Acta neuropath. (Berl.) 31,229--242 (1975) �9 by Springer-Verlag 1975

Electron Microscopic Study of the Developing Capillaries of Human Brain

Jean-Jacques Hauw, Brigit te Berger*, and R a y m o n d Escourolle

Laboratoire de Neuropathologie Charles Foix C. H. U. Piti6-Salp6tribre, Paris, France

Received August 29, 1974; Accepted January 7, 1975

Summary. The ultrastructural characteristics of the cerebellum and olfactory bulb capil- laries were studied in 12 human specimen measuring 25--200 ram.

Type I capillaries, whose lumina were wider than 8 ~, contained no pores or fenestrations. The basement membrane was sometimes discontinuous. In that case, junctions without

quintuple-layered zones could be observed. Such capillaries may correspond to primary vessels in which the sinusoid character has disappeared.

Elsewhere, the basement membrane was continuous and the interendothelial junctions always contained quintuple-layered zones.

Type I I capillaries, whose lumina measured between 2 and 8 tx had basement membranes that were either discontinuous or continuous. They were identical to the immature capillaries described previously in numerous species including man.

Type I I I capillaries, rarely observed, were characterized by the presence of several endo- thelial cells with abundant cytoplasm, which limited the lumina from 0.5--3 ~ in diameters. They had continuous basement membranes.

Type 1V and V capillaries has small or non-patent lumina. The basement membrane was absent or rudimentary. These capillaries appeared to correspond to simple and complex endothelial sprouts.

In Types II, III, V and most Type IV capillaries, the interendothelial junctions contained quintuple-layered zones.

No mature capillaries were observed. The hypothetical pathways of development of cerebral capillaries and the degree of perme-

ability of immature interendothelial junctions are discussed. The long duration of maturation in man probably accounts for the diversity of capillary formations observed as compared with animals. The co-existence over a long period of time of several capillary types with probably varying permeability may play a significant role in the maturation of the blood-brain barrier.

Key words: Capillary -- Maturation -- Human -- Cerebral Development -- Electron ~icroscopy.

Introduction

The growth of h u m a n cerebral capillaries has been studied by electron micro- scopy by Dahl (1963) in embryos and fetuses measur ing 90 to 200 mm Crown- R u m p (C-R) and by Pappas and Pu rpu ra (1964) and Bauer and Vester (1970) at 7 - -12 weeks and 12--16 weeks gestat ional age, respectively. The u l t ras t ruc tura l pa t t e rn of development of the cerebral vascular network, however, could no t be accurately analysed, both in h u m a n and an imal embryos, un t i l the most recent works, in animals, of Breipohl (1968), Caley and Maxwell (1970) and, especially, Bar and Wolff (1972), who in t roduced a more comprehensive classification. Never-

* Maitre de Recherche g I'I.N.S.E.t~.M.

16 Acta neuropath. (Berl.) Bd. 31

230 J.-J. ttauw et al.

theless, m a n y problems concerning h u m a n cerebral vascular embryogenesis remained unresolved. Owing to the obvious impor tance of the development of cerebral capillaries in relat ion to blood-brain barrier, this s tudy was unde r t aken in an a t t emp t to be t te r unde r s t and h u m a n cerebral capillary development by electron microscopy on specimens measur ing 25--200 m m C-I~.

Materials and Methods

Twelve human embryos and fetuses, measuring 25, 30, 72, 75, 100, 105, 135, 145, 170 and 200 mm C-R, whose tissue preservation was acceptable for ultrastructural study were selected for this work.

The embryos and fetuses, obtained between 45 rain to 15 hrs following surgical intervention (ectopic pregnancies, spontaneous or therapeutic abortions) were measured for C-R. The cerebellum and olfactory bulbs were dissected, sectioned in Eagle's medium into 1 • 1 • 0.2 mm cubes and fixed by immersion in 3.5 or 5o/0 glutaraldehyde in Millonig or S6rensen buffers for 1 hr. The tissue was post-fixed in 2~ osmium tetroxide for 45 rain and subsequently embedded in epon. Sometimes prior to embedding, the tissue was treated with uranyl acetate (Karnovsky, 1967). Ultra-thin sections were stained either with uranyl acetate and lead citrate or with bismuth by the method of Ainsworth and Karnovsky (1972).

Results

Several types of developing capillaries were classified according to the size of their luminal diameters [this somewhat artificial classification was al ready adopted by Bfi.r and Wolff (1972) and seemed to us to be the most convenien t for this material]. They are analyzed morphologically below, wi thout regard to the stage of m a t u r i t y to which they correspond.

A. Type I Capillaries

This type of capillary was observed in embryos whose size ranged from 25--100 ram, in the olfactory bulb as well as in the cerebellum and in the lepto- meningeal clefts as well as in the bra in tissue. These capillaries were large and readily recognizable on thick sections under l ight microscope: the lumen was

Fig. 1. Leptomeningeal space of the olfactory bulb of a 170 mm fetus. Type I capillary. Note the wide lumen, many endothelial junctions (arrow-heads) and the inconspicuous basement membrane. A plump eell, presumably a pericyte (P) is seen. Uranyl acetate and lead eitrate

stain. Scale: 5 ~ • 6000

Fig.2--5. Interendothelial junctions of type I and II capillaries

Fig. 2. Cerebellar molecular layer of a 200 mm fetus. Junctional complex of the same type II capillary as illustrated in Fig. 6. Note the quintuple-layered zone (arrow). L lumen. Uranyl

acetate treatment of the block. Scale: 0.1 V • 167500

Fig. 3. Same complex as shown in Fig. 2 at an approximately 1500 A distance (3 serial sections). Note the different pattern of the quintuple-layered zone (arrow) and the appearance of another

junction not demonstrated in Fig.2 (arrow-head). Scale: 0.1 ~t • 137500

Fig.4 and 5. Cerebellar intermediate layer of a 25 mm fetus. Contacts without quintuple- layered zone in type I capillary with discontinuous basement membrane. L lumen. Uranyl

acetate and lead citrate. Scale: 0.1 ~z. Fig.4 • 76000, Fig.5 • 81000

T

o

232 J.-J. Hauw et al.

greater than 8 ~ in diameter, wide enough to contain several blood cells and was lined by several endothelial cells with thin processes that sometimes overlapped in irregular patterns. Their nuclei were fusiform and dense, often with peripheral chromatin. The cytoplasm was flat and electron dense whatever the stain used (uranyl acetate and lead citrate or bismuth). I t contained few organelles such as free ribosomes and rough endoplasmic reticulum, pinocytotic vesicles, mito- chondria, lipid droplets and some microfilaments, which appeared to be evenly distributed. Microvilli often projected towards the lumen. No mitosis was dis- covered.

Junctional complexes were often observed between two adjacent marginal folds. Although sometimes short, they were often very elongated. I t was some- times difficult to analyze them precisely on account of the poor contrast of endo- thelial membranes in embryo. Each junctional complex comprised different zones: i. quintuple-layered segments (often these were reduced to a simple pin- point fusion of the external leaflets of the plasma membranes ; rarely they were elongated, sometimes attaining i ~ in length); 2. alternating zones in which the intercellular cleft between the two plasma membranes, whose internal leaflets were reinforced by hyaloplasmic densifieation, was reduced but still measurable (from 6--12 rim); 3. lastly, zones where the intercellular cleft attained 20--40 nm or more and whose adjacent cytoplasm was sometimes dense, sometimes not.

These junctional complexes were observed whatever the age of the embryo studied. But, in embryos measuring 25 ram, some interendothelial contacts did not contain a quintuple-layered zone (Figs. 4 and 5). Neither pore nor fenestration was observed in the capillaries, even in the youngest embryos.

The basement membrane, which stained better with nranyl acetate and lead citrate than with bismuth, was either continuous in some cases or discontinuous in others. When continuous (Fig. 11), the basement membrane was readily recognizable but the lamina densa and rata were inconspicuous although they were usually clearly seen in astrocytic or subpial marginal glial basement mem- branes. In some cases, it was possible to recognize the fusion between capillary and glial basement membranes. However, the endothelial basement membrane some- times appeared as a thin amorphous, discontinuous, or thinly fibrillar, less electron dense deposit, predominant in front of the pericapillary processes (Fig. 12). I t was in the capillaries containing this type of basement membrane that junctions without the quintuple-layered zone were observed in embryos measuring 25 mm.

Pericytes were consistently present. They were usually flat and had often electron dense, fusifbrm nuclei. However, some sections of discontinuous basement

Fig. 6 and 7. Type II and III capillaries

Fig. 6. Cerebellar molecular layer of a 200 mm fetus. Type II capillary. Section of the peripheral and parajunctional zones. Note the many organelles and continuous basement membrane. The junction (arrow) is magnified in Fig.2. P pericyte. Uranyl acetate treatment of the

block. Scale: 1 ~ • 17500

Fig.7. Olfactory bulb of a fetus measuring 105 ram. Type III capillary. Note the numerou s junctions, the narrow lumen and continuous basement membrane. Uranyl acetate and lead

citrate. Scale: 1 ~ • 13000

rQ

234 J.-J. Hauw et al.

membrane capillaries revealed irregularly-shaped plump nuclei (Fig. 1). The cyto- plasm was often distributed in thin strips; it contained rough endoplasmic reti- culum, rather well developed in the nuclear region, some mitochondria and pino- cytotic vacuoles.

B. Type I I Capillaries

These capillaries were observed in the olfactory bulb and cerebellum of embryos and fetuses measuring 25--200 nm. They were characterized by a small lumen measuring 3--8 ~z, thus probably permitting only a single blood cell. They were lined by a few endothelial cells, whose cytoplasm, sometimes more abundant in the organelle region 1, projected into the lumen. The cytoplasm often contained numerous organelles as well in the organelle region as in the peripheral zone1: mitochondria, rough endoplasmic reticulum and Golgi or pinocytotic vesicles, a small number of free ribosomes, often grouped into po]ysomes and a few dense bodies and lipid droplets. The cytoplasmic membrane sometimes exhibited several microvilli that projected into the lumen. Although the fairly clear nucleus always extended parallel to the capillary axis, it plainly projected into the lumen.

Two or three junctions were usually recognizable in cross sections of the capillary. Often tortuous, these junctions appeared identical to the complex junctions observed in Type I capillaries (Figs. 2 and 3).

The thickness of the basement membrane varied from 0--130 nm and was always less than that of the basement membrane of the mature capillaries. In most cases it was discontinuous. Its density varied greatly and it appeared quite distinct in some cases and less distinct in others. A continuous basement membrane was observed only in three cases of this type of capillary in fetuses measuring 72, 75 and1200 mm (Fig. 6). I t was always contiguous with the astrocytic processes. Pericytes could be observed. Frequently their cytoplasm appeared to be more electron dense than that of the endothelial cells and, especially, of the adjacent nerve and glial cells (Fig. 6). I t often covered the endothelium to a great extent. The dense nucleus exhibited thick peripheral chromatin and extended parallel to the lumen; it often had several indentations.

C. Type I I I Capillaries

This structure was found in the superficial olfