THE JOURNAL OF COMPARATIVE NEUROLOGY 306~307-319 (1991)

Dendritic Morphology of Pyramidal Neurones of the Visual Cortex of the Rat: I. Branching Patterns ALAN U. LARKMAN University Laboratory of Physiology, Oxford University, Oxford OX1 3PT, United Kingdom

ABSTRACT The aim of this study was to provide quantitative descriptions of the branching patterns of basal and apical dendrites of pyramidal neurones from the visual cortex of the rat. Thirty-nine neurones from cortical layers 213 and 5, that had been injected with horseradish peroxidase, reconstructed, and measured with the light microscope as part of an earlier study (Larkman and Mason, '90; J . Neurosci.10:1407-1414), were used. The cells had previously been divided into three classes, layer 213 cells and thick and slender layer 5 cells, on the basis of their dendritic morphology. The branching pattern of the basal and apical oblique dendrites was similar for all the cells. Between 3 and 9 basal trees arose from the soma and the number of tips in each tree varied widely, between 1 and 13. The path lengths of all the basal dendrites of a given cell were relatively constant, however. Most basal dendritic branching occurred close to the soma, such that terminal segments were much longer than intermediate segments and contributed approximately 90% of the total dendritic length of each tree. Terminal segments showed only a narrow range of diameters. Most apical oblique trees arose from the proximal part of the apical trunk. They tended to be less highly branched but were otherwise extremely similar to basal trees. Distal oblique trees were unbranched or branched only once, and their terminal segments tended to be shorter and thinner than those of basal trees. The branching pattern of the apical terminal arbors was different, with many longer intermediate segments. The terminal segments tended to be thinner than those of basal or proximal oblique trees. Slender layer 5 cells were without obvious terminal arbors. The basal and proximal oblique dendrites jointly sampled a roughly spherical volume of cortex centred about the soma, and together they accounted for the substantial majority of the cell's total dendritic shaft membrane area. Comparisons with previous studies suggest that intracellular HRP injection can yield a more complete visualization of dendritic morphology than is obtained using Golgi-based methods (unless cells are reconstructed across tissue slabs), and can therefore result in a different view of the relative importance of the various components that make up the cell's dendritic system. Key words: pyramidal cell, morphometric analysis, dendritic structure, intracellular HRP, light microscopy

Pyramidal cells are the numerically dominant neuronal type of the neocortex and may account for up to 90% of the neuronal population of the visual cortex of the rat (Peters and Kara, '85). The soma-dendritic morphology of these cells has been investigated using a range of techniques for more than a century (Feldman, '841, and each improvement in technique has permitted more complete or more detailed qualitative or quantitative descriptions. Much recent work has made use of Golgi-impregnatedneurones, in which the structure of the soma and dendrites is revealed in exquisite detail. Such neurones are traditionally examined in sections cut at a thickness of 100 to 150 pm and oriented at right B

1991 WILEY-LISS, INC.

angles to the pial surface (Peters and Jones, '841, and the majority of studies have been restricted to those parts of the cell contained within a single section. The extent of pyramidal neurone dendritic spread frequently exceeds the section thickness, so such a procedure often leads to an underestimation of cell size. A recent technical advance has been the use of intracellular injection of the enzyme horseradish peroxidase (HRP) as a neuronal marker (e.g., Snow et al., Accepted November 26,1990. Address reprint requests to Dr. A.U. Larkman, University Laboratory of Physiology, Parks Road, Oxford OX1 3PT, U.K.

A.U. LARKMAN

308

'76). This technique allows individual neurones to be stained singly, which not only facilitates observation but also makes it relatively easy to reconstruct cells across tissue section boundaries. In this way more complete representations of pyramidal neurones can be obtained. This technique also permits intracellular records to be obtained from the same cells. Previous quantitative studies of Golgi-impregnated neocortical pyramidal neurones have largely been restricted to relative measures, in which groups of cells are compared after various experimental manipulations or at different stages of development (e.g., Greenough and Chang, '88; Petit et al., '88).For such purposes, the possible incompleteness of the neurones may not be a serious problem. However, quantitative information about the dendritic structure of more complete neurones from normal cortex is of importance for many other areas of research. For example, the extent and disposition of the dendrites will influence the overall number and the composition of the mix of synaptic inputs available to an individual neurone. This is of particular interest in the neocortex, in which afferents from different intrinsic and extrinsic sources synapse in a laminar specific fashion (Gilbert, '83; Burkhalter, '89). The geometry of a dendritic arbor will also influence the way in which excitatory and inhibitory inputs to the various parts will be integrated to determine the output of action potentials by the cell. A detailed description of the dendritic structure is therefore important for the development of models of the active and passive integrative properties of the neurone. As part of a study of the morphological and electrophysiological properties of pyramidal neurones from slices of the visual cortex of the rat maintained in vitro, a sample of 39 neurones from layers 213 and 5 was studied electrophysiologically and filled with HRP. It was shown that these cells showed variation in a number of aspects of their intrinsic electrophysiology and that much of this variation could be correlated with features of their morphology (Larkman and Mason, '90; Mason and Larkman, '90). The present work and subsequent studies in this series make use of the same sample of injected neurones and examine their somadendritic morphology in more detail. In the previous works, attention was focused on the differences between cells, but in the present and following studies the emphasis is shifted to include features common to all. In this work, the dendritic branching patterns and the lengths and diameters of dendritic segments are considered. In the following two works, correlations among some of these measures and the distribution of dendritic spines are discussed. In future studies the passive electrical geometry of the cells will be considered. A discussion of the advantages and limitations of the use of the HRP injection technique in slices maintained in vitrw has been given previously (Larkman and Mason, '90).

MATERIALS AND METHODS The same sample of 39 HRP-injected pyramidal neurones from layers 213 and 5 of the rat visual cortex was used as in a previous study (Larkman and Mason, '90; Mason and Larkman, '90). Details of slice preparation and maintenance were given in Larkman et al., ('88) and are only briefly summarized here. Male Wistar albino rats, 130-160 g in weight, were used. Coronal slices, 400 pm thick, were prepared from the occipital cortex of one hemisphere and trimmed to consist entirely of visual cortex, as determined

from published cytoarchitectonic studies (Zilles et al., '84). Slices were maintained in an interface-type recording chamber at 34.5 t 0.5"C. Neurones were impaled using HIIPcontaining electrodes and injected using pressure (Mason et al., '88). Details of the reaction procedure have been given previously (Larkman et al., '88). Slices were resectioned at 60 pm thickness to facilitate penetration of reagents and improve the visualization of neurones under high power objectives. Injected neurones were reconstructed from camera lucida drawings made at x 1000 magnification using a x 100 oil immersion objective. Dendrograms showing the branching pattern of each dendritic tree were prepared for each neurone. The length and diameter of every dendritic segment were measured making use of the reconstructions and a microscope equipped with an eyepiece graticule, using a magnification of x 1875. Segment lengths were corrected for projection errors by Pythagoras' theorem, using depth values obtained from the calibrated fine focus control of the microscope. All segments were taken to be circular in cross section. The diameter of each segment (not including spines) was measured at three points along its length and the average diameter recorded. The small diameters of particularly distal dendritic segments means that considerable errors are possible in segment diameter measurements. Segments that showed substantial taper were subdivided and were not used for analyses requiring unique values for segment diameter. The dendritic branching patterns and measurements were entered into a computer program that calculated dendritic shaft membrane areas and performed the concentric shell analyses. No correction was made for tissue shrinkage during histological processing. A discussion of this and other methodological difficulties has been given previously (Larkman and Mason, '90).

