, down. Barrels from the five major rows (labeled A-E) are seen moreor-less distinctly with portions of the "straddlers" visible in the upper part of the figure. Note that in the center of the photomicrograph individual barrels c a n be clearly distinguished, and that some Golgi-impregnated neurons as well a s nonspecific Golgi precipitates are also present. Bar = 500 p ; thickness = 50 p ; GolgiCox preparation counterstained with thionin.

MEASUREMENTS OF MOUSE BARREL N E U R O N S

In addition, the photomicrograph shows neurons which have been impregnated by the Golgi method but the relative frequenc y of these cells in contrast to the Nissl stained neurons is low. An indication of the quality of the cytology obtained in these Golgi-Nissl preparations is given in figure 4. In the large cell marked by the arrow one can easily identify a nucleus containing a sharp nucleolus a s well as some additional clumps of chromatin. The pattern of nuclear chromatin created occasional problems in determining the optimal plane in which to draw the cell, but it was usually possible to identify the nucleolus which was larger

295

(about 2.5 k in diameter in this case), occupied a more nearly central position in the nucleus and frequently exhibited a metachromatic pinkish color. The cytoplasm of these neurons showed variable intensity of Nissl staining. The large cell shown in figure 4 has dark cytoplasm with a few Nissl granules, an appearance which is typical for all neurons with cross-sectional areas 2 S.D. greater than the mean. Other neurons seen in the field, while showing a clearly recognizable cytoplasm, are obviously paler staining. Unfortunately we have been unable to correlate cytological characteristics with neuron location within the barrel.

Fig. 3 A somewhat higher power view to show barrel C - I . The barrel is clearly recognizable because of the Nissl counterstain. Note that the barrel apparently consists of a region of greater cell density (the side) surrounding a n area of lesser cell density (the hollow). Barrel C - l is more-or-less distinctly segregated from the neighboring barrels by a narrow, nearly acellular septum. Occasional neurons are impregnated by the Golgi method, but have been truncated by a microtome. The rectangular outline in the center of C - l identifies the area shown in figure 4. Bar = 100 f i ; section thickness = 52 j ~ Orientation . a s in figure 2.

296

JOSEPH F. PASTERNAK AND THOMAS A WOOLSEY

Fig. 4 Three photomicrographs of the field indicated in figure 3 , taken a t different focal planes to illustrate the cytology obtained in the combined Golgi-Nissl preparations and the criteria for tracing a neuron (indicated by +). A , A photomicrograph taken at a focal plane slightly above that shown in B. The nuclear architecture of the large central neuron cannot be appreciated. B, A photomicrograph taken at a focal plane of the nucleolus of the large central neuron. Note the relatively central position of the nucleolus and peripheral position and smaller size of clumps of nuclear chromatin. The cytoplasm of this large neuron c a n be seen surrounding the nucleus. The outline of this cell was traced i n this focal plane. C, A photomicrograph of the same neuron taken a t a focal plane below that shown in B. Again note the absence of a clear nuclear profile. All photomicrographs are reproduced at the same magnification. For further details see text.

Photographs of the large cell in figure 4 show the appearance of a single neuron at different focal planes. The cross-sectional area that one would obtain by tracing the outline of this cell a particular plane of focus might be expected to vary considerably. Therefore, the plane in which all of the cells were drawn was operationally defined as that in which the nucleolar diameter was maximal. When proximal dendrites were stained (as in fig. 4A), the cell perimeter was drawn to include only portions of cytoplasm which had the staining characteristics of the soma.

Barrel size and cell n u m b e r In table 1 we have listed the eleven specimens in which barrel C - l was examined in a tangential plane, the barrel C - l cross-sectional areas, the section thickness, the number of neurons in barrel C - l and a n estimate of the total number of cells in barrel C-1. There is considerable range in the barrel cross-sectional areas from 47,000 to 68,000 p'. The standard

deviation of perikaryal counts is about 20% of the mean. From these counts, we estimated the total number of neurons in barrel C-1, to be 2,000 (S.D. 14.3%), assuming that the thickness of Layer IV is 120 p and that there are no significant changes in cell number as a function of depth. Examination of the data from the coronal section (fig. 7) shows that both of these assumptions are justified as well as why it is impractical to actually count all of these neurons using adjacent sections from these specimens. The variability in the barrel size measurements and in the neuron counts presented in table 1 may be in part biological, but also may have a contribution from various technical problems. The first of these is due to differences in the angle of section through the barrel. Sections deviating from the truly tangential plane would increase the apparent cross-sectional area; 1 0 " error in section plane would only increase the cross-sectional area by 1.595, but a 20" error would increase the area

297

MEASUREMENTS OF MOUSE BARREL NEURONS TABLE 1

S i t m m r i v y of brcrrel C-1 cttrtcc Specimen

Barrel area pz

Section thickness p

Number of neurons

Estimated neuronslbarrel

2 LZ 4 LZ 14 L 22 L 31 L 31 R 32 R 33 L 33 R 34 L 34 R

47713 64556 55379 54673 47920 46830 66525 57600 61282 68113 52047

44 42 44 56 40 41 52 40 44 48 51

666 886 68 1 909 506 588 987 68 7 70 7 696 886

1816 2531 1857 1948 1518 1721 2278 2061 1928 1740 2085

-

8199 745 k 150

1953 +- 280

Total Mean

S.D. I 2

56603 -t 7700

45.6 i 5.37

I

-

Assumes thickness of Layer IV is 120 u Nissl stain only.

by 6 . 4 % . We evaluated the angle of section through barrel C-1 by locating where our sections first encountered the barrel field and observing the order in which the remaining barrels were seen in deeper sections. In this way we were able to determine crudely the angle of section; however, we could not reduce the variability of barrel areas by correcting for these angles. In fact, the largest C-1 barrel (34Lj was approached directly. A second problem is determining the boundaries of each barrel. Although we used sections showing C - l optimally in each case, certain portions of the septum were identifiable only with great difficulty and often we had to examine the section immediately deep or superficial to the optimal section to help determine the exact location of the septum-side interface. However, the contribution of this error is probably small and is likely to be systematic. A third contribution to the variation could be true differences in cross-sectional diameter as a function of depth in Layer IV. Previous studies suggest that the barrel has maximal cross-sectional area in the middle of Layer IV with decreasing values as one nears Layer I11 or Layer V (Woolsey and Van der Loos, '70). From the analysis of the spatial distribution of the neurons in barrel C-l in the vertical direction, the way in which the optimal sections were chosen, and the use of relatively thick sections, it is likely that all of the material

analyzed was from nearly the same level in the barrel. The fourth difficulty comes from fixation and subsequent processing. From the results of the individual cell measurements this factor is probably quite small. Controls are present in the data in table 1 in the pairs of hemispheres taken from three animals 31, 33 and 34 in which both hemispheres were fixed and imbedded together presumably minimizing any artifacts due to processing. Our conclusion is that barrel C-1 has a cross-sectional area and corrected neuronal counts which vary among different specimens with a S.D. of 13.6% and 14.3% of the mean value, respectively, but that some proportion of this variability may be artifactu al. In table 1, it can be seen that the sections ranged in thickness from 40-56 p. These thick sections are necessary to facilitate the recognition of barrel C-1 and to gather substantial numbers of cells from a single section. Perhaps the greatest value of these sections was to reduce the overestimation of the numbers of neurons in each specimen. Since only cells with nucleolus were drawn, using Abercrombie's correction (Abercrombie, '46), our figures overestimate the real density of nucleolar points by, at most, 5 % .

Spatial distribution of neurons The packing density of neurons within the barrels was examined qualitatively and

.... .......... ............. .............. ............... ............... ............... .............. ............

......... ............ .............. ................ . .............. ............. ......... ....... .......

298

JOSEPH F. PASTERNAK AND THOMAS A . WOOLSEY

0 0 0

0 . 0

0

0

0

0.0.0.

0 0

0.0..

0

0 . .

0 .

31R

......... ............ ................ . .................. .................. .. - .'..'.. .................. ............... ............ 0.0

0 . 0 . 0 . -

0 . 0 .

.a*.

0 . 0 .

0 . .

0..

0 .

0 . .

0 . 0 0 . .

