Freeze-fracture of developing neuronal plasma membrane in postnatal cerebellum

Brain Research, 208 (1981) 19-33 © Elsevier/North-Holland Biomedical Press

19

F R E E Z E - F R A C T U R E OF D E V E L O P I N G N E U R O N A L PLASMA M E M B R A N E IN P O S T N A T A L C E R E B E L L U M

LUIS MIGUEL GARCIA-SEGURA and ALAIN PERRELET Institute of Histology and Embryology, University of Geneva Medical School, 1211 Geneva 4 (Switzerland)

(Accepted August 14th, 1980) Key words: neuron -- development -- membrane -- cerebellum -- freeze-fracture

SUMMARY The structure of the plasma membrane of two neuronal cell types was quantitatively assessed in freeze-fracture replicas of rat cerebellar cortex from 2 to 73 postnatal days. The measurement of the number and size of intramembrane particles (IMP) in different regions of the Purkinje cell and of the granule cell showed that between 2 and 21 postnatal days there was a differential increase in the number of IMP; several regions of the cell characterized by a different pattern of synaptic interconnections showed differences in particle size and in particle number: neuronal processes had lower numbers of IMP than perikarya, whereas dendritic spines contained fewer IMP than dendritic trunks. These differences in IMP content appeared at the time at which specific synaptic contacts are established in these various regions of the cells.

INT RODUCTION One of the basic unsolved questions in neurobiology is how a neuron can recognize its specific target to make and maintain synaptic contacts. An increasing body of evidence suggests that the neuronal plasma membrane plays a central role in neuronal recognition and in the formation of synaptic junctionsS-7,0,12, is. In the central nervous system, a neuron receives several types of presynaptic axons that make synapses in specific regions of the neuronal body. If a recognition between pre- and postsynaptic plasma membranes takes place, differences in membrane composition and/or arrangement of different membrane components between these specific regions can be expected. Freeze-fracture is the method of choice for the study of macromolecular arrangement of membranes and application of this technique to the nervous system has revealed an aggregation or inhomogeneous distribution of intramembrane

20 particles at presynaptic and post-synaptic sites of mature synapses l°,1t,lz,14,15,25,a°. In developing neurons, it has been shown that the membrane of growing nerve fibers contains few particles; a gradient of density of particles then forms between the relatively richly particulate perikaryon and the less particulate growth cone, and the nerve fibers are progressively enriched in particles as a function of time ~7,19,28. The purpose of this investigation was to extend this structural approach by studying, with freeze-fracture, two precisely identified neuronal types in vivo during postnatal growth. It reveals that in the maturing Purkinje cell and granule cell of the rat cerebellum, different regions of these ceils show different patterns of intramembrane particle size and particle number and that these patterns change during postnatal life. MATERIALS AND METHODS

Animals, anesthesia and fixation Female rats (Sprague-Dawley) aged 2, 5, 10, 21 and 73 days post-partum were used. Three rats of each age were anesthetized with ether and perfused through the heart with 2 ~ glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) for 15 min following rinsing of the vascular system with oxygenated Ringer solution. Perfused cerebellum was removed and small blocks of 3 cortical regions (flocculus, hemispheres and vermis) were further fixed in 6 ~ glutaraldehyde in 0.1 M phosphate buffer, pH 7.4, for 2 h at 4 °C and then washed overnight in 0.2 M phosphate buffer with 6.8 sucrose at 4 °C.

Conventional electron microscopy Tissue blocks from each animal and each cerebellar region were dehydrated in ethanol and embedded in Epon 812. Semi-thin sections were stained with toluidine blue and thin sections with uranyl acetate and lead citrate.

Freeze-fracturing Pieces of fixed cerebellar cortex were soaked for 2 h in 0.1 M phosphate buffer, pH 7.4, containing 2 0 ~ glycerol, rapidly frozen in Freon 22, cooled with liquid nitrogen, fractured at --110 °C and replicated 15 in a Balzers BAF 301 apparatus (Balzers, Lichtenstein). Three to six freeze-fracture replicas were prepared from each cerebellar cortical region.