The terminology employed for dendrites is as follows. Dendrites branch at branch points, which in this sample were always recorded as binary; no clear examples of trifurcations were encountered, although some intervening segments were very short. The portion of a dendrite between two branch points, between the soma and a branch point or between a branch point and the end of a dendrite is called a dendritic segment. The termination of a dendrite is called its tip, and a segment ending in a tip is called a terminal segment. Segments adjoining the soma are termed stem segments and segments between two branch points are intermediate segments. Intermediate segments branching to form two terminal segments are termed preterminal segments. The segments arising from a common ,stem constitute a dendritic tree. A collection of trees is termed a dendritic system (preferred to forest; Percheron, '79). The apical dendrite of a pyramidal cell consists of a trunk, which may give off oblique branches that are usually shorter and of smaller diameter, before possibly dividing to form a terminal arbor (Uylings et al., '78). A distance from the soma to a point on a dendritic tree measured along the course of the intervening dendritic segments is termed a path length. This was generally greater than the physical distance between that point and the soma, although usually not greatly so, since major changes of direction in the course of dendrites were rare. Minor, short wavelength undulations were extremely common, but these may be in part artfactual, caused by greater shrinkage of the overall tissue volume than the volume of the labelled neurone.

PYRAMIDAL NEURONE DENDRITIC BRANCHING PATTERNS

309

Path lengths rather than physical distances have been used in this study. For most measures, means and standard deviations have been quoted. In some cases, however, the data were clearly not distributed normally and the use of mean values would be inappropriate. In these cases median values have been given, and this has been indicated in the text. Least-squares linear regression techniques have been employed throughout (for a discussion of possible limitations, see Hofman et al., ’86).

RESULTS This study makes use of a sample of 39 neurones, 18 from layer 2/3 and 21 from layer 5, injected and reconstructed as part of a previous study (Larkman and Mason, ‘90; Mason and Larkman, ’go), and a typical HRP-injected neurone is shown in Figure 1. The layer 2/3 cells were considered together as a single cell class, but the layer 5 cells were divided into 2 classes, termed thick and slender L5 cells. Examples of each class are shown in Figure 2, and the criteria used for classification are summarized in the legend to that figure. The dendritic systems of neocortical pyramidal cells can be divided into basal and apical parts (Fig. 2). The apical part may be further subdivided into an apical trunk, apical oblique dendrites, and generally a terminal arbor, although the slender L5 cells were unusual in lacking an obvious terminal arbor. Each of these dendritic subdivisions is considered separately.

Basal dendrites A dendrogram illustrating the pattern of branching of the basal dendrites of a typical neurone from layer 213 is shown in Figure 3A. Five basal stem segments arise from the soma (sample range 3-9) and most of these branch repeatedly to produce a total of 28 tips (sample range 19-49). The degree of branching, as measured by the number of tips arising from each stem, varied widely with one stem not branching at all and another forming 9 tips. The distribution of tips per stem segment for the whole sample is broad (Fig. 3B), with a range from 1 to 13. Mean values for each cell class are given in Table 1. Dendrograms may be quantified using the method of Sholl (’53),in which vertical probe lines are laid across the dendrogram at regular intervals and the number of dendritic intersections with each line plotted as a function of dendritic path length from the soma. Figure 4A shows such a “Sholl plot” for the cell shown in Figure 3A. The number of dendritic intersections increases rapidly for the first 50 pm, then more slowly to reach a plateau between 70 and 130 pm. The intersections then decline abruptly between 130 and 150 pm. This pattern suggests that most basal dendritic branch points occur close to the soma and most basal dendrites terminate at similar path lengths. This pattern is repeated when data from the whole sample were pooled by cell class (Fig. 4B), although the rising and falling phases appear slightly less steep than for the single cell plot. This is because there is some variation in mean path lengths between cells within a class, which were not normalized prior to pooling for this plot. It can also be seen that thick L5 cells had higher peak intersection numbers than slender L5 cells, but their basal dendrites were generally shorter, confirming the impression given by Figure 2.

Fig. 1. Light micrograph of an HRP-injected neurone from layer 2/3. Scale bar 100 km.

The frequency distribution of path lengths to each basal branch point for all cells is shown in Figure 4C. The range of path lengths over which branching can occur is very wide, but it is clear that most branching occurs close to the soma. The median value for branch point path length was very similar for all three cell classes at about 25 pm, in spite of the differences in mean path lengths to tips between classes (see Table 1). This suggests that the differences between classes were in the lengths of the terminal rather than the intermediate segments. Figure 4D shows the distribution of basal tip path lengths for a single layer 213 cell. The lengths were approximately normally distributed and showed relatively little variation, with 17 out of the 23 dendrites terminating between 140 and 160 pm from the soma. This relative constancy of basal path length to tip within individual cells was quantified by calculating the coefficient of variation of the basal path lengths for each cell, and the results, pooled by cell class, are shown in Table 1. The mean coefficients of variation were generally low. The variation was highest among slender L5 cells, some of which had one or more basal path lengths substantially longer than the majority. Segment lengths and diameters. The proximity of most dendritic branch points to the soma suggests that intermediate segments are generally shorter than terminal segments. Figure 5A shows the frequency distribution of basal segment lengths for a single neurone from layer 213. The

A.U. LARKMAN

310

C 2/ 3

do’

li

Fig. 2. Camera lucida reconstructions of HRP-injected pyramidal neurones, showing typical examples of cells from each of the 3 cell classes. Abbreviations: ba-basal dendrites, do-distal oblique dendrites, po-proximal oblique dendrites, ta-terminal arbor. Scale bars 100 pm. A. Layer 213 cell. B. Thick L5 cell, characterized by the thick apical dendrite, which branches repeatedly near the top of layer 2/3 to form a

terminal arbor in layer 1. C. Slender L5 cell. Cells of this type were distinguished from thick cells by the thinner apical trunk, which tapers to a fine diameter and terminates, without forming a terminal arbor, before reaching layer 1.Although not used as a diagnostic feature, the basal dendrites of slender L5 cells tended to be longer but less numerous than those of thick L5 cells.