33 R

Tang entia I

........... ............... ............... ................. ............... -......* ....... 111 -

Neuron5

/ 4 0 0 y2

0 . 0

0.0

0 0 0 0 0 0 0 . 0

0.

21 R

V

0-3

*

4-7 8-11 12-15

0

0

C o ronal Fig. 5 To show the spatial distribution of neurons in barrel C - I . The upper four specimens are tangential specimens through barrel C - I . The bottom center of the figure is representative of the coronal section through barrel C-I. For each of these specimens a n arbitrary grid of squares 10 u , on a side was placed over the camera lucida drawing of the cells in the barrel. The number of neurons falling within each square was counted and the results of these counts are indicated schematically by dots of different sizes according to the key at the lower righthand corner of the figure. Notice that there are more cells in the periphery of each barrel than i n the hollow.

quantitatively. Figure 5 shows the distribution of neurons in four tangential C-1 specimens by schematically displaying the number of cells in 100 pz squares of a grid arbitrarily drawn on the barrel. Greater neuronal packing in the barrel side is

demonstrated by the predominance of larger dots in the periphery. A similar pattern is seen in the plot for the coronal section but the contrast between side and hollow is not as great. The location of each of the 687 neurons

MEASUREMENTS OF MOUSE BARREL N EU RO N S

counted in C-1 from specimen 33L is indicated by the dots in figure 6. Also depicted in this figure are the outlines of the 20 p concentric rings (Rl-R5), used to determine regional densities. Qualitatively there may be more neurons in R1, which includes the side, than in R2-R3, but the difference is not particularly striking. Since the section thicknesses are known as well as the areas of each ring, neuronal densities could be calculated for the five concentric rings in each tangential specimen and the means and S.D.’s are shown in table 2. On average, the cell density in R1 was 1.58 times the density in the remaining rings. Since RI was arbitrarily defined, it probably contains the entire side and perhaps a small portion of the hollow and septum, the neuronal packing in the true side must be at least 1.58 times that in the hollow. Packing density was consistent

299

from R2 through R:, implying that neurons are evently distributed in the hollow itself. These data indicate that the neuronal density drops abruptly at the side-hollow interface rather than gradually declining from the periphery centripetally. A coronal section through C-1 was examined to determine neuronal distribution patterns along the vertical axis. Figure 7A shows a drawing of this barrel seen in Layer I V divided into the customary five rings and seven vertical levels from Layer I11 to Layer V. Level a undoubtedly includes some cells from cortical Layer I11 and may in fact be predominantly from Layer 111. Figure 7 (parts B,C) shows neuronal density and size data from the coronal section. The mean neuronal size changed little a s a function in depth; likewise overall neuronal density showed little vertical change, being on average about

Fig. 6 A plot of the actual position of each neuron traced for barrel C - I . Small dots were placed over each neuron specimen 33L. T h e larger dots indicate the position of the four Golgiimpregnated neurons in this specimen. To determine differences in regional neuronal distribution, many of the specimens studied were divided into 20 c~ concentric rings from without in: here numbers 1 through 5. Notice that R,,which includes the side, apparently has a greater density of neurons but the difference is not striking.

300

JOSEPH F. PASTERNAK AND THOMAS A. WOOLSEY

~

Specimen

Ring 1

2L’ 4L’ 14 L 22 L 31 L 31 R 32 R 33 L 33 R 34 L 34 R 1 21 R ”

4.34 4.44 3.95 3.74 3.35 4.08 4.04 3.91 4.02 3.58 3.30 4.74 3.85 i 0.28

’

Mean S.D. I

2

Ring 2

Ruig 3

Ring 4

Ring 5

Hollow

-

-

-

-

2.47

2.43

2.55

2.15

-

-

-

2.55 2.55 2.79 2.70 1.74 2.53

2.17 2.63 2.34 2.71 2.14 2.21

2.43 2.83 2.11 2.45 1.92 2.04

-

3.36 2.38 2 0.23

3.13 2.33 t 0.32

2.64 2.84 2.42 2.52 2.37 2.59 2.46 2.59 2.03 2.23 2.37 3.15

-

3.37 2.48 f 0.34

-

-

2.18 2.77 2.39 2.31 1.85 -

3.05 2.28 t 0.31

-

Ratio sideihollow

1.64 1.56 1.63 1.48 1.41 1.58 1.64 1.51 1.98 1.61 1.39 1.50 1.58 k 0.15

Values excluded from computation of m e a n s a n d standard deviations Coronal section

3.3 x 10 4/p:’ if level a is ignored.:’ This constancy indicates that overall density figures from tangential sections are relatively independent of the depth in Layer IV from which they are taken. The density ratio between side and hollow did change as a function of depth with the deeper levels (especially e,f) of the barrel shown to have increased side to hollow ratio (fig. 7D). This indicates that a portion of these deeper levels must be included in the section for the barrel to be recognized. Cells in the coronal specimen have smaller cross-sectional areas, and are more densely packed (fig. 7) than those in the tangential specimens (tables 2, 3 ) . This is apparently due to differential tissue shrinkage during processing and possibly compression during sectioning since the side to hollow ratio in this coronal specimen is comparable to that observed in the other brains. As a result, Layer IV may be thicker than 120 p measured in this brain. The principal value of the coronal specimen is to show that neuronal numbers in the barrel do not change a s a function of depth in Layer IV, suggesting that our estimate of number of neurons in a barrel while low is not grossly inaccurate.

Neuronal sizes A plot of neuronal cross-sectional areas (in 2.5 p? wide bins) against frequency was made for each tangential section studied. Four examples are shown in figure 8. Although the absolute frequencies

varied, the curve shapes were quite consistent among the ten different specimens. Each distribution was unimodal with a slight skew to the right in the larger cells. The means and S.D.’s for the neuronal areas in all specimens studied are shown in table 3 . The means for the tangential specimens are quite consistent; the mean for the coronal section is somewhat smaller but still within one S.D. limit of all other specimens. The neuronal sizes and densities are roughly comparable to those determined by Haddura (’56) in the visual cortex of the adult mouse. We sought sub-populations of neurons based on size which were selectively localized to the side or hollow. From table 3 it is clear that there are no significant differences in the mean neuronal sizes when one compares the side and hollow. When the frequencies of neurons are plotted against neuronal size for both hollow and side from the total population, as in figure 9, the resultant curves have nearly identical shapes, means and S.D.’s. If sub-populations exist they might tend to be burried in the larger population. In particular, we wondered if the relative density of “large neurons” (two S.D.’s above the mean) depended upon barrel ring. We were unable to find any signifi:$Expressing neuronal density In this w a y seems curious but to OUF minds no odder t h a n the statement t h a t there are 3.3 X 10% neuronimms. Perhaps stating t h a t there a r e 3.3 neuronsi10.000 cL:$or t h a t the cortical volume for each neuron is about 3,030 p?gives a better. if less conventional, feel for these numbers.

301

MEASUREMENTS OF MOUSE BARREL NEURONS

1

pia

-111

A

IV

loop V Meon

Neuron

S i z e pz

57.04

C

D

I

5 7.07 56.03

53.51 54.05 54.51 55.23

1 Neurons

X

2

3

Side / H o l l o w

Fig. 7 To show the results obtained in a coronal section traversing the long axis of barrel C-I. A. Shows the arbitrary division of the barrel into the customary five rings, 20 p in width, and a further subdivision from Layer 111 to Layer V, into seven different levels labeled ( I through g. T h e direction of the pia and the approximate boundaries of Layer IV with adjacent cortical Layers 111 and V are indicated on the right. B, This indicates the neuronal density according to t h e arbitrary levels shown in Part A. Each bar indicates the value for that level, and the stipples indicate o n e standard deviation about the mean of all these values. With the exception of level (I (which possibly includes a bit of cortical Layer 1111, there is no slgnficant change in cell density a s a function of vertical depth through Layer IV. C, This table summarizes the mean cross-sectional area of neurons in barrel C-1 a s a function of cortical depth. Neuronal size does not vary appreciably with vertical position in this barrel. D, This figure summarizes the vertical variations in the ratio of neuronal densities in the side ( R , ) to the hollow (Re-R:). The hatching indicates the standard deviation about the mean of this ratio, taken from the tangential specimens a n d summarized in table 2. Note that there is considerable variation of the ratio a s a function of vertical cortical depth and that only in levels rl through g do the values fall close to those determined for the tangential sections. For further discussion, see text.

