THE JOURNAL OF COMPARATIVE NEUROLOGY 309~141-149 (1991)

Postnatal Development of Thalamic Recipient Neurons in the Monkey Striate Cortex: 111. Somatic Inhibitory Synapse Acquisition by Spiny Stellate Neurons of

Layer 4C

JENNIFER S. LUND AND THOMAS R. HARPER Department of Psychiatry and Center for Neuroscience, University of Pittsburgh,

Pittsburgh, Pennsylvania 15261

ABSTRACT The development of type 2 (Colonnier, '81) synapses on the cell bodies of thalamic recipient

spiny stellate neurons in layers 4C alpha and 4C beta of primary visual cortex neurons was examined over the first 36 postnatal weeks and in the adult monkey. The type 2 synapses, known to be GABAergic (Ribak, '78) and therefore presumed to be inhibitory, developed faster on the alpha neurons than the beta neurons. Both neuron groups show a marked increase and then decline in the percentage of the somatic membrane covered by type 2 synaptic appositions during this 36-week time period. The time course of the type 2 synapses development is compared to that of the spine synapse development described in previous studies (Lund and Holbach, '91; Lund et al., '911, and it is clear that on both neuron groups this inhibitory synapse population is put in place and refined later than the spine synapses. These findings suggest that each cortical neural circuit has a unique time course for its early development within an overall time window (Rakic et al., '861, or sensitive period (Hubel and Wiesel, '70). Visual deprivation, although causing the alpha and beta neurons to adopt a more similar temporal and numerical developmental pattern than normal, did not prevent acquisition and loss phases of type 2 synapses or the assumption of a normal numerical loading by 36 weeks of age.

Key words: vision, cerebral cortex, sensory deprivation inhibition, electronmicroscopy

The companion studies by Lund and Holbach ('91) and Lund et al. ('91) define two phases of dendritic spine formation and refinement in the early postnatal period on thalamic recipient neurons of the monkey primary visual cortex. The spines are known to be the sites of type 1 synapses (Colonnier, 'Sl), which are presumed to be excita- tory. The two groups of cells examined in these studies were spiny stellate neurons lying in layer 4C alpha (postsynaptic to relays from the magnocellular portion of the lateral geniculate nucleus-dLGN) and layer 4C beta (postsynap- tic to relays from the parvicellular dLGN). The alpha neurons were found to precede the beta neurons in develop- ment of spines and dendritic arbor growth during the first 30 postnatal weeks. By 30 weeks of age, the alpha and beta neurons become synchronized in their development and acquire equal numerical loading of spines and, therefore, perhaps of excitatory spine synapses. These alpha and beta neurons then move into a second period of spine acquisition and dendritic growth now maintaining equal synapse num- bers. Change in afferent driving, induced by rearing under a

variety of conditions of visual deprivation, was found to modify the progression of these phases of growth but not to markedly change the final spine loading or alter the 30- week time window available for accomplishing the first phase of spine addition and refinement.

The present study examines the normal development of type 2 synapses (usually GABAergic-Ribak, '78- and, therefore, presumed to be inhibitory) on the somata of these same two neuron groups over the same developmen- tal period (birth to 36 weeks) and in the adult in many of the same animals used for the spine development studies. Four visually deprived (by bilateral eye lid suture or dark rear- ing) animals are also examined.

In an earlier attempt to examine this same population of synapses (Mates and Lund, '831, we concluded that type 2 synaptic contacts (containing flattened or pleomorphic vesi-

Accepted April 1, 1991. Address reprint requests to Jennifer S. Lund, University of Pittsburgh,

336C Scaife Hall, Pittsburgh, PA 15261.

0 1991 WILEY-LISS, INC.