distribution is clearly U-shaped, with most intermediate segments shorter than terminal ones. There is, however, a degree of overlap; a small number of intermediate segments are substantially longer than the majority. The same pattern was seen when the data for the basal segments from all the cells in the sample were pooled (Fig. 5B), in spite of the wide range of terminal segment lengths encountered. Within individual cells, the coefficients of variation for terminal segment lengths were higher than those for overall path lengths (Table 1).This suggests that dendritic paths including long intermediate segments tend to have relatively short terminal segments and vice versa, preserving a more constant total path length. Terminal and intermediate segment lengths are summarized in Table 2. The generally greater lengths of terminal as opposed to intermediate segments means that a high proportion of the combined dendritic length of the basal dendrites is contributed by terminal segments. Values close to 90% were obtained for all cell classes (Table 2). Figure 5C shows the frequency distribution of basal segment diameters for the same layer 213 cell whose segment lengths were shown in Figure 5A. The terminal segments occupy a narrow range of values at the low end of the distribution, whereas intermediate segments show a much wider range of diameters. Values for segment diameters, pooled by cell class, are summarized in Table 2. Figure 6A shows the relationship between segment lengths and diameters for the same cell as in Figure 5. There was a

positive correlation between length and diameters for terminal segments and a weak negative correlation for intermediate segments, which was not significant in some cases. This pattern was preserved when data from all the cells in a class were pooled (Fig. 6B for thick L5 cells). Those terminal segments that showed substantial taper (mainly from some slender L5 cells) were excluded from this analysis. The weak negative correlation seen for intermediate segments was largely due to a minority of relatively very long preterminal segments (arrowed in Fig. 6A), which were of low diameter. Among the larger diameter, more proximal intermediate segments there was little correlation between length and diameter. Daughter branch ratios. The daughter branch ratio (DBR) of a branch point, here defined as the ratio of the diameter of the larger daughter segment to the diameter of the smaller daughter segment, is a measure of the symmetry of dendritic branching; ratios close to unity imply equal or symmetrical branching. Measured DBRs showed a wide range, from 1to 5, but the majority were close to unity (Fig. 6C). A clear difference was found between branch points on preterminal segments, which had DBR values close t o unity, and those on other intermediate segments, which showed a much wider range. This reflects the restricted range of diameters shown by terminal segments compared with intermediate segments, although the values may be subject to considerable measurement error. Daughter

PYRAMIDAL NEURONE DENDRITIC BRANCHING PATTERNS

A

311 TABLE 1.

6

a

Basal stemsicell

Layer 213

Thick L5

Slender L5

4.3 t_ 0.8 n = 18 6.7 i- 3.0 n = 78 24 n = 467 142 + 30 n = 526 11.7 -c 3.4 n = 18 22.4 2 5.6 n = 18 180 (35382) n = 17 57 (5-244) n = 96 32.9 (3.3-97.1) n = 96

5.8 ? 1.8 n = 11 6.0 f 2.7 n = 64 26 n = 313 156 2 29 n = 381 14.9 f 1.5 n = 11 25.3 f 4.6 n=ll 698 (553-769) n = 11 111 (8446) n = 131 16.3 (1.1-63.8) n = 131

4.4 f 0.7 n = 10 5.9 2 2.9 n = 44 24 n = 208 185 ? 53 n = 258 19.7 4.8 n = 10 27.5 z 7.0 n = 10 634 (447-849) n = 10 105 (3-548) n = 89 16.0 (0.5-74.5) n = 89

12.0 i 2.0 n = 11 2.1 2 1.4 n = 131

8.9 z 4.9 n = 10 1.5 -+_ 1.0 n = 89

bz8 Basal tipsistem

Path length to basal branch points (pm; median) Path length to basal tips (pm)

1

C

d

r

4

9

Coefficient of variation of basal path lengths to tips (%)I Coefficientof variation of basal terminal segment lengths (70)' Length of apical trunk (bm; median, range in brackets)

Distance of apical trunk branch points from soma (pm; median, range in brackets) Distance of apical trunk branch points from soma (as %ofapical trunk length; median, range in brackets) Oblique stemslcell Oblique hpslstem

5.7

? 1.8 n = 17 1.9 2 1.2 n = 96

'Mean of individual cell values

1

2

3

4

5

6

7

8

9 10111213

Tips per Stem Fig. 3. A. Dendrogram of the 5 basal trees (a-e) of a layer 213 cell. The number of tips in each tree is indicated to the right. Note the wide variation between trees in the degree of branching, with tree (c) remaining unbranched. Segment lengths are drawn to scale, but line thickness is only an approximate guide to segment diameter. Scale bar 50 pm. B. Frequency histogram showing the distributionof the number of tips per stem for all the basal trees in the sample. Wide hatchinglayer 213 cells; stippling-slenderL5; close hatchingthick L5.

branch ratios for the three cell classes are summarized in Table 3 .

Apical dendrites A dendrogram of the apical dendritic system of a typical neurone from layer 213 is shown in Figure 7A. The dendritic branching pattern is clearly different from the basals, and the apical trunk, apical oblique branches, and the terminal arbor can all be distinguished clearly in this case. This distinction was clear for all thick L5 and most of the layer 213 cells. Some layer 213 cells situated in the upper part of the layer had apical trunks that bifurcated close to the soma, and the distinction between oblique and terminal arbor branches was not obvious. In these cases it was often helpful to examine the distribution of path lengths to all of the apical tips (Fig. 7B). The path lengths usually fell into two clear groups, the shorter of which represented the obliques and the longer the terminal arbor tips. In some cases, branches that were defined as obliques on this

criterion arose after the apical trunk had divided. For these cases the convention was adopted that all segments giving rise to oblique branches were considered as part of the apical trunk rather than the terminal arbor. One cell, located at the upper boundary of layer 213, had two apical dendrites and resembled a modified superficial pyramid (O'Leary, '41).For both dendrites, the path lengths showed a unimodal distribution, and no distinction could be made between oblique and terminal arbor tips. This cell was eliminated from all analyses involving such a distinction. Slender L5 cells had clearly distinguishable apical trunks and obliques but lacked an obvious terminal arbor. The apical trunks of these cells gradually tapered to a small diameter before terminating. Sholl plots of apical dendritic systems highlighted the differences from the basals and showed clear differences between the cell classes. In all cases there was an initial increase in the number of intersections as oblique dendrites arose from the trunk and branched (Fig. 8). In most layer 213 cells, the terminal arbor branching began before all the obliques had terminated (Fig. 8A). For thick L5 cells this was not usually the case, so there was a region where the number of intersections fell to unity, representing the apical trunk, before increasing again in the terminal arbor (Fig. 8B). Some slender L5 cells had oblique branches along much of the length of the apical trunk, but the number of intersections eventually fell to unity and remained so until termination, reflecting the absence of a terminal arbor (Fig. 8C).