cant regional differences and thus were unable in this case to find any selective localization of neurons by size within the barrel. A s Leowy ('72) has shown this finding is not incompatible with the existence of different cell types with the population; our own Golgi studies (Woolsey et al., '75) indicate that there are at least six cell types within the barrels. Finally, i t should be noted that the mea-

surements made on the paired hemispheres from animals 31, 33 and 34 apparently do not exhibit significantly less variability when left and right are compared than is observed when hemispheres from different animals are compared (tables 1-3). DISCUSSION

T h e cytoarchitectonic basis of the barrel We have presented evidence that neu-

302

JOSEPH F PASTERNAK A N D TIIOMAS A WOOLSEY TABLE 3

Sumrnciiy of cell cioss-secttowrl Side

(Iit’(l5 112

11’

Hollow

Total population

Specimen

2L 4 L

14 L 31 L 31 R 32 R 32 L 33 R 34 L 21 L

‘

Pooled data I

Mean

S.D.

Mean

S.D.

Mean

S.D.

67.08 60.50 62.01 66.87 61.47 64.67 66.49 63.32 60.90 51.40

2 15.82 2 16.18 2 11.28 -t 16.60 2 12.09 & 13.53

t 15.96

-c 15.94 f 11.81 f 16.67 215.13 f 13.73 i 13.50 f 15.07 i 11.78 i 13.89

66.69 61.24 63.33 66.53 62.24 65.66 66.45 63.21 60.49 55.33

t 15.86

14.83 2 12.02 f 13.04 2 10.30

66.56 61.68 64.22 66.27 62.79 66.26 66.43 63.12 60.23 56.31

62.33

c 14.25

62.63

i 14.67

62.51

?

I 16.03

t 11.64 t 16.48 I 13.94

t 13.68 I 14.01

t 13.77 t 12.29 t 13.49 t 14.51

Coronal section

60

40

10

0

0

80

1_!\

20

1

40

60

80

100

120

140

I

H

33 1

33 R

20

0

20

40

60

80

100

110

140

160

Neuron Areas p 2 Fig. 8 Histograms to show the size distribution of neurons in barrel C - l from four different tangential specimens. Along the abcissa neurons a r e grouped into 2.5 square micron bins and the frequency is plotted on the ordinate. T h e curved arrows indicate the means and the solid triangles denote one S.D. above and below the mean. Note that while the absolute n u m ber of cells varies among the different specimens, that the shape of all histograms is roughly comparable with a skew towards the larger cells. For fuller description, see text.

303 10

V

8

C

.2c

7

0 I

3

6

Hollow; N = 4 2 3 4

P 0

a

0

5

0

*

0

4

c

I

3

2

1

0

10

20

30

40

50

60

70

80

Neuron

90

100

110

120

130

140

.

.

150

160

170

Areas y 2

Fig. 9 To test whether there is any regional segregation of neurons on the basis of soma size, the distribution of neuronal areas in the side (R,) is compared directly with the neuronal areas for the hollow. Both populations have been normalized, the frequencies being expressed on the ordinate as percentages of the total population. The distribution for the side is indicated by the stippling and the distribution for the hollow is indicated by the open lines. Note the near identity of the two histograms and that the means (curved arrows) and S.D. (solid triangles) for both populations are the same.