Quantitative evaluation of freeze-fracture replicas Freeze-fracture replicas were photographed at 55,000 × magnification (calibrated with a reference grid: Fullham 2130 lines/mm) in a Philips EM 300 electron microscope. Negatives were projected at a suitable magnification onto a graphic tablet (Tektronix, type 4953) connected to a microprocessor (IMSAI, type 8080) allowing the different parameters to be recorded and statistically analyzed. In each replica, the number of intramembrane particles (IMP) per square micrometer of membrane was assessed in the different neuronal regions, together with the size of individual particles. The size (diameter) of the particles was measured on the graphic tablet as the length of

21 a line drawn at the base of the triangular shadow projected by the particle, perpendicularly to the direction of the shadow. Statistical comparisons of mean values were carried out using Student's unpaired t-test. RESULTS

Identification of neuronal membranes The adult cerebellar cortex has two cell layers: an external or molecular layer and an internal granular layer (Figs. 1 and 2). A third layer, the external granular layer, is temporarily present at the cerebellar surface during development. In the rat, the internal granular layer becomes fully established only after birth1, 2,23 and the external granular layer can still be recognized at the cerebeUar surface during early days of postnatal development. Several types of neurons have been described in the cerebellar cortexl~,2L In this study, the quantitative assessment of freeze-fracture membrane organization was restricted to the perikaryon, dendrite and dendritic spine of the Purkinje cell and to the perikaryon and parallel fiber of the granule cell (Figs. 1-5). The majority of granule cells are formed between 8 and 15 days but from two days onwards, primitive granule cells can be detected in the internal granular layer s . Differences in size and cytoplasmic organization were used to distinguish Purkinje and granule cells from the other cells of cerebellar cortex z-4,16 (Figs. 3 and 4). Parallel fibers are ramifications of the axon of the granule cell which run horizontally in the molecular layer, perpendicular to the dendritic tree of the Purkinje cell. Parallel fibers establish synapses with the spine of the Purkinje cell dendrites. This characteristic geometry was present at day 2 and was used in freeze-fracture replicas to identify parallel fibers. Enlarged segments of the parallel fibers (cf. Fig. 5) were not evaluated since they are assumed to represent presynaptic active zones and have already been shown to present a specific pattern of IMP 3°. Dendrites and dendritic spines of the Purkinje cell show intensive growth from the tenth to the thirtieth postnatal days 8. Typical dendritic spines were not recognized before day 5. The few spines present on day 5 could, however, be distinguished from the collateral filopodia* by their characteristic drumstick appearance. Continuity of spines with a fractured dendrite profile was used as criterion for identification.

Particle numbers in Purkinje and granule cells Quantitative assessment of IMP did not reveal any difference in IMP size or IMP number between cells of the 3 cortical regions studied (flocculus, hemispheres and

* On the fifth postnatal day, the dendritic tree of Purkinje cellsdoes not show the characteristic aspect of later periods, and freeze-fractured dendrites have a large number of small processes that probably correspond to collateral filopodia described with the Golgi method in developing Purkinje cell dendritesa. These processes were observed from 2 to 10 days and were characterized by a low content (93 dr 12 IMP/sq./~m, P-face; 22 ± 5 IMP/sq./~m, E-face) of IMP with large mean diameters (11.4 ± 0.1 nm, P-face; 11.7 q- 0.3 nm, E-face). The role of collateral filopodia is unknown and it has been suggested that they represent immature spines, primordia of collateral dendritic branches and/or growth cones that are being reabsorbed8.

r-

~Q

~N

~T

J~ !I

F"

Fig. 5. Freeze-fracture replica of the molecular layer showing the P-face of a Purkinje cell dendrite (PCD) giving rise to several spines (PCS). The field of the replica is crossed by the exposed profiles of the parallel fibers (PF). In the upper part of the picture, two enlarged regions of the parallel fibers (*) not considered for the evaluation of particle number and size (see text). Sixty-four-day-old rat. × 32,700.