The individual components of the apical dendritic system are considered separately in more detail. Apical trunk. In all of the cells in our sample, the apical trunk was oriented orthogonal to the pial surface. Some of the slender L5 cells had displaced somata (Miller, '88) in that the soma and in some cases the extreme proximal portion of the apical trunk were aligned along a slightly different axis. The lengths of the apical trunks for the three cell classes are summarized in Table 1. The branch points giving rise to apical oblique dendrites were not uniformly distributed along the apical trunk, but were more frequent close to the soma. This is perhaps most

A.U. LARKMAN

312 v)

C

0 .c 0

a,

30

za,

20 -

c -

15-

c

10 -

c

0 L

a,

5-

3

0

a E

z

A

-

25

’ ~ . ‘ “ ” ’ ’ . “ . ‘ “ ’ ~ 0

100

50

150

0

Y

200

0

-

0

N

0

0

*

m

m

0

a

0

h

w

0

u

0

l

0

o

0

T 7

-

7

Y

0

0

N -

m l

-

Segment Length (pm)

40 v)

800

c

0 .c 30 0

v) c

a,

za,

c

C -

5

v)

C

.-cd

’0 a,

600

kl al

20

10

400

0

&

I 0

Segment Diameter (prn)

o

E 200

0

100

200

300

Path Length from Soma (pm)

3

2

-

n

C

Segment Length (pm)

400

v)

c

C .-

0

a

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 N * a a 3 o N - t u J w o N * a a 3 o T - - - F . - N N N N N m

300

1 0 C

Sl m 200 c

0

120130140150160170180

Fig. 5. Frequency histograms showing the distributions of basal segment lengths and diameters. Intermediate segments shown hatched, terminal segments stippled. A. Basal segment lengths for a single layer 213 cell. Note the “U”-shape distribution, with terminal segments generally longer than intermediate segments. B. Basal segment lengths for cells of all classes pooled, with the same overall shape as in A. C.. Basal segment diameters for the same layer 213 cell as in A, showing the narrow range of diameters of terminal segments.

Path Length from Soma (pm)

z

a I00 E

TABLE 2. Segment Lengths and Diameters’

3

z

0

20 40 60 80 100120140160180200220240

Path Length from Soma (pm)

BASAL DENDRITES Intermediate segment length (median) Terminal segment length (mean i SD) Intermediate segment diameter

Fig. 4. A. “Sholl plot” showing the number of intersections made by the basal dendrites of the cell shown in Figure 3A with probe lines or surfaces at various path lengths from the soma. Note the “plateau” between 70 and 130 km. B. Sholl plots showing the median values of the number of intersections made by the basal dendrites for each cell class. Layer 2/3-solid circles; thick L5-hollow circles; slender L5-hollow squares. Note the greater peak number of intersections made by thick compared with slender L5 cells. C. Frequency histogram showing the distribution of the path lengths from the soma at which basal branch points occur, for all cells pooled. Note that the overwhelming majority of branch points occur within 40 +m from the soma. D. Frequency histogram showing the distribution of the path lengths from the soma at which basal dendrites terminate, for a single layer 2/3 cell.

Terminal segment diameter %total dendritic length as terminal segments OBLIQUE DENDRITES Intermediate segment length (median) Terminal segment length (mean i SD) Intermediate segment diameter Terminal segment diameter % total dendritic length as terminal segments

TERMINAL ARBOR DENDRITES Intermediate segment length ( m e l a n ) Terminal segment length (mean i SD)

clearly shown if the distance from soma to branch point is expressed as a percentage of the overall length of the trunk (Fig. 9A-C). Median distances to trunk branch points, expressed in both absolute and relative terms, are given in Table 1. It was found that 50% of the branch points

Intermediate segment diameter Terminal segment diameter % total dendritic length as terminal segments

Layer 213

Thick L5

Slender L5

10 n = 446 102 t 31 n = 523 1.3 i 0.5 n = 446 0.6 i 0.1 n = 523 88 2 3.4

11 n = 314 117 t 33 n = 388 1.5 ? 0.6 n = 314 0.8 f 0.2 n = 388 89 2 2.9

11 n = 215 143 f 48 n = 255 1.6 i 0.7 n = 215 0.7 2 0.1 n = 255 90 t 2.7

12 n = 92 94 i 30 n = 187 1.1 i 0.3 n = 92 0.6 i 0.1 n = 187 91 ? 5.4

9 n = 140 107 i 31 n = 270 1.3 i 0.4 n = 140 0.7 i 0.2 n = 270 92 t 1.8

15 n = 48 129 t 50 n = 135 1.1 i 0.4 n = 48 0.6 i 0.2 n = 135 93 i 7.5

35 n = 120 92 2 4 0 n = 155 0.8 i 0.3 n = 120 0.4 t 0.1 n = 155 75 i 11.5

39 n = 37 110 2 4 3 n = 43 1.0 2 0.5 n = 37 0.5 2 0.1 n = 43 69 i 10.5’

‘All lengths and &lameters given in pm. ‘Calculated for 5 cells with most complete terminal arbors only.

-

-

PYRAMIDAL NEURONE DENDRITIC BRANCHING PATTERNS

-

200

5 120-

150

5

100-

100

a,

80-

50

E

60-

140

313

A

6

v

0)

c

-I

W 0 X Z ". - "-

a,

& a,

03

40-

200 -0.0

,

9 ° " m m

/ .Ii ,

,

F

"

.

0.5

I

1.5

1.0

2.0

3.0

2.5

tA

Segment Diameter (prn) 500 v1

.E

400

0

a

300

F

200

r

2

m 100 0

.

.

.

N ? = ? " " y ? ~ m N ? ? ? * -

-

r

m

(\I

-

m

.

m

.

o

Daughter Branch Ratio Fig. 6. A. Relationship between segment length and diameter for basal segments of a single layer 2/3 cell. Hollow circles-terminal segments; solid circles-intermediate segments. Regression line fitted through terminal segments only, r2 = 0.524. Note two relatively long preterminal segments (arrowed). B. As A, but for segments from all thick L5 cells pooled, showing similar pattern to A. r2 = 0.216. C. Frequency histogram showing the distribution of daughter branch ratios for basal branch points from all cells. Preterminal segment branch points-hatched; other branch points-stippled. Preterminal segment branch points all had DBR values close to one, but other branch points showed a much wider range. TABLE 3. Daughter Branch Ratios' BASAL DENDRITES Preterminal segments Other segments

Layer 213

Thick L5

Slender L5

1.00 (1.0-1.33) n = 182

1.00 (1.00-1.33) n = 139 1.38 (1.00-3.63)

1.00 (1.00-2.00) n = 86 1.50 (1.00-5.00)

1.40 (1.0-3.29) n = 242

OBLIQUE DENDRITES Preterminal segments Other segments

1.00 (1.00-1.50) n = 62

1.40 (1.OO-2.10) n = 30

A P I C A L TRUNK All segments

TERMINAL ARBOR SEGMENTS Preterminal segments

174

n = 107

1.00 (1.00-2.30) n = 51 1.30 (1.00-2.30)

1.00 (1.00-1.30) n = 19 1.50 (1.00-2.20) n = 19

=

n = 51

2.00 (1.08-4.00) n = 93

2.50 (1.13-6.71) n = 132

1.00

1.00 (1.00-1.33)

(1.00-2.50)

n = 56 Other segments

n

1.33 (1.00-2.70) = 55

"

2.00 (1.00400) n = 88

-

n = 17 1.50 (1.00-2.00) n = 20

'Medm values quoted, m t h ranges in brackets

occurred in the proximal 30%of the trunk for layer 213 cells, and in the proximal 16%for both types of layer 5 cells. The daughter branch ratios for apical trunk branch points varied between 1and 7, with median values greater

0

0

0

0

0

0

-

T

0

0

0

0

0

0

C

Q

O

0

0

0

o l D m N m ~ T - - = t ~ o r Y w m N m Y

-

-

N

N

N

C

~

O

W

Path Length from Soma (pm) Fig. 7. A. Dendrogram of the apical dendrites of a layer 213 cell showing the division into apical trunk (tr), apical obliques (ob) and terminal arbor fta) components. Segment lengths are to scale, but line thickness gives only an approximate guide to segment diameters. Scale bar = 100 pm. B. Frequency histogram of the path lengths from soma to tip for the apical dendrites of a different cell, from the upper part of layer 213. The distinction between oblique and terminal arbor segments was not obvious from the camera lucida reconstruction of this cell, but the distribution of path lengths is clearly bi-modal. The dendrites comprising the group with the shorter path lengths were taken to be obliques and those of the longer group the terminal arbor.

than 2 (see Table 3). The distribution of these DBR values for all cells is shown in Figure 9D, which may be contrasted with the distribution for basal dendrites shown in Figure 6C. Apical oblique dendrites. The distribution of apical oblique segment lengths for the same layer 2/3 cell used for Figure 5A is shown in Figure 1OA. As was found for the basals, the terminal segments were generally much longer than the intermediate segments. Oblique segment diameters also showed similarities with the basals, with slender terminal segments showing a narrower range of diameters than the thicker intermediate segments. The relationship between oblique segment lengths and diameters was also strikingly similar to the basal case. There was a positive correlation between length and diameter for terminal segments and a weak negative correlation, not significant in some cases, for intermediate segments (Fig. 10B; compare with Fig. 7A). Not only was the pattern of branching similar

~

A.U. LARKMAN

314 Layer 213

0

ICIO

200

300

400

1 0 2 0 30 4 0 5 0 6 0 7 0 8 0 9 0 1 0 0

500

50

20

0

0

200

400

600

800

1

i

1 0 20 30 4 0 5 0 6 0 7 0 8 0 9 0 1 0 0

1000

Slender Layer 5

I 0

200

400

600

800

1 0 20 30 4 0 5 0 6 0 7 0 8 0 9 0 1 0 0

1000

% Distance Along Trunk

Path Length from Soma (pm) Fig. 8. Sholl plots showing the number of intersections made by apical dendrites with probes at various path lengths from the soma, for individual cells from each class. A. Layer 213 cell. B. Thick layer 5 cell. C. Slender layer 5 cell.

-

100 m ._

.S

80

0-

60

If o t

40

0

to that of the basals, but the absolute values of their segment dimensions were also very similar. Data for oblique segment lengths and diameters for the three cell classes are summarized in Table 2. In each case, the mean terminal segment lengths of the obliques were slightly shorter than the basals and usually of slightly lower diameter. This difference was thought to be at least partly caused by the presence of some distal obliques that were shorter and thinner than the basal or more proximal oblique dendrites. Daughter branch ratios for branch points within oblique trees were close to unity for branch points giving rise to two terminal segments, and ranged between 1and 2.3 €or other branch points (Table 3). This is again similar to the situation described for basal dendrites. Each oblique segment branching from the apical trunk may be regarded as the stem segment of an oblique dendritic tree. The number of tips in each of these trees was measured and was found to vary with distance along the apical trunk, proximal stems giving rise to more tips than distal ones (Fig. lOC). Even among the proximal oblique trees, the number of tips per stem segment was generally lower than found for the basal dendrites. The distal oblique stems gave rise to no more than two terminal segments and most were mbranched. Since the oblique stems themselves

2 m

I

1

20

fl -

9 T-

c

u ? cu

o

9 @-J

b

9 d-

m

~ m

c

D

~

(D

Daughter Branch Ratio Fig. 9. A-C. Frequency histograms showing the distribution of the path lengths from the soma to the apical trunk branch points, expressed as percentages of the total length of the trunk, for cells of each class pooled. A. Layer 213 cells. B. Thick layer 5 cells. C . Slender layer 5 cells. D. Frequency histogram of daughter branch ratios for apical trunk branch points, for all cells pooled. Note the wide range of values obtained and the different shape of the distribution compared with the basal branch points (Fig. 6C).

are more numerous proximally, this means that the oblique dendritic system is much more profuse close to the soma than farther along the trunk. Oblique stems per cell and tips per stem values for the three cell classes are given in Table 1. The proportion of the oblique combined dendritic length contributed by the terminal segments was even higher than

r

-

PYRAMIDAL NEURONE DENDRITIC BRANCHING PATTERNS

0

-

N

0

m

0

b

0

t

0

n

0

w

h

0

c

0

n

0

a

l

0

0

0

r

r

7

O

r

N

315

0

m

0

-

N

Segment Length

0

0

b

0

a

m

0

0

O l

0

N

-

T

b

-

T

0

0

a

-

0

m l

-

0

O

-

N

N

N

Segment Length (pm) 200

-5

E,

5

150

0

v

B

0

2000

0 0 0

.

0)

C

2

100

c.

C

Q,

El a,

50

-. I- :

v)

0 0.0

0.5

1 .o

1.5

2.0

2.5 0.0

Segment Diameter (prn) in, I "

c/ v)

20 40 60 80 % Distance Along Apical Trunk

1.o

1.5

2.0

Segment Diameter (pm)

I

0

0.5

.

!

Fig. 11. A. Frequency histogram showing the distribution of terminal arbor segment lengths for layer 213 cells. Intermediate segmentshatched, terminal segments-stippled. The distribution is unimodal with considerable overlap between intermediate and terminal segment lengths, in contrast to basal and oblique dendrites (Figs. 5A, 1OA). B. Relationship between segment lengths and diameters for terminal arbor segments of layer 2/3 cells. Hollow circles-terminal segments; solid squares-intermediate segments. Regression line fitted through terminal segments only; r2= 0.314.