142 J.S. LUND AND T.R. HARPER

Fig. 1. Illustration of a sample of layer 4C somatic synaptic contacts from animal K9 (12-week-old normal). (a-f) Type 2 contacts (symmet- ric contact sites with very little postsynaptic density plus an aggrega- tion of pleomorphic vesicles); small arrowheads indicate ends of the length of contact site typically measured; p (in (a)) indicates a region suspected of being a puncta adherens, such contacts were not included

in the counts of synapses. (g-h) asterisks indicate type 1 synapses (slight asymmetry to contact site with round synaptic vesicles); these cells and their contacts would be rejected from the sample since spiny stellate neurons are believed not to receive type 1 synaptic contacts on their somata; arrow indicates type 2 contact in (g). Magnification for (a-h) ~27 ,500 .

DEVELOPMENT OF INHIBITORY SYNAPSES IN STRIATE CORTEX 143

TABLE 1. Normal and Visually DeDrived Animals Used for This Studv. ~_____ ~~

No. of cells Animal no. Species Age (wks.) Condition Alpha Beta Region

1199 M. nemestrina 3 Normal 99 (0) 121 (1) Perimac 1188 M. nemestrina 5 Normal 97 (0) 104 (0) Perimac 1192 M. nemestrina 8 Normal 80 (0) 103 (0) Perimac T78303 M. nemestrina 8 Bilateral Suture 91 (1) 102 (2) Perimac K9 M. nemestrina 12 Normal 103 (41 106 (1) Perimac M1 M. nemestrina 12 Dark Reared 56 (0) 77 (0) Perimac

1274 P2

M. nemestrina M. nemestrina

24 24

Normal Dark Reared

56 (0) 58 (1) CalC 100 (1) 125 ( 2 ) Calc 113 (0) 136 (1) Mac

1316 M. fascicularis 24 Normal 93 (0) 104 (2) Perimac 1299 M. fascieularis 24 Bilateral Suture 85 (0) 95 (3) Perimac R1 M. nemestrina YO Normal 108 ( 2 ) 124 (1) Macular MI3025 M. nemestrina 36 Normal 89 (11 91 (1) Perimac 1204 M. nemestrina 5 years Normal adult 84 (1) 100 (0) Perimac

Animals 1316 (normal, 24 weeks old) and 1299 (bilaterallid suture from day 14 to 24 weeks) wereMacaca fascicularis; all the rest wereMacma nemestrina monkeys. Region sampled: Mac = macular representation; P. Mac = perimacular representation; Calc. = peripheral field representation (cortex from root of calcarine fissure). After the number of cells sampled, the number of cells encountered but not included in the sample is shown in brackets. Cells were eliminated when they had both type 1 and type 2 synapses on their surface, or had four or more type 2 contacts; see Methods), AU Macaca nemestrina monkeys except the adult (1204) were used for the study of spine and dendrite maturation in the previous studies (Lund andHolhach, '91; Lund et al., ,911,

cles and making symmetric apposition sites-Gray, '59; Colonnier, '68, '81) go through an increase followed by a decrease in number on the somata of the spiny stellate neurons of layer 4C during the early postnatal weeks. However, although the data of our earlier study showed a peak in percentage of somatic membrane covered by synap- tic apposition sites during the early weeks of postnatal development of the beta neurons, it did not indicate any clearly defined peak in development of these synapses on neurons in lamina 4C alpha. We have undertaken a reinves- tigation of these somatic synapse populations over age since we were particularly interested to see if the development of inhibitory contacts would relate in any way to the periods of excitatory synapse formation that we have defined more recently. As can be seen, the alpha and beta neurons are both found to undergo an increase, then decrease in percentage coverage of their somatic membrane by type 2 contacts during early postnatal development, but the t,ime course differs for alpha and beta neurons and is not synchronous with spine development.

MATERIALS AND METHODS The 13 animals used for this investigation of type 2

synaptic contacts are listed in Table 1. All but two of the infants were Macaca nemestrina monkeys, which had been used for spine counts and dendritic arbor measures in the preceding studies. Two M. fascicularis monkeys were in- cluded to allow examination of the effects of visual depriva- tion by bilateral lid suture taken to 6 months. Conditions of rearing are outlined in the preceding study (Lund et al., '91).

Tissue blocks from the primary visual cortex were taken after the animals had been anaesthetized with Nembutal and perfused with 4% paraformaldelyde (sometimes with the addition of 0.5% glutaraldehyde) in 0.1 M phosphate buffer (Millonig) with 0.54% dextrose (pH 7.4). The blocks were allowed to remain in the fixative for a period of at least 24 hours. Blocks of visual cortex approximately 1-2 mm were placed in 2% osmium tetroxide in 0.1 M phosphate buffer for 1-1.5 hours. Following a brief rinse in phosphate buffer, the tissue blocks were dehydrated in an ascending alcohol series to 70%. In 70% alcohol the blocks were further trimmed by hand, dehydrated completely in alcohol or acetone, followed by propylene oxide, and infiltrated with Epon or Polybed. Sections 1 pm thick were cut from the pia

to white matter extent of the cortical block; then, using a light microscope drawing tube, the pia to white matter section was drawn, giving the laminar boundaries of layers 4C alpha and 4C beta relative to the position of the pattern of blood vessels. The blocks were then trimmed to include layers 4C alpha and 4C beta plus part of adjacent layers 5 and 4B and thin sections cut for electronmicroscopy. The EM block was sectioned once at 1 km following thin sectioning to check and remap the laminar boundaries.

Thin sections were obtained at four intervals spaced at 50 pm through one or more blocks per animal. Comparison was made for two animals (K9 and M1) of sections taken from macular, perimacular, and peripheral field representa- tions; since no consistent difference could be found between these regions in single animals, the comparisons made for this study were taken from sections of the perimacular region from each animal as far as possible (see Table 1). In some cases, however, only macular or calcarine blocks were available. The perimacular cortex was the region used for the Golgi studies in the preceding studies.

Sets of measurements were made directly at the electron- microscope using an eyepiece graticule. The graticule was calibrated each time that measurements were made against a standard carbon replica calibration grid. The laminar boundaries of 4C alpha and beta were recognized by comparison with the detailed light microscope drawings of the 1-pm-thick sections made immediately before and following each thin-section set. Each cell body position was marked on the map relative to patterns of blood vessels and other landmarks so that no cell body was measured twice and the neurons could be assigned correctly to alpha or beta divisions. Measurements were made of two axes at 90" to one another across each neuron cell body encountered in layer 4C alpha and layer 4C beta. These measures were used to estimate average neuron diameter for each layer. The length of the apposition sites of all type 2 contacts onto each cell body were measured (see Fig. 1) and the total number of cells and the total number of type 2 contacts on their surfaces were noted for each of the samples taken.

It is known that GABAergic cells comprise approximately 12-15% of neurons in the alpha and beta division of lamina 4C (Fitzpatrick et al., '87) and the spiny stellate neurons appear to comprise the great majority of the rest of the neurons of the layer (Lund, '73, '84). Whereas the spiny stellate neurons appear to have solely type 2 contacts on their somata (LeVay, '73; Mates and Lund, '83; Saint Marie

144

0 48 - 046-

0 4 4 -

0 42 - 0 4 0 -

0 38 - 1 0 3 6 -

; 0 3 4 -

: 0 3 2 -

030-

j 0 2 8 -

! 0 2 6 - B . F 0 2 4 -

f 0 22 - 0 2 0 -

0 18-

0 1 8 -

0 1 4 -

0 12-

J.S. LUND AND T.R. HARPER

70 -

60 - 5

5 50 - 0

m -

P E 4 0 - v)

z 30 -

20 -

10 -

a. Normal Alpha and Beta Neurons. Cell Diameter

Cell diameter alpha normal

cell diPmeIer-oBIa norma! i i 1

6 . , . , . , . , . , . , . l . , . , . , . , . , _ , . , _ I _ , . , 0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0 2 2 2 4 2 6 2 8 3 0 3 2 3 4 3 5

Age (weeks)

I

A d d l

C .Normal Alpha and Beta Neurons. Number of SvnaDses/Cell Bodv

1 T

O L - n . , . , . , . , . , . , . ,., . ,. , . , . , . , . I . , 1 , I 0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0 2 2 2 4 i 6 2 8 3 0 3 2 3 4 3 6

Plge (weeks) Adult

b.Norrnal Alpha and Beta Neurons. Synaptic Length

0 50

I

7 010 . , . , . , . , . , . , . , . , . , . , . , . , . , . , . , . , . I , 0 2 4 6 8 1 0 1 2 ' 4 1 6 1 8 2 0 2 2 2 4 2 6 2 8 3 0 3 2 3 4 3 6

A 6 U l l Age (weeks)

d . N o r m a l Alpha and Beta Neurons. '70 of Cell Surface Covered bv TvDe 2 SvnaDses

I O -

09-

P

; 08- >

0 $ 0 7 -

> c 0 6 - n . ' 0 5 -

04-

0 3 -

0 2 -

U % C S Alpha ..*. % c s Bela

o t

Adult 0 2 4 8 8 1 0 1 2 1 4 1 6 1 8 2 0 2 2 2 4 2 6 2 8 3 0 3 2 3 4 3 6

Age (weeks)

Fig. 2. (a) Mean cell diameter of neurons sampled in layers 4C alpha and 4C beta over age. (b) Mean type 2 synaptic apposition length on neurons in layers 4C alpha and 4C beta over age. (c) Calculated mean number of synapses per neuron in layers 4C alpha and 4C beta over age. (d) Percentage of cell surface area covered by type 2 synaptic apposition sites for neurons of layers 4C alpha and 4C beta over age. Standard deviation is indicated for each sample in all graphs. ** P < .01.

and Peters, '85), some at least of the nonspinous, GABAer- gic neurons have both type 1 and type 2 contacts (Uchizono, '65; Colonnier, '68; Lund and Lund, '70; Levay, '73; Ribak, '78) on their surfaces. For this reason we discarded from our samples the few cells that were observed to have both type 1 and type 2 contacts (see Fig. lg-h). In addition, we discarded a very few cells that were observed to have four or more type 2 contacts since we came to suspect they were also indicative of cells that were heavily loaded with both

type 1 and type 2 synapses and therefore probably GABAer- gic. The encounter rate for such cells was very low (approx- imately 1% of the cells sampled). The numbers of cells discarded for any reason are shown in brackets in Table 1.

The rest of the cells encountered were used for our sample of neurons, and for the reasons outlined above they are believed to be largely spiny stellate cells with some small number (under 15%) of GABAergic or other types of nonspinous interneurons. The average circumference and

DEVELOPMENT OF INHIBITORY SYNAPSES IN STRIATE CORTEX

200 -

145

0

b. Normal Beta Neurons. a .Normai Alpha Neurons.

Dendritic Arbor Size and % Cell Surface Covered by Type 2 Synapses

Dendritic Arbor Size and % Cell Surface Covered by Type 2 Synapses

, 1 2 70 --O-- Inter no Alpha , 1 2

- 0 4 ' aJ 4 0 - be-: - 0 3 0

- 0 2

+ --*-

----a,

I lnler no Bela

, L o 1 30 . I . I . I , I . I , I , I . I . I , ' 0 1 0 4 8 1 2 1 6 2 0 24 2 8 3 2 3 6

Adull 0 4 8 1 2 1 6 2 0 2 4 28 3 2 3 6

Age (weeks) Adult Age [weeks)

Fig. 3. Comparison of development of dendritic arbor (measured by Sholl ring analysis of dendritic intersections; see Lund and Holbach, '91) and percent of somatic membrane occupied hy type 2 synapses on spiny stellate neurons in layer 4C alpha (a) and layer 4C beta (b).

a,Normal Alpha Neurons. Total Spine Coverage (Spines x Intersections) and % Cell Surface Covered by Type 2 Synapses

800 I

I U Sxl Alpha

t- % C S Alpha

0

1 2

1 1

1 0

0 9

0 8

b.Norrnal Beta Neurons. Total Spine Coverage (Spines x Intersections) and Ya Cell Surface Covered by Type 2 Synapses

800

700

600

500 0 7 2 z; 06 v) *

v 400

z 0 5

0 4 300

0 3

200

0 2

0 1 100

1 2 U Sxl Beta --*- % C S Beta

- 1 1

_. , , , . , , I , , , , , , , , , , i;r 4 8 1 2 1 6 20 2 4 2 8 3 2 36

Age (weeks) Adult

Fig. 4. Comparison of development of total observed spine coverage per neuron (calculated from observed spine density and dendritic intersection measures (S x I); see Lund and Holhach, '91) and percentage of somatic membrane occupied by type 2 synapses on spiny

stellate neurons of layers 4C alpha (a) and 4C beta (b). (Note: The observed values of spine density may be increased by a constant of + 75% of observed value to give a figure closer to actual spine density for both alpha and beta neurons at all ages; see Lund and Holbach, '91).

1 2

1 1 -

1 0 -

0 9 -

f n .

B 0 8 -

C z . v)

2 0 7 - TI

2 0 6 - 0

2 .

E - 2 0 5 -

0 . ' 0 4 -

-

0 3 -

U Alp9a ?o-maI

A Alpha b atera su'ure

0 A oha -3 , ru I tssc

A ApPa D la t t rd I %sc - -B- Alpia dark re ed

0

1 2

1 1 -

1 0 -

0 9 -

f n .

B 0 8 -

C z . v)

2 0 7 - TI

2 0 6 - 0

2 .

E - 2 0 5 -

0 . ' 0 4 -

-

U Alp9a ?o-maI

A Alpha b atera su'ure

0 A oha -3 , ru I tssc

A ApPa D la t t rd I %sc - -B- Alpia dark re ed

0

02-

0 1

* * T

'lo of Cell Surface Covered by Type 2 Synapses

, . , . , . , . , . I . , . , . , . , . , . ( . , . , . I , , . , . , . , 0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0 2 2 2 4 2 6 2 8 3 0 3 2 3 4 3 6

0 3 -

02-

0 1

1

, . , . , . , . , . I . , . , . , . , . , . ( . , . , . I , , . , . , . , 0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0 2 2 2 4 2 6 2 8 3 0 3 2 3 4 3 6

0 1 . , . l . I . l . , . , . , . , . / . , . 1 . / . / . , . , . $ . , . I

0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0 2 2 2 4 2 6 2 8 3 0 3 2 3 4 3 6 Age (weeks) Ad-11

Figure 5a, b

1 2

1 1 -

: 0-

--O-- Be'a i c m i ~ A Bela b,'a'e,a S J ' ~ W

Beta n o r m i , t i lsc

A Bela D 8citer,?l tCis:

- - - Be:a rea.rd

0 9 -

m . al

n m 0 8 -

z v ) .

z 0 7 -

U

0

8 0 6 -

m .

v) 0 5 -

L

5 - 0

bp 0 4 -

0 3 -

I

DEVELOPMENT OF INHIBITORY SYNAPSES IN STRIATE CORTEX

1 1 -

1 0 -

0 9 -

v ) .

v)

0

> v)

r n

2 0 8 -

P 0 7 - ? > 8 0 6 - m E . v, 0 5 - 0 - ;? '

0 4 -

03-

0 2 -

147

C.Normal and Deprived Alpha and Beta Neurons. Oh of Cell Surface Covered by Type 2 Synapses

0 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . , . , . , . I . I

0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0 2 2 2 4 2 6 2 8 3 0 3 2 3 4 3 6 Age (weeks) Adult

Fig. 5. Percentage of somatic membrane covered by type 2 synapses on spiny stellate neurons in normal and visually deprived monkeys. (a) Layer 4C alpha. (b) Layer 4C beta. The deprived animals include a bilateral lid-sutured Macaca nemestrina (sacrificed at 2 months of age A); two dark-reared M. nemestrina monkeys (sacrificed at 3 and 6

cell body surface area were calculated for each sample. The percentage of occupancy of single EM section cell profile membrane by synaptic apposition sites was calculated for each sample, and this was assumed to directly reflect the average percentage coverage of the cell body by synaptic appositions for that sample of neurons. Assuming each apposition to be a circular disc, the average number of synapses per neuron for each sample was calculated from this percentage occupancy of surface membrane, taking the average length of the apposition sites for each sample as the average diameter of the synaptic apposition. These figures may not be exact, but they do allow comparison over age of changes in a number of features bearing on the maturation of this particular inhibitory synapse population.

RESULTS In the collected data for cell diameter (Fig. 2a) and

synaptic apposition length (Fig. 2b), no particular trends were observed to occur over age and in general the alpha and beta neurons showed similar values at any particular age. The calculated average number of type 2 synapses per cell body over age (Fig. 2c) had some suggestion of different trends for alpha and beta neurons-the alpha neurons having somewhat more synapses than beta neurons in the

--Q- --+-

A A 0

A A 0

Alpha normal

Beta normal

Alpha-bilateral suture

Bela-bilateral suture

Alpha-normal lasc

Beta-normal Iasc

Alpha- bilateral lasc

Beta- bilateral f a x

Alpha-dark reared

Bela~dark reared

0

0

months m); and one M . fascicularis monkey reared with bilateral lid suture to 6 months of age A). The normal animals were 8 M. nemestrina monkeys 0, plus one normal 6 month old M. fuscicularis 0. In (c) the data of (a) and (b) is combined for comparison.

first 12 weeks postnatally but considerable variation was found between samples, making it uncertain if this data revealed any significant developmental pattern. However, when the average percentage of somatic membrane covered by synapses was calculated (Fig. 2d), the data showed more consistent trends over time during the first 36 weeks of life. The percentage of coverage rose sharply after birth on the alpha neurons, reaching a peak in our sample of different age monkeys around 12 weeks of age. This percentage of coverage fell markedly by 24 weeks and continued to decline to at least 36 weeks. The beta neurons were slower to initiate a rise in percentage coverage and only after 8 weeks of age was a rise in coverage found. The data suggest that a peak in coverage was achieved on the beta neurons some- time between 12 and 24 weeks with a decline between 24 and 36 weeks; the coverage on alpha and beta neurons had very similar values at 36 weeks of age.

We have plotted the percentage somatic coverage by type 2 synapses together with a measure of the total size of the average dendritic tree (see Lund and Holbach, '91) on these same alpha (Fig. 3a) and beta (Fig. 3b) neurons over age. It is clear from this comparison that the percentage of cover- age by inhibitory contacts on the soma does not directly reflect the overall size of the cell during early development.

148 J.S. LUND AND T.R. HARPER

A similar lack of correlation between synaptic coverage is found if only the size of the cell body (Fig. 2a) is considered.

The percentage of somatic coverage was also compared to the development of total observed spine populations on the same neurons (see Lund and Holbach, '91; note that a correction factor of + 75% of observed spine value is needed if actual spine densities are desired). Figure 4 illustrates this comparison separately for alpha and beta neurons. It can be seen that the peak in type 2 synapse coverage occurs later (probably 7-8 weeks later) than the first peak in total observed spine coverage on both alpha and beta neurons. The decline in type 2 synapse coverage continues to at least 36 weeks even as dendritic growth resumes around 24 weeks after birth (Fig. 3) and total observed spine synapse coverage again increases (Fig. 4). The somatic synapse coverage appears to increase again as the animal matures- even though spine populations (the site of type l-pre- sumed excitatory synapses) show a marked decrease be- tween 36 weeks of age and maturity.

The data from the four visually deprived animals showed no striking changes from the normal animals in synaptic length or cell diameter. A rise in percentage of coverage by type 2 synapses (Fig. 5)had evidently occurred on alpha and beta neurons in all animals despite visual deprivation. It is noticeable, however, that in the bilaterally deprived 8-week- old animal, the percentage of coverage on beta neurons is significantly higher (P < .01) than the normal, matching the alpha neuron percent coverage (Fig. 5C). This suggests that a faster than normal rise in synapse coverage had occurred on the beta neurons. The dark-reared animal at 12 weeks shows rather similar values for the alpha and beta neuron percent coverage, a figure between the normal alpha and beta neuron values, which were significantly different at the same age. At 24 weeks the dark-reared animal shows percentage coverage very close to the normal. Bilateral lid suture carried from 14 days to 24 weeks in aM. fascicularis monkey produced little change from alpha and beta neuron values compared to a normal animal of the same age and species (Fig. 5).