Fig. 2. Semi-thin, toluidine blue-stained section of the cerebellar cortex of a 36-day-old rat. The upper part of the picture represents the molecular layer (ML) containing Purkinje cell dendrites and parallel fibers; the lower part of the picture corresponds to the granular layer (GL) with the perikarya of the granular cell. The perikarya of Purkinje cells are situated at the limit between the molecular and granular layers, x 480. Fig. 3. Low power micrograph of a freeze-fracture replica showing the typical appearance of a granule cell perikaryon. Note the scarce cytoplasm (CYT) between the nucleus (N) and the plasma membrane (PM). Seventy-three-day-old rat. x 10,600. Fig. 4. Low power micrograph of a freeze-fracture replica with a part of a Purkinje cell perikaryon. At the same magnification as Fig. 3, note the large size of the Purkinj¢ cell cytoplasm (CYT) which contains numerous membrane profiles of cytoplasmic organdies. N, nucleus; PM, plasma membrane. Seventy-three-day-old rat. x 10,600.

i

i

25 PURKINJE CELL P - FACE IMP/.urn ;=` 1500 1400 1300 1200 I100 I000 900 800 700 600 500 400 300 200 I00 0

f

SPINE

I

I

i

i

5

I0

21

7 POSTNATAL

DAYS

Fig. 12. Mean numbers (± S.E.M.) of IMP per sq./~m of P-face in different membrane regions of the Purkinje cell, plotted against postnatal age (in days). Note the difference in IMP density between the 3 regions compared and the sharp increase in the density of particles at the 3 levels between 2 and 21 days. vermis). Thus, the results from the 3 regions were pooled in tables and figures. On the contrary, differences were found between several regions of the neuronal cell body. As shown in Figs. 6-13 and Tables I and II, the quantitative evaluation of I M P on these identified membrane faces revealed an increase in the number of P-face particles in both granule and Purkinje cells, with a characteristic pattern for each membrane region. At 5 days of postnatal development there were approximately 6 times more particles in the perikaryon of Purkinje cells than in spines (1028 4- 66 IMP/sq. #m, P-face; vs 173 4- 44 I M P / s q . / , m , P-face). This difference decreased steadily with increasing age and by day 21 and for the remainder of the experimental period, there were only twice as many particles on the perikaryon than in spines. At two days, the number of particles in the P-face of the Purkinje cell perikaryon was only 1.5 times higher than in the dendrite but by day 73, the two membranes contained similar numbers of I M P (Fig. 12 and Table I). On the second day of postnatal development, the number of I M P in the Pface of the granule cell perikaryon was approximately 7 times that in the parallel fibers (859 ± 48 IMP/sq./~m, vs 120 ~ 10 IMP/sq./~m). As in Purkinje cells, this difference

Figs. 6-11. Examples of fracture faces (such as those evaluated quantitatively) of different membrane regions of the two cell types studied at various postnatal ages. Fig. 6. Purkinje cell dendritic spine, age 5 days, P-face. Fig. 7. Purkinje cell dendritic spine, age 73 days, P-face. Fig. 8. Purkinje cell dendrite, age 2 days, P-face. Fig. 9. Purkinje cell dendrite, age 73 days, P-face. Fig. 10. Parallel fiber, age 10 days. Fig. 11. Parallel fiber, age 73 days. PF (P), P-face; PF (E), E-face. All figures × 69,100.

lMP/sq. I~m P-face Spine

947--42(n 17) * 617±21(n--66) -1028~66(n-49) ** 8 0 7 5 - 2 6 ( n 60) * 1 7 3 ± 4 4 ( n - 3 3 ) 1 1 2 6 ± 5 4 ( n - - 1 8 ) *** 9 1 1 ± 6 1 ( n : 10) * 4 8 1 ± 1 3 ( n 29) 1231±63(n 10) * 959±27(n 13) *** 6 7 8 ± 1 0 0 ( n 46) 1346±55(n 19) 1204±56(n 19) * 6 7 6 i 3 6 ( n 143)

Dendrite 131±06(n 145) 242±28(n-32) 233±25(n--25) 342±27(n 11) 3 9 7 ± 3 1 (n 6)

Perikaryon

lMP/sq, l¢m E-face

120±14(n 36) 174±09(n 23) 222±27(n--29) 306i13(n 35) 430i34(n 14)

Dendrite

*

*

* Significant differences P < 0.001.

* *

*

1 2 0 i I0 ( n - 8 6 ) 295±30(n 53) 451 ± 3 8 ( n 77) 656±44(n--30) 6 8 3 ± 2 8 (n 59)

* *

2 5 10 21 73

9 5 9 ± 4 8 (n 10) 1 2 8 7 ± 6 8 (n 36) 1364±92 (n-29) 1443±72(n 28) 1 4 4 1 i 8 8 (n 53)

Axon 154±15 232±19 249 ± 12 260±16 266±14

(n 48) (n 49) (n -- 54) (n 20) (n 13)

Perikaryon

lMP/sq. /era E-face

n u m b e r of m e m b r a n e s studied.

Postnatal IMP/sq. l~m P-face age in days Perikaryon

D a t a are m e a n ± S.E.M. ; n

IMP numerical density in granule cell plasma membranes

T A B L E lI

* * *

23±03 85=13 97±12 273--71 269±40

Axon (n~ 30) (n 56) (n 59) (n 16) (n--37)

* Significant differences P < 0.001. ** Significant differences P < 0.002. *** Significant differences P < 0.01. P c o m p a r e s the 2 values o n either side of the asterisk(s) in this a n d the following tables

2 5 10 21 73

age in days Perikaryon

Postnatal

D a t a are m e a n ± S.E.M. ; n = n u m b e r o f m e m b r a n e s studied.

IMP numerical density in Purkinje cell plasma membrane

TABLE I

6 8 ± 16 (n 143_+29 (n 254±46 (n-218±24(n

Spine

25) 22) 19) 122)

t~

27 GRANULE CELL P - FACE IMP 500

-

400

-

300

-

200

-

/

pm 2

PERIKARYON

I00 000 900 800 700 600 500

-

400

-

500

-

200

-

I00

/

AXON (PARALLEL

FIBERS)

-

0

!

!

2 5

I

I

I0

21

/

73 POSTNATAL

DAYS

Fig. 13. Mean number (± S.E.M.) of IMP per sq./zm of P-face in different regions of the granule cell, plotted against postnatal age (in days). Note the differencein particle density between the two regions compared and the sharp increase of IMP density occurring at both levels between 2 and 21 postnatal days. decreased during development, and from the twenty-first day on, the number of particles per granule cell perikaryon was only double that in the granule cell axon (Fig. 13 and Table II). As shown in Tables I and II, variations in the number of IMP in E-faces of the different membrane regions of the Purkinje and granule ceils were present but were less marked.

Size of particle on Purkinje and granule cell membranes One problem encountered during this evaluation was the high degree of 'granularity' observed in the lipid domain of many freeze-fractured neuronal membranes (Figs. 7-9, 11). This granularity, which may hamper the identification of the smallest IMP, was also apparent in previously published work on developing axons (Fig. 19 of ref. 14), as well as in other cell types, and its nature has been variously interpreted as water vapor contamination or as a specific membrane substructure 2 0 ,2 1 , 24 The mean IMP diameter values for each age and each plasma membrane region are shown in Tables III and IV, and significant differences of IMP diameter were found between the different plasma membrane regions, both in the E- and P-faces. In Purkinje cells, dendritic spines contained IMP of larger diameter than dendrites or perikarya. Also in granule cells, the mean diameter of IMP in the axon plasma membrane was larger than in the cell body. When the comparison between different cell regions was restricted to IMP with diameter larger than I 1 nm (class C and D of Fig. 14), one observed that in dendritic spines and in axons, such particles represented approximately half of the total IMP population. By contrast, in dendritic trunks and