100

Fig. 10. A. Frequency histogram showing the distribution of apical oblique segment lengths for a cell from layer 213. Intermediate segments shown hatched, terminal segments stippled. Note the similarity with Figure 5A, which shows the basal segment lengths for the same cell. B. Relationship between segment length and diameter for apical oblique segments from all thick layer 5 cells. Hollow circles-terminal segments; solid squares-intermediate segments. Regression line fitted through terminal segments only; rz = 0.251. C. Relationship between the number of tips arising from an oblique stem segment and the position of its origin along the apical trunk, expressed as a percentage of the total length of the trunk, for thick layer 5 cells. Distal oblique trees are less highly branched than proximal ones; r2 = 0.18.

for the basals (Table 2). This was because the relative lengths of terminal and intermediate segments were similar in the two cases, but the oblique trees were less highly branched and so contained relatively fewer intermediate segments. Terminal arbors. The apical terminal arbors were studied in rather less detail than the basal or oblique dendrites. Slender L5 cells did not have obvious terminal arbors, and for some of the thick L5 cells the terminal arbor had clearly been reduced by amputation during slice preparation (see Larkman and Mason, '90). The following analysis is therefore restricted to layer 213 cells and five of the thick L5 cells whose terminal arbors appeared reasonably complete. The distribution of terminal arbor segment lengths is shown for layer 213 cells in Figure 11A, and a similar

pattern was found for the thick L5 cells. In contrast to the basal and oblique dendrites (cf. Figs. 6, 10A), the distribution is unimodal rather than U-shaped. Although the terminal segments are generally longer than intermediate segments, there is a high degree of overlap. Both exhibit a very wide range of lengths compared with basal and oblique segments. The occurrence of relatively short terminal segments could possibly be the result of incomplete staining or amputation during slice preparation, but more significant is the high proportion of relatively long intermediate segments ( > 4 0 Fm), which are rare in basal or oblique trees. The median values for intermediate segment lengths were greater for terminal arbor than basal or oblique dendrites (Table 2). The terminal segments thus contribute a lower proportion of the total dendritic length than in basal or oblique trees (Table 2). The terminal segments of terminal arbors were generally of lower diameter than those of basal or oblique trees (Table 2 ) . The relationship between segment lengths and diameters was, however, very similar to the basal and oblique pattern, with a positive correlation for terminal segments and no significant correlation for intermediate segments (Fig. 11B). Terminal arbor daughter branch ratios were similar to those of basals and obliques in that branch points on preterminal segments had DBR's close to unity, whereas other branch points were more variable (Table 3). Overall features. Having examined the different components separately, the way in which they combine to make up the dendritic system of the cell as a whole is now

A.U. LARKMAN

316 40 v)

5 30 .-c u

$

20

L

al

2 -

10

n 0

100

200

300

400

500

0

200

400

600

800

1000

200

400

600

800

1000

50 v)

.c

40

30 v)

&

20

c

c

-

10 0

417

0

Path Length from Soma (pm)

basal and oblique dendrites. Only a relatively small number of oblique intersections occur beyond the extent of the basal dendrites. Inspection of the HRP-injected neurones indicates that the basal dendrites together with the proximal majority of the obliques and the proximal part of the apical trunk appear to sample a roughly spherical volume of cortex centred about the soma and having a radius of 150-250 pm, depending on cell class and individual cell size. The relatively small number of more distal obliques, the distal part of the apical trunk and the terminal arbor, if present, sample a different volume of cortex of more complex shape, often extending across several cortical layers. The quantitative importance of the “proximal” portion of the cell’s dendritic system was estimated by a variation of Sholl’s procedure. The dendrites of a given cell were artificially straightened and tested using concentric spherical probes. Instead of counting intersections with the probes, the combined dendritic shaft membrane area lying within each of the “shells” between adjacent probes was calculated. In Figure 12D these values are expressed as cumulative percentages of the cell’s total dendritic shaft area with increasing path length from the soma, for one example cell from each of the cell classes. It was found that the cumulative percentage increased steeply and approximately linearly for some 150-200 pm, then showed an abrupt decrease in slope. The inflection point coincided in each ease with the rapid decline in intersection number seen in conventional Sholl plots as the basal and most of the oblique dendrites terminate. This point can be considered to mark the transition from the proximal to distal portions of the overall dendritic system of the cell. The proximal portion was found to include 70-80% of the cell’s total dendritic shaft area. This proportion was not quantified more precisely because of the difficulty in determining the exact point of transition, but nevertheless it represented a substantial majority of the total shaft area for every neurone in the sample.

DISCUSSION

0

200

400

600

800

1000

Path Length from Soma (pm) Fig. 12. A-C. Sholl plots showing the number of intersections made by all dendrites with probes at various path lengths from the soma. Open squares-apical dendrites; solid circles-basal dendrites. A. Layer 213 cell. B. Thick layer 5 cell. C. Slender layer 5 cell. In each case there is considerable overlap in the path lengths of the basal and oblique dendrites. D. Relationship between the cumulative percentage of the total dendritic shaft membrane area and path length from the soma. Solid circles-layer 213 cell; open circles-thick layer 5 cell; open squaresslender layer 5 cell. In each case, 70% or more of the total shaft membrane area is situated within a path length of 200 wm from the soma.

considered. Figure 12A-C shows Sholl plots for the entire dendritic systems of example cells from each of the three cell classes, with the contributions from the basal and apical dendrites indicated separately. A striking feature in each case is the considerable overlap in the path lengths of the

Neocortical pyramidal cells show considerable variation in their soma-dendritic morphology, to the extent that the presence of an apical dendrite may be the only feature they all share (Feldman, ’84).In previous works (Larkman and Mason, ’90; Mason and Larkman, ’901, based on the same sample of neurones used here, we drew attention to the morphological and electrophysiological differences between the three major classes of pyramidal neurone we encountered. In this and the companion studies, the morphological features common to all the classes are considered, although some interesting differences are also discussed. This study may be regarded as an extension of the work of Sholl ( ’53), who pioneered many of the analytical strategies employed here. An important finding was that the terminal segments of the basal and oblique dendritic trees showed similar length and diameter distributions, confirming the findings of Hillman (’79). Terminal segments within a given cell showed relatively less variation in their lengths and diameters than intermediate segments. The terminal segments were generally much longer than the intermediate segments and accounted for approximately 90% of the total combined dendritic length of the basal and oblique trees for all of the cell classes. The difference was more marked than