DISCUSSION A clear change in percentage of cell body membrane

covered by type 2 synaptic appositions occurs on both layer 4C alpha and beta neurons, 85% of which at least should be spiny stellate neurons. These neurons show an increase and decline in coverage over the first 36 postnatal weeks beginning and ending this period with very similar coverage but differing significantly from one another around 12 weeks of age when the alpha neuron coverage exceeds that of the beta neurons. The time course of the increase and decrease in coverage is not synchronous for the alpha and beta neurons, the alpha peak occurring earlier (by about 7-8 weeks) than the beta peak. This is reminiscent of the different rates of spine synapse development on the same two groups of neurons described in the first work of this series (Lund and Holbach, '91) and reinforces the sugges- tion that alpha neuron synaptic development precedes beta neuron development during the first 30 weeks of postnatal life.

It is evident from this study that the degree to which the surface of the cell body is covered in type 2 synapses during development is not a simple balance between excitatory load on the dendrites and inhibitory load on the soma. Nor does the somatic coverage relate in a simple fashion to size

of either the dendritic tree or soma. It is also apparent that there is not a simple, single, temporal framework for excitatory and inhibitory synapse development for neurons of the same class; on these neurons spine synapses can be increasing, whereas synapse coverage on the soma remains unchanged, or somatic type 2 synapse coverage may be declining while dendritic growth occurs and new spine synapses are being added. It does, however, appear that inhibition at the soma is built and refined in step with but somewhat delayed relative to the building and refining of dendritic spines and presumably their type 1 synapses.

Whereas normal visual input does not appear to be necessary for rise in coverage by type 2 synapses to occur, the temporal and numerical patterning of the type 2 synapse formation may be changed when visual deprivation is carried out. It is reminiscent of the spine development under conditions of bilateral lid suture (Lund et al, '91) that the beta neuron type 2 synapse coverage changes from normal and resembles the alpha neurons at 8 weeks, i.e., it seems that in both excitatory and inhibitory development that the beta neurons move to a more alphalike anatomical state under conditions of diffuse light stimulation. The animal dark-reared to 12 weeks showed a closer similarity of alpha and beta neuron type 2 synapse coverage than the normal animal, midway between the normally disparate values; this again resembles the changes observed in spine synapse coverage on the same neurons in the dark-reared animal. By 6 months, the somatic coverage in dark-reared and bilaterally lid sutured animals seems close to normal values. This may indicate that an innate regulation of these synapses occurs around 24-30 weeks of age (much as we observed for spine development) that is not dependent on visual driving. Reorganization of circuitry may well occur to sustain a normal level of activity in these synapses, now independent of visually driven input (Mower et al., '88).

As we concluded from the study of maturation of den- dritic spine populations, the temporal differences between the alpha and beta neurons in addition and refinement of somatic synaptic contacts during early postnatal life indi- cates that separate factors control time and rate of develop- ment of specific synapse populations in developing cortex. Whereas there may be a time window during which overall synaptic addition and attrition can occur (Rakic et al., '861, when the cortical neuropil is examined for specific neural circuits each element is found to have its individual, unique developmental time course. Moreover, the continuing de- cline of somatic inhibitory synapses up to at least 36 weeks suggests that the 30-week postnatal period that we outlined as a first phase of spine synapse development may not be a time period that is particularly meaningful for the develop- ment of somatic inhibition on the same neurons. It is possible that renewed addition of somatic inhibitory cover- age occurs in animals a little older than 36 weeks of age, i.e., showing a delay compared to the second period of addition of spine synapses, which starts around 24-30 weeks (much as occurred in the first phase prior to 30 weeks where spine synapse formation preceded type 2 synapse addition to the soma). It remains to be seen how development of somatic inhibition may interrelate to type 2 synapse formation on the dendritic shafts and spines. It is possible that total type 2 synapse coverage and total type 1 synapse coverage each assumes equal numerical loading on the alpha and beta neurons by maturity.