9.31 ±0.13 8.53±0.13 9.01 ±0.14 9.40±0.17 9.41 ±0.10

11.15±0.15 (n--232) 9.95±0.13 ( n - 3 7 8 ) 12.43±0.30(n--146) 10.99±0.11 ( n - 4 3 0 ) 11.65±0.36 ( n = 1 4 6 )

9.64±0.10(n--294) 9.374-0.14 ( n - 3 0 8 ) 9.39±0.15 ( n - 1 9 8 ) 9.01 ±0.13 ( n = 3 0 8 ) 9.02±0.14 ( n = 3 1 6 ) 9.34±0.17 9.69±0.20 10.17±0.19 9.61 i 0 . 1 4 9.32±0.10

Perikaryon

Axon

Perikaryon * ** * * *

E-face

P-face

* Significant differences P < 0.001. ** Significant differences P < 0.002.

2 5 10 21 73

Postnatal

Data are m e a n ± S.E.M. ; n -- number of I M P measured.

IMP mean diameter (nm) in granule cell plasma membranes

TABLE IV

Spine

( n = 152) * (n 196) (n--152) (n--336) * (n--256) *

Dendrite

10.75±0.12 9.98±0.13 10.69±0.29 11.32±0.18 11.05±0.34

(n-378) (n 298) (n=100) ( n - - 148) (n 126)

9 . 2 4 ± 0 . 1 5 ( n - - 1 7 4 ) * * 1 0 . 0 0 ± 0 . 1 8 (n 200) 9.42±0.18(n--126)* 10.14±0.11(n--402) 9.46±0.19(n=194) * 10.51±0.29(n--228) 9.49±0.21(n 140)* 10.99±0.22(n 172) 9 . 0 9 ± 0 . 2 4 ( n - - 8 4 ) * 10.07±0.19 ( n - 2 0 4 ) *

Perikaryon

E-face

A xon

( n - - 3 ~ ) * 10.97±0.15 (n 648) ( n = 2 7 4 ) * 10.17±0.10 (n--682) * 11.38±0.13(n--116) ( n - 2 7 6 ) * 10.69±0.24 ( n = 5 ~ ) 11.35±0.25(n--222) (n--160) 9 . 5 8 i 0 . 1 2 (n 584) *** 1 0 . ~ ± 0 . 1 0 ( n = 7 6 2 ) (n=404) 9.22±0.08 (n-- 1004) * 10.82±0.06 (n-- 1888)

Dendrite

* Significant differences P < 0.001. ** Significant differences P < 0.002. *** Significant differences P < 0.01.

2 5 I0 21 73

Postnatal P-face ageindays Perikaryon

Data are mean ± S.E.M.; n -- n u m b e r of I M P measured.

IMP mean diameter (nm) in Purkinje cell plasma membranes

TABLE llI

10.54±0.11(n--64) 10.88±0.30(n=210) 10.93±0.15(n 320) 10.86±0.07 (n--598)

Spine

tO 00

198±07 263±09 307±21 292±08 2714-13

(n=66) (n=60) ( n = 10) (n=13) (n=19) *

NM 744-19 ( n = 3 3 ) 2214-06 ( n = 2 9 ) 324±48 ( n = 4 6 ) 321 ± 17 (n = 143)

Spine 274-01 694-08 694-07 89±07 964-07

6 5 ± 0 5 (n--86) 125±13(n=53) 266±22 (n=77) 354±24 ( n = 30) 337± 14 ( n = 5 9 )

* Significant differences P < 0.001.

* *

2 5 10 21 73

2494-12 ( n = 1 0 ) 3264-17(n=36) 331 ± 2 2 ( n = 2 9 ) 3374-17 ( n = 2 8 ) 331 :k22 (n--53)

Axon

Postnatal P-face age in days Perikaryon 444-+-04 ( n = 4 8 ) 83±07(n=49) 98±05 (n=54) 91 ± 0 6 ( n = 2 0 ) 944-05 (n-- 13)

Perikaryon

E-face

Data are mean 4- S . E . M . ; n = number of membranes studied.