PYRAMIDAL NEURONE DENDRITIC BRANCHING PATTERNS in previous Golgi studies (e.g., Jursaka, '82). The terminal segments were, without exception, densely covered with dendritic spines, whereas many of the intermediate segments were nonspiny or only sparsely spiny. Thus the overwhelming preponderance of spines, and hence excitatory synaptic inputs, to these trees were situated on terminal segments. The dominance of basal and oblique trees by terminal segments is associated with a dendritic branching pattern rather different from some other well-studied types of neurone, for example, retinal ganglion cells (Boycott and Wassle, '74) or spinal motoneurones (Cullheim et al., '87) in which terminal and intermediate segments are of more similar lengths. The pyramidal pattern is different to the situation in cerebellar Purkinje cells, where the terminal and preterminal segments tend to consist of short branchlets (Hollingworth and Berry, '75; Shelton, '85; Sadler and Berry, '89). Relay cells in the lateral geniculate nucleus, however, are more like visual cortical pyramidal cells in having relatively long terminal segments (Leuba and Garey, '84; Fritschey and Garey, '88). Uylings et al. ('86) provide a table of terminal and intermediate segment lengths for a range of neuronal types, based on Golgi studies. Among the terminal segments of basal and apical trees, segment lengths and diameters were generally positively correlated. Since the electrotonic length of a cylindrical dendrite of given physical length is an inverse function of its diameter (Rall, '771, such a correlation will mean that the electrotonic lengths of terminal segments will tend to vary less than their physical lengths. The present findings for segment lengths and diameters are consistent with those of Hillman 1'79, Fig. 5 ) . The values for intermediate segment lengths obtained in the present study at first sight appear shorter than those reported by Juraska ('82). However, in the present study, median values were quoted as opposed to means, and given the heavily skewed distribution of intermediate segment lengths, these are very different. Our mean values are, in fact very close to hers. We found that most basal and oblique terminal segments showed little taper along their length, again in agreement with Hillman ('79). Most of the exceptions were provided by slender L5 cells. Some of these cells had very long terminal segments that were of fairly constant diameter for perhaps half their length but then tapered considerably. For reasons of simplicity, we excluded these segments from any analysis involving segment diameters, but such tapering may be of functional significance and might repay more detailed study. For example, tapering segments might be electrotoncally longer than nontapering segments of the same physical length and hence might provide exceptions to the generalization given above that all the basal dendrites of a single cell are likely to be of similar electrotonic lengths. A small proportion of basal and oblique dendrites did not follow the normal pattern but showed an exceptionally long preterminal segment, which divided to form two short terminal segments. In these cases the long preterminal segment was in most respects similar to a normal terminal segment but underwent branching, the significance of which is unclear. The apical trunk and terminal arbor displayed a different branching pattern from the basals and obliques. Branch points along the apical trunk showed higher daughter branch ratios than found in the basal and oblique trees. This is in agreement with Hillman ('791, who stressed the importance of the DBR in determining the form of a

317

dendritic tree. The high DBR's at oblique origins are associated with maintaining the identity of the apical trunk through successive branchings. Within the terminal arbor, terminal segments were generally of smaller diameter than those of the basal and oblique trees of the same cell. The intermediate segments were longer relative to the terminal ones and were always covered in spines. These factors combined to produce a tree less dominated by its terminal segments, in terms of both combined dendritic length and numbers of synaptic inputs, than was the case for basal and oblique trees. The results of our concentric probe analysis of basal dendritic trees are rather different from those of previous authors. For all of our cells individually, and for each cell class when cells were pooled, the number of dendritic intersections rose to a rounded peak or plateau which extended between approximately 60 and 130 pm from the soma before declining abruptly. In some previous studies there was no plateau, but rather a more pointed peak that occurred closer to the soma and was followed by a progressive decline to zero. The peak values were also generally lower than in the present study. Median peak values for our sample ranged from 23 intersections for slender L5 cells to 36 intersections for thick L5 cells, which are in excess of the peak values of approximately 18 intersections obtained in the early studies of Sholl ('53) and Eayrs and Goodhead ('591, or more recent studies of the rat auditory cortex (Vaughn, '77) and the rat occipital cortex (Davies and Katz, '83; Green et al., '83). Perhaps most striking is the difference between our mean peak value of 36 intersections for thick L5 cells and the peak value of 9 recently obtained for large layer 5 pyramidal cells of rat sensorimotor cortex (Petit et al., '88). Several factors may contribute to such differences. Different variants of Sholl's approach have been used in the various studies. We, like Sholl, artificially straightened the dendrites before analysis by using path length values, whereas others have applied concentric probes to camera lucida reconstructions or directly to cells visualized under the microscope (for a discussion of the relative merits see Berry et al., '72). We attempted to correct for two-dimensional projection errors, and many previous authors have not. However, perhaps the most important factor is the thickness of the section or slab of cortex containing the neurone under study. We reconstructed neurones from several 60-pm sections cut from the nominally 400-pm thick brain slices. Thinner slabs have generally been used in Golgi studies-200 pm (Eayrs and Goodhead, '591, 100-200 pm (Sholl, '53), 125 ym (Vaughn, '771, 100 pm (Davies and Katz, '83) or as thin as 90 pm (Petit et al., '88) and cells are not usually reconstructed across slabs. Interestingly, the above ranking by thickness of slab approximately mirrors the similarity of the results obtained to our own. The differences in methodology reflect the different aims of the various studies. We were concerned to achieve quantitative descriptions of dendritic systems which were as complete as possible. The other studies mentioned, with the possible exception of Sholl ('531, were concerned with relative measures for comparisons during development or to assess the effects of environmental factors. The underestimation of dendritic extent resulting from the use of thin tissue slabs will not affect the different components of individual dendritic trees or the cell's overall dendritic system equally. The importance of terminal segments relative to intermediate segments will be underestimated. Radially oriented trees, such

A.U. LARKMAN

318

as basals, obliques, and the distal parts of the terminal arbor, will be underestimated, whereas the apical trunk, which lies in the plane of section, will not. This may have contributed to the view that the basal and oblique dendrites are small compared to the apical trunk (Marin-Padilla and Stibitz, ’68), which is not consistent with present findings. The present results will not, of course, be free from errors. We did not correct for tissue shrinkage during histological processing, nor for any possible dimensional changes that might have occurred during the period of maintenance in vitro. Additionally, the “wiggle factor,” caused by small undulations, orthogonal to the plane of section, in the course of dendrites, can lead to a substantial underestimation of dendritic lengths (Desmond and Levy, ’821, for which we have not corrected. Nevertheless, we are confident that the HRP injection procedure can yield a more complete picture of the dendritic system than is readily obtainable using Golgi-based methods unless the difficult task of reconstruction across adjacent slabs is performed. This study has highlighted the similarity in branching pattern between the apical oblique dendritic trees and those of the basals. It was also found that most oblique segments arose from the apical trunk relatively close to the soma. We propose that it may be useful, at least for some purposes, to consider the dendritic systems of pyramidal neurones in two parts. The “proximal” part consists of the basal dendrites, the proximal obliques and the proximal part of the apical trunk, and samples a roughly spherical cortical volume centred about the soma. The “distal” part consists of the remainder of the apical trunk, the small number of distal obliques, and the terminal arbor if present. This displays a different branching pattern and samples a more complex volume of cortex. We found that the proximal portion was considerably larger than the distal in terms of dendritic shaft membrane area, although the proportion varied between cell classes. This will have implications for the distribution of synaptic inputs to the cell, which are considered in more detail in the third study of this series (Larkman, ’91).

ACKNOWLEDGMENTS The author thanks Mr. K.J. Stratford for writing computer software for morphometric analysis, Mrs. L.J. Annetts for photographic assistance, and Dr. J.J.B. Jack for critical reading of the manuscript. A.U.L. was a Beit Memorial Research Fellow; additional support was provided by the Wellcome Trust and MRC Programme Grant No. PG7900491.