ron perikaryal packing density is greater in the barrel side than in the hollow. The overall mean side hollow ratio was 1.58, with individual variation from 1.39 to 1.98. Are these ratios sufficient to account for the striking side-hollow contrast which produces the distinctive barrel form? We could not find any selective localization of neurons by size to either the side or hollow; and the individual neurons in the side do not appear more deeply stained than neurons of similar size in the hollow, although at present such comparisons are qualitative. Thus, the major factor responsible for barrel appearance in Nissl preparations is related to regional neuronal density. It is probable that the side-hollow density ratio of 1.58 is a n underestimate of the true ratio. The side-hollow ratio was calculated using rings arbitrarily constructed in the barrels. Because R1 was constructed 20 p wide to be certain that

all of the side was included, it almost certainly extends into the hollow and possibly to a lesser extent the septum diluting the neuronal density estimated for the side. However, the true side-hollow ratio may not be much greater than 1.58 since the amount of light transmitted through different areas of the barrel is related to the concentration of absorbing material (Nissl dye and hence neuronal packing density) by a logarithmic function. Small changes in neuronal density could produce large changes in the quantity of light transmitted. This spatial distribution of neurons in the mouse is quite different from what has recently been observed in comparable barrels in the rat (Welker and Woolsey, '74). The unusual arrangement of the neuronal somata in the barrel suggests that there is a corresponding segregation of certain elements comprising the neuropil.

304

JOSEPH F. PASTERNAK AND T H O M A S A . WOOLSEY

In his classic paper, Lorente de NO (’22) described and illustrated a number of neuron types which had part or all of their processes (both dendrites and axons) confined to the region of the barrel hollow. In Golgi-Cox preparations (Woolsey et al., ’75) over 90% of the impregnated neurons in Layer IV have their dendritic fields restricted to the barrel hollows. A second major source of processes in Layer IV is extrinsic to the cerebral cortex from the ventro-basal thalamus. In degeneration studies the terminals of these fibers are extremely dense and seem to be confined to the region of the hollow (Killackey, ’73). Their appearance as described by Lorente de NO (’22) is quite remarkable: y ?&to7 1 0 1’s ? l t l d t l i 7 1 , f 1 ~ K 1 l t ~ 1 l t 1 ’ , irnprc~giitrii t v i totttlirlttd ltrs rtimcis dc. (I?horizacion, ciptrrrcr Itr ctrpcc IV diuitlitltc (’71 ccimpos ouciludos, reple,tos dt, jihrilltis P O j i z c t s - ( I ctriis(r tli, s i l tlcdgtrdrz - tcoi tcprr“ClllllldO,

se

t t l d t l s , qZlC’ t l p C ’ l l ( t S S I dC’jtl71 1 ’ S p C l C i O S pCtl-(I

los cicerpos celitlnres.”

4

Dez*elopmentul considerations It is of some interest to consider how the uneven distribution of neurons that make up the barrels comes about. A t birth, Layer IV of the somatosensory cortex of the mouse is a homogeneous sheet of cells (Rice, ’73). Presumably some program of cell migration, process formation with passive displacement of the somata and/or cell death causes the first barrels to form by postnatal day 3 and produce a recognizable barrel field by day 5. An hypothesis that would explain part of this sequence is that the neocortical neurons undergo a horizontal migration within Layer IV so that the hollows become less densely populated. Perhaps the thalamo-cortical afferents upon arriving in Layer IV displace neurons away from the hollow into the sides with extensive axon terminal arborization or by stimulating cortical neurons to spin out dendrites (and axons) increasing the volume of neuropil (e.g., Kelly and Cowan, ’72). The exact time when thalamic afferents arrive in Layer IV of SmI cortex is unknown but they have been seen a s early as the eighth postnatal day (Lorente de NO, ’22). On the other hand, the barrels could be formed by programmed patterns of cell death (see Cowan, ’73, for an excellent review of this subject). Of

particular interest is the pattern of neuronal death in the isthmo-optic nucleus of the chicken in which an amorphous cell mass is converted into a highly structured, convoluted, neuronal sheet (Cowan and Wenger, ’68). A similar sequence of events could likewise convert homogeneous cortex into a barrel field. A major problem is that neuronal death during development has not been seen in cortex; all the known examples are not only in subcortical structures but in sites having direct connections to the periphery. Although we do not know at present whether intracortical neuronal migration or programmed cell death play a role in the formation of the cortical barrels, both of these hypotheses have the attraction of being potentially demonstrable in the mouse cerebral cortex by studying animals of different postnatal ages. We now have some knowledge of the end-point reached in the adult animal, how variable that final result is, and that methods exist to directly determine the sequence of events leading up to it.