DEVELOPMENT OF INHIBITORY SYNAPSES IN STRIATE CORTEX 149

ACKNOWLEDGMENTS Support for this work was provided by the National Eye

Institute (Grants EY01086 and EY05282). We thank Drs. Jonathan Levitt, David Lewis, and Ray Lund for their helpful comments.

LITERATURE CITED Colonnier, M. (1968) Synaptic patterns on different cell types in the different

laminae of the cat visual cortex. An electron microscopy study. Brain Res. 9268-278.

Colonnier, M. (1981) The electron-microscopic analysis of the neuronal organization of the cerebral cortex. In F.O. Schmitt, F.G. Worden, G. Adelman, and S.G. Dennis (eds): The Organization of the Cerebral Cortex: Proceedings of a Neurosciences Research Program Colloquium. Cambridge: MIT Press, pp. 125-152.

Fitzpatrick, D., J.S. Lund, D.E. Schmeckel, and A.C. Towles (1987) Distribu- tion of GABA-ergic neurons and axon terminals in macaque striate cortex. 3. Comp. Neurol. 264~73-91.

Gray, E.G. (1959) Axosomatic and axodendritic synapses of the cerebral cortex. An electron microscopic study. J. h a t . 93:420-433.

Hubel, D.H., and T.N. Wiesel (1970) The period of susceptibility to the physiological effects of unilateral eye closure in kittens. J. Physiol. (Lond.) 206t419-436.

LeVay, S. (1973) Synaptic patterns in the visual cortex of the cat and monkey. Electron microscopy of Golgi preparations. J. Comp. Neurol. 150:53-86.

Lund, J.S. (1973) Organization of neurons in the visual cortex, area 17, of the monkey (Macacu Mulattu). J. Comp. Neurol. 147:455496.

Lund, J.S. (1984) Spiny stellate neurons. In E.G. Jones and A.A. Peters (eds): Cerebral Cortex, Vol. 1. New York Plenum Press, pp. 255-308.

Lund, J.S., and R.D. Lund (1970). The termination of callosal fibers in the paravisual cortex of the rat. Brain Res. 1 7 3 5 4 5 .

Lund, J.S., and S.M. Holbach (1991) Postnatal development of thalamic recipient neurons in monkey striate cortex: I. Comparison of spine acquisition and dendritic growth of layer 4C alpha and beta spiny stellate neurons. J. Comp. Neurol. 309t115-128.

Lund, J.S., S.M. Holbach, and W.-W. Chung (1991) Postnatal development of thalamic recipient neurons in monkey striate cortex: 11. Influence of afferent driving on spine acquisition and dendritic growth of layer 4C spiny stellate neurons. J. Comp. Neurol. 309:129-140.

Mates, S.L., and J.S. Lund (1983) Developmental changes in the relationship between type 2 synapses and spiny neurons in the monkey visual cortex. J. Comp. Neurol. 221~98-105.

Mower, G.D., R. Rustad, and W.F. White (1988) Quantitative comparisons of gamma-aminobutyric acid neurons and receptors in the visual cortex of normal and dark-reared cats. J. Comp Neurol. 272293-302.

Rakic, P., J.-P. Bourgeois, M.F. Eckenhoff, N. Zecevic, and P.S. Goldman- Rakic (1986) Concurrent overproduction of synapses in diverse regions of the primate cerebral cortex. Science 232232-235.

Rihak, C.E. (1978) Aspinous and sparsely spinous stellate neurons in the visual cortex of rats contain glutamic acid decarboxylase. J. Neurocytol. 7~461478.

Saint Marie, R.L., and A. Peters (1985) The morphology and synaptic connections of spiny stellate neurons in monkey visual cortex (area 17). A Golgi-electron microscopic study. J. Comp. Neurol. 233r213-235.

Uchizono, K. (1965) Characteristics of excitatory and inhibitory synapses in the central nervous system of the cat. Nature 207~642-643.