LITERATURE CITED Berry, M., T. Hollingsworth, R.M. Flinn, and E.M. Anderson (1972) Dendritic field analysis-a reappraisal. T-I-T-J. Life Sci. 2: 129-140. Boycott, B.B., and H. W&sle (1974) The morphological types of ganglion cells of the domestic cat’s retina. J. Physiol. 240:397419. Burkhalter, A. (1989) Intrinsic connections of rat primary visual cortex: laminar organization of axonal projections. J. Comp. Neurol. 279: 171186. Cullheim, S.,J.W. Fleshman, L.L. Glenn, and R.E. Burke (1987) Membrane area and dendritic structure in type-identified triceps surae alpha motoneurones. J . Comp. Neurol. 255:68-81. Davies, C.A., and H.B. Katz (1983) The comparative effects of early-life undernutrition and subsequent differential environments on the dendritic branching of pyramidal cells in rat visual cortex. J. Comp. Neurol. 218:345-350.

Desmond, N.L., and W.B. Levy (1982) Aquantitative anatomical study ofthe granule cell dendritic fields of the rat dentate gyrus using a novel probabilistic method. J. Comp. Neurol. 212:131-145. Eayrs, J.T., and B. Goodhead (1959) Postnatal development of the cerebral cortex in the rat. J. Anat. 93:385-402. Feldman, M.L. (1984) Morphology of the neocortical pyramidal neuron. In A. Peters and E.G. Jones (eds): Cerebral Cortex, Vol. 1: Cellular Components of the Cerebral Cortex. New York: Plenum Press, pp. 123-200. Fritschey, J.M., and L.J. Garey (1988) Postnatal development of dendrites of relay neurons in the lateral geniculate nucleus of the marmoset (Callithrix jacchus); A quantitative Golgi study. J. Comp. Neurol. 268:234247. Gilbert, C.D. (1983) Microcircuitry of the visual cortex. Ann. Rev. Neurosci. 6217-247. Green, E.J., W.T. Greenough, and B.E. Schlumpf (1983) Effects of complex or isolated environments on cortical dendrites of middle-aged rats. Brain Res. 264.233-240. Greenough, W.T., and F.-L.F. Chang (1988) Plasticity of synapse structure and pattern in the cerebral cortex. In A. Peters and E.G. Jones (eds): Cerebral Cortex, Vol. 7: Development and Maturation of Cerebral Cortex. New York: Plenum Press, pp. 3 9 1 4 4 0 . Hillman, D.E. (1979) Neuronal shape parameters and substructures as a basis of neuronal form. In F.O. Schmitt and F.G. Worden (eds): The Neurosciences: Fourth Study Program. Cambridge: MIT Press, pp. 477498. Hofman, M.A., A.C. Laan, and H.B.M. Uylings (1986) Bivariate linear models in neurobiology: problems of concept and methodology. J. Neurosci. Meth. 18:103-114. Hollingworth, T., and M. Berry (1975) Network analysis of dendritic fields of pyramidal cells in neocortex and Purkinje cells in the cerebellum of the rat. Phil. Trans. Roy. SOC.B. 270:227-264. Juraska, J.M. (1982) The development of pyramidal neurons after eye opening in the visual cortex of hooded rats: a quantitative study. J. Comp. Neurol. 212:208-213. Larkman, A.U. (1991) Dendritic morphology of pyramidal neurones of the visual cortex of the rat. 111. Spine distributions. J. Comp. Neurol. 306:332-343. Larkman, A.U., and A. Mason (1990) Correlations between morphology and electrophysiology of pyramidal neurones in slices of rat visual cortex. I. Establishment of cell classes. J. Neurosci. 1011407-1414. Larkman, A.U., A. Mason, and C. Blakemore (1988) The in vitro slice preparation for combined morphological and electrophysiological studies of rat visual cortex. Neurosci. Res. 6:l-19. Leuba, G., and L.J. Garey (1984) Development of dendritic patterns in the lateral geniculate nucleus of monkey: a quantitative Golgi study. Dev. Brain Res. 16:285-299. Marin-Padilla, M., and G.R. Stihitz (1968) Distribution of the apical dendritic spines of the layer V pyramidal cells of the hamster neocortex. Brain Res. 11:580-592. Mason, A., and A.U. Larkman (1990) Correlations between morphology and electrophysiology of pyramidal neurones in slices of rat visual cortex. 11. Electrophysiology. J. Neurosci. 10:1415-1428. Mason, A,, A. Larkman, and J.L. Eldridge (1988) A method for intracellular injection of horseradish peroxidase by pressure. J. Neurosci. Meth. 22181-187. Miller, M.W. (1988) Maturation of rat visual cortex: IV. The generation, migration, morphogenesis, and connectivity of atypically oriented pyramidal neurons. J. Comp. Neurol. 274:387-405. O’Leary, J.L. (1941) Structure of the area striata of the cat. J. Comp. Neurol. 75131-164. Percheron, G. (1979) Quantitative analysis of dendritic branching. I. Simple formulae for the quantitative analysis of dendritic branching. Neurosci. Lett. 14287-293. Peters, A,, and E.G. Jones (1984) Classification of cortical neurons. In A. Peters and E.G. Jones (eds): Cerebral Cortex, Vol. 1: Cellular Components of the Cerebral Cortex. New York: Plenum Press, pp. 107-121.. Peters, A., and D.A. Kara (1985) The neuronal composition of area 17 of rat visual cortex. I. The pyramidal cells. J. Comp. Neurol. 2341218-241. Petit, T.L., J.C. LeBoutillier, A. Gregorio, and H. Libstug (1988) The pattern of dendritic development in the cerebral cortex of the rat. Dev. Brain Res. 41:209-219. Rall, W. (1977) Core conductor theory and cable properties of neurons. In Handbook of Physiology. The Nervous System-Cellular Biology of

PYRAMIDAL NEURONE DENDRITE BRANCHING PATTERNS Neurons, Section 1,Vol. 1. Baltimore: American Physiological Society, pp. 39-97. Sadler, M., and M. Berry (1989) Topological link-vertex analysis of the growth of Purkinje cell dendritic trees in normal, reeler, and weaver mice. J. Comp. Neurol. 289.960-283. Shelton, D.P. (1985) Membrane resistivity estimated for the Purkinje neuron by means of a passive computer model. Neuroscience 14:111131. Sholl, D.A. (1953) Dendritic organization in the neurons of the visual and motor cortices of the cat. J. Anat. 87:387406. Snow, P.J., P.K. Rose, and A.G. Brown (1976) Tracing axons and axon collaterals of spinal neurons using intracellular injection of horseradish peroxidase. Science 191:3 12-313.

319

Uylings, H.B.M., A. Ruiz-Marcos, and J. van Pelt (1986) The metric analysis of three-dimensional dendritic tree patterns: A methodological review. J. Neurosci. Meth. 18:127-151. Uylings, H.B.M., K. Kuypers, M.C. Diamond, and W.A.M. Veltman (1978) Effects of differential environment on plasticity of dendrites of cortical pyramidal neurons in adult rats. Exp. Neurol. 62:658-677. Vaughn, D.A. (1977) Age-related deterioration of pyramidal cell basal dendrites in rat auditory cortex. J. Comp. Neurol. 171:501-516. Zilles, K., A. Wree, A. Schleicher, and I. Divac (1984) The monocular and binocular subfield of the rat's primary visual cortex: A quantitative morphological approach. J. Cornp. Neurol. 226:391-402.