T h e question of variability Our figures indicate fixed genetic and developmental programs leading to barrels and perhaps neocortex in general. I t could be reasoned that comparing the two hemispheres of the same animal should correct for any existing variation (genetic and/or developmental) unless each hemisphere is genetically programmed separately or certain environmental inputs affect the animal asymmetrically. We have data for pairs of hemispheres from three animals-Nos. 31, 33 and 34. The first two pairs seem to indicate a tighter correlation of barrel size and neuronal density between the hemispheres of the same animal than is seen in hemispheres from different animals. There is no apparent reduction in the overall variability when specimens from littermates are compared. The variability which we report for the number of mouse barrel neurons is comparable to that seen in the mammalian 4 I n English, “When, ( a n d this is not uncommon) the terminal arborizations [of the specific thalamic afferelits] a r e completely impregnated, Layer 1V appears divided into oval fields, very full of reddish (because of their fineness) fibers so compact that they scarcely allow spaces for t h e cell bodies.”

MEASUREMENTS O F MOUSE BARREL NEURONS

systems where the parameter has been considered (eg., Fry and Cowan, ’72). For size, it is less than that found in at least one study in which the variation in measurements of cat lateral geniculate neurons was attributed to at least two factors: (1) genetic, and (2) tissue processing (Guillery, ’73). There can be little doubt that one of the principal advantages that the mouse offers to the mammalian biologist is that a large number of genetically “pure” strains have been produced by inbreeding (Wimer et al., ’69). For this reason we were careful to restrict our studies to individuals taken from only one strain. The second factor is nearly impossible to measure directly. However, the contribution of processing to the variability we have observed apparently is not so substantial as to rule out the subsequent pursuit of further studies of a similar kind. The values we obtained are not absolute in so far a s perikaryal size is concerned, and this should be remembered when similar measures are made on tissues prepared in other ways.

Future studies Finally, we should like to emphasize some of the practical uses to which our results can be put. We have already alluded to potential for understanding some of the developmental events that result in the unusual arrangement of Layer IV neurons in the mouse SmI cortex and these data may also be used to assess the affects of early damage to the sensory periphery ( V a n der Loos and Woolsey, ’73). Our results indicate that barrel C-1, one of the largest in the barrel field, has approximately 2,000 neurons. This quite manageable number indicates that mouse SmI cortex is a good site for comprehensive quantitative studies of neocortical structure and function, having many of the properties that have attracted students of invertebrate nervous systems (Kater and Nicholson, ’73). Work in this laboratory is in progress to identify various types of neurons with the Golgi method (Woolsey et al., ’75) and to correlate these types with perikaryon size and location within the barrel. Eventually it should be possible to extend observations to the ultrastructural level using serial EM reconstructions (Levinthal and Ware, ’72).