114-01 ( n = 3 0 )

324-05 (n--56) 41 ± 0 5 (n = 59) 1404-36 ( n = 16) 141 4-21(n--37)

*

* *

Axon

(n=145) (n=32) (n-25) (n=ll) (n=6)

Perikaryon

E-face

Numerical density of lMP of large size ( > 11 nm) in granule cell plasma membrane

TABLE VI

* Significant differences P < 0.001. N M = not measurable.

( n = 17) (n=49) (n--18) ( n = 10) (n=19)

2 5 10 21 73

212-1-09 218±14 252±15 292±15 326±13

Dendrite

Postnatal P-face age in days Perikaryon

Data are mean 4- S.E.M. ; n = number of membranes studied.

Numerical density of IMP of large size ( > 11 nm) in Purkinje cell plasma membrane

TABLE V

* *

41 4-05 62±03 99±12 1394-06 164±13

Dendrite (n-36) (n=23) (n =29) (n--35) ( n = 14) *

*

NM 30±07 534-11 1254-23 101±11

Spine

(n--25) (n=22) ( n = 19) (n=122)

I'O

30 PURKINJE CELLS PERIKARYA ~ ] DENDRITES ABCD ABCD !

i

I

=

i

i

i

GRANULE

ABCD

r

I

i

I

i

CELLS

~ SPINES [ ~ PARALLEL FIBERS ABCD ABCD I

!

I

I

I

i

r

i

!

i

I

I

60

3O

30

2 DAYS 0 .

.

.

.

.

.

.

.

.

.

.

.

.

0 2

3O

30

6O

60

60

60

DAYS

30 5 DAYS 0

-

o

-

0 5

DAYS

50 60 60

60

30

50

I0 DAYS O.

-0

30

30

60

60

60

60

50

30

21 DAYS 0

I0 DAYS

.0 2 I DAYS

30

30

60

6O

60 30

50

73 DAYS O.

.

.

.

.

.

o 73 DAYS

.

30

r-3° 6O =

=

J

i

ABCD

i

I

I

A

I

A BCD

i

1

=

J

J

ABCD

i

=

=

I

ABC

f

i

D

i

i

I

ABC

I

A ~

- 6 0

D

Fig. 14. Histogram of IMP size in function of age and region of plasma membrane studied in Purkinje and granule cells. Numbers on vertical lines indicate the per cent of particle in each size class. Letters on horizontal solid lines represent the different IMP size classes (diameters): A, < 6.9 rim; B, 7-10.9 nm; C, 11-14.9 nm; D, > 15 nm. Columns above the horizontal dotted lines are P-face values; columns below horizontal dotted lines are E-face values.

31 in perikarya, large IMP made up in most cases less than a third of the total IMP population (Fig. 14). When the numerical density of IMP larger or smaller than 11 nm was plotted against postnatal age in all membrane regions, it was found that from the tenth postnatal day on, large particles ( > 11 nm) were distributed in the P-face at the same density in all regions of the neuronal plasma membrane studied (Tables V and VI). DISCUSSION Using a quantitative analysis of the freeze-fracture organization of the neuronal membrane, we have shown that the plasma membranes of two main cell types of the cerebellar cortex, the Purkinje cell and the granule cell, undergo marked changes during postnatal development: (1) in early stages (2-21 days), there is an increase in the number of IMP; (2) the increase is not identical in all regions of the same neuron and results in regional differences in IMP content; (3) the regions of the neuron which develop first (perikarya) have more IMP than regions appearing later (dendrites, axon); and (4) the differences in the number of particles in different regions of the neuron disappear progressively and in the P-face a homogeneous distribution of IMP larger than 11 nm is reached by the tenth postnatal day. Increases in IMP number during neuronal development has been reported in the P-face of axons 17,19,2s. Our study allows us to extend this finding to the other processes of the neuron, i.e. the dendrites. Moreover, the evaluation of IMP in both fracture faces indicate that the changes observed in one membrane leaflet are not only due to change in the particle partition coefficient26, a finding suggested from previously published data on developing axons 17. In growing axons, a gradient of IMP density has been described by Pfenninger and Bunge 17, Pfenninger and Reesela and Small et al.28; this may be interpreted as reflecting a synthesis of the particles in the perikaryon, followed by their transport to remote areas of the axon either by diffusion in the plane of the membrane or by cytoplasmic migration followed by insertion in the membrane at the axonal growth cone. Our results, which show for the first time structural differences between several regions of the postsynaptic membrane receiving different synaptic inputs 2-4,16, raise another question in that not all particles seem to have the same rate of insertion in the membrane; in the P-face IMP larger than 11 nm reach equilibrium from the tenth day on, while IMP smaller than 11 nm still appear not to be distributed homogeneously at the seventy-third day. According to this new data, the specificity of the postsynaptic membrane which is thought of primary importance for the establishment of suitable synaptic contactsT, za would be expressed not only by differences in IMP concentration along the several segments of the neuron, but also by a distinct repartition of certain IMP during development. The hypothesis of Pfenninger and Bunge 17 that a gradient of IMP density in dendritic membrane may explain the acceptance of specific synaptic inputs in different dendritic zones seems thus established on direct morphological evidence and it is also indirectly supported by the fact that the regional differences in IMP density and size were maximal between 2 and 21 days, the period of highest synaptogenesis in rat cerebellar cortex~-4, 27.

32 ACKNOWLEDGEMENTS

We are grateful to Dr. M. Amherdt for help during quantitative evaluation and to Drs. L. Orci and D. Brown for reviewing the manuscript. We also acknowledge the technical assistance of P. Sors, N. Dupont, C. Perroton and P.-A. Riittiman. This work was supported by the Swiss National Science Foundation, Grant 3.120.77.

REFERENCES 1 Addison, W. 14. F., The development of the Purkinje cells and of the cortical layers in the cerebellum of the albino rat, J. comp. Neurol., 21 (1911) 459-487. 2 Altman, J., Postnatal development of the cerebellar cortex in the rat. I. The external germinal layer and the transitional molecular layer, J. comp. NeuroL, 145 (1972) 353-398. 3 Altman, J., Postnatal development of the cerebellar cortex in the rat. II. Phases in the maturation of Purkinje cells and of the molecular layer, J. comp. Neurol., 145 (1972) 399-464. 4 Altman, J., Postnatal development of the cerebellar cortex in the rat. III. Maturation of the components of the granular layer, J. comp. Neurol., 145 (1972) 465-514. 5 Barbera, A. J., Adhesive recognition between developing retinal cells and optic tecta of the chick embryo, Develop. Biol., 46 (1975) 167-191. 6 Barbera, A. J., Marchase, R. B. and Roth, S., Adhesive recognition and retinotectal specificity, Proc. nat. Acad. Sci .(Wash.), 70 (1973) 2482-2486. 7 Barondes, S. H., Neuronal Recognition, Plenum Press, New York, 1976. 8 Berry, M. and Bradley, P., The growth of the dendritic trees of Purkinje cells in the cerebellum of the rat, Brain Research, 112 (1976) 1-35. 9 Gottlieb, D. I., Rock, K. and Glaser, L., A gradient of adhesive specificity in developing avian retina, Proc. nat. ,4cad. Sci. (Wash.), 73 (1976)410-414. 10 Gulley, R. L. and Reese, T. S., Freeze-fracture studies on the synapses in the organ of Corti, J. comp. NeuroL, 171 (1977) 517-544. 11 Hanna, R. B., Keeter, J. S. and Pappas, G. D., The fine structure of a rectifying electrotonic synapse, J. Cell BioL, 79 (1978) 764-773. 12 Hausman, R. E. and Moscona, A. A., isolation of the retina-specific cell-aggregating factor from membrane of embryonic neural retina tissue, Proc. nat. A cad. Sci. (Wash.), 73 (1976) 3594-3598. 13 Landis, D. M. D. and Reese, T. S., Differences in membrane structure between excitatory and inhibitory synapses in the cerebellar cortex, J. comp. Neurol., 155 (1974) 93-126. 14 Landis, D. M. D., Reese, T. S. and Raviola, E., Differences in membrane structure between excitatory and inhibitory components of the reciprocal synapse in the olfactory bulb, J. comp. NeuroL, 155 (1974) 67-92. 15 Moor, H. and Muhlethaler, K., Fine structure in frozen-etched yeast cells, J. Cell Biol., 17 (1963) 609-628. 16 Palay, S. L. and Chan-Palay, V., Cerebellar Cortex: Cytology and Organization, Springer-Verlag, Berlin, 1973. 17 Pfenninger, K. H. and Bunge, R. P., Freeze-fracturing of nerve growth cones and young fibers. A study of developing plasma membrane, J. Cell Biol., 63 (1974) 180-196. 18 Pfenninger, K. H. and Mayle-Pfenninger, M. F., Characterization, distribution and appearance of surface carbohydrates on growing neurites. In Neuronal Information Transfer, Karlin Academic Press, New York, 1978, pp. 373-386. 19 Pfenninger, K. H. and Reese, T. S., From the growth cone to the synapse. Properties of membranes involved in synapse formation. In S. H. Barondes (Ed.), Neuronal Recognition, Plenum Press, New York, 1976, pp. 131-178. 20 Pinto da Silva, P., Translational mobility of the membrane intercalated particles of human erythrocyte ghosts, pH-dependent, reversible aggregation, J. Cell Biol., 53 (1972) 777-787. 21 Pinto da Silva, P. and Martinez-Palomo, A., Induced redistribution of membrane particles, anionic sites and Con A receptors in Entamoeba histolytica, Nature (Lond.), 249 (1974) 170-171.

33 22 Ramon y Cajal, S., Histologie du Systdme Nerveux de rHomme et des Vertebras, Maloine, Paris, 1911, (reprinted 1972, Instituto Ramon y Cajal, Madrid). 23 Ram6n y Cajal, S., Studies on Vertebrate Neurogenesis, Thomas, Springfield, Ill. 1960, (L. Guth, trans.). 24 Rash, J. E., Graham, W. F. and Hudson, C. S., Sources and rates of contaminationin a conventional Balzers freeze-etch device. In J. E. Rash and C. S. Hudson (Eds.), Freeze Fracture: Methods, Artifacts, and Interpretations, Raven Press, New York, 1979, pp. 111-122. 25 Sandri, C., Akert, K., Livingstone, R. B. and Moor, H., Particle aggregations at specialized sites in freeze-etched postsynaptic membranes, Brain Research, 41 (1972) 1-16. 26 Satir, P. and Satir, B., Partition coefficient of membrane particles in the fusion rosette, Exp. Cell Res., 89 (1974) 404-407. 27 Shimono, T., Shoichiro, N. and Sasaki, K., Electrophysiological study on the postnatal development of neuronal mechanisms in the rat cerebellar cortex, Brain Research, 108 (1976) 279-294. 28 Small, R., Strichartz, G. R. and Pfenninger, K. H., Membrane properties of the growing axon: intramembrane particles and saxitonin binding sites, J. Cell Biol., 83 (1979) 279a. 29 Sperry, R. W., Chemoaffinity in the orderly growth of nerve fiber patterns and connections, Proc. nat. Acad. Sci. (Wash.), 50 (1963) 703-710. 30 Venzin, M., Sandri, C., Akert, K. and Wyss, V. R., Membrane associated particles of the presynaptic active zone in rat spinal cord. A morphometric analysis, Brain Research, 130 (1977) 393-404.