305

ACKNOWLEDGMENTS

We wish to thank Drs. W. J . Crossland and W. M. Cowan for their advice on the manuscript; Mrs. V. Whittington for assistance with some of the histology; Mr. M. Rhoades for photography; and Miss C. Keller and Mrs. D. Stevenson for typing the manuscript. LlTERATURE CITED Abercrombie, M. 1946 Estimation of nuclear population from microtome sections. Anat. Ree., 94: 239-247. Betz, A. U. 1874 Anatomischer Nachweis zweier Gehirnez. Zbl. Med. Wiss.. 1 2 . 578-580; 595599. Blinkov, S . M.. and I . 1. Glezer 1968 The Hum a n Brain in Figures and Tables. Basic Books, New York, pp. 172-213. Campbell, A. W. 1905 Histological Studies on the Localisation of Cerebral Function. Univ. Press, Cambridge, 360 pp. Cowan, W. M. 1973 Neuronal death as a regulative mechanism in the control of cell numbers in the nervous system. In: Development and Aging in the Nervous System. M . Rockstein and M. L. Sussman, eds., Academic Press, New York. pp. 1 9 4 1 . Cowan, W. M , and D. F. W a n n 1973 A corn puter system for the measurement of cell and nuclear sizes. J. Micros., 99. 331-348. Cowan, W. M., and E. Wenger 1968 The development of the nucleus of origin of centrifugal fibers to the retina in the chick. J. Comp. Neur.. 133: 207-240. Fry, F. J., and W. M. Cowan 1972 A study of retrograde cell degeneration in the lateral mammillary nucleus of the cat, with special reference to the role of axonal branching in the preservation of the cell. J . Comp. Neur., 144: 1-24. Guillery, R. W. 1973 The effect of lid suture upon the growth of cells in the dorsal lateral geniculate nucleus of kittens. J . Comp. Neur., 148: 41 7 4 2 2 . Haddura, M. 1956 A quantitative study of the postnatal changes in the packing density of the neurons in the visual cortex of the mouse. J. Anat. (London), 90: 494-501. Hubel, D. H., and T. N. Wiesel 1972 Laminar and columnar distribution of geniculo-cortical fibers in the macaque monkey. J. Comp. Neur.. 146: 421-450. Kater, S. B., and C. Nicholson, eds. 1973 lntracellular Staining in Neurobiology. Springer, New York. Kelly, J . P., and W. M. Cowan 1972 Studies on the development of the chick optic tectum. 111. Effects of early eye removal. Brain Res.. 4 2 . 263-288. Killackey, H . P. 1973 Anatomical evidence for cortical subdivision based on vertically discrete thalamic projections from the ventral posterior nucleus to cortical barrels i n the rat. Brain Res., 51: 326-331. Konigsmark, B. W. 1970 Methods for the counting of neurons. I n : Contemporary Research Methods in Neuroanatomy. W. J. H. Nauta and

306

JOSEPH F. PASTERNAK AND THOMAS A. WOOLSEY

S. 0. E. Ebbesson, eds. Springer, New York, pp. 315-339. Lassek, A. M. 1940 The h u m a n pyramidal tract. 11. A numerical investigation of the Betz cells of the motor area. Arch. Neur. & Psychia. (Chicago), 44: 718-724. Levinthal, C.. and R. Ware 1972 Three dimensional reconstruction from serial sections. Nature, 236: 207-210. Loewy, A . D. 1972 The effects of dorsal root lesions on the Clarke neurons in cats of different ages. J. Comp. New., 145: 141-164. Lorente d e NO, R. 1922 La corteza cerebral del raton. Trab. Lab. Invest. Biol., 20: 41-78. Mall, F. P. 1909 On several anatomical characters of the human brain, said to vary according to race and sex, with special reference to the weight of the frontal lobe. Am. J. Anat., 9: 1-32. Meynert, T. 1872 Von Gehirn der Saeugitiere. In: Handbuch der Lehre der Gewebe. Vol. 2. Stirker, ed. p. 694. Rice, F. L. 1973 Somatosensory cortex of mouse: Development of barrels and of barrel field. Anat. Rec., 175. 4231124. Van der Loos, H. 1956 Une combinaison d e deux vielles methodes histoliques pour le systeme nerveux central. Mschr. Psychiat. NeuroI., 132. 33G334.

Van der Loos, H., and T. A. Woolsey 1973 Somatosensory cortex: Structural alterations following early injury to sense organs. Science, 179: 3 9 5 3 9 8 . Von Bonin, G. 1973 About quantitative studies on the cerebral cortex. J. Micros., 99: 75-83. Welker, C., and T. A . Woolsey 1974 Structure of Layer IV in the somatosensory neocortex of the rat: Description and comparison with the mouse. J. Comp. Neur., 158: 4371154. Wimer, R. E., C. C. Wimer arid T. H. Roderick 1969 Genetic variability in forebrain structures between inbred strains of mice. Brain Res., 16: 257-264. Witelson, S. F., and W. Pallie 1973 Left hemisphere specialization for language in the newborn: Neuroanatomical evidence of asymmetry. Brain, 96: 641-646. Woolsey, T. A,, and H. Van der Loos 1970 The structural organization of Layer I V in the somatosensory region (SI) of the mouse cerebral cortex: The description of a cortical field composed of discrete cytoarchitectonic units. Brain Res., 17: 205-242. Woolsey, T . A , , M L. Dierker and D. F. Wann 1975 Mouse SmI cortex: Qualitative and quantitative classification of Golgi-impregnated barrel neurons. P.N.A.S.. (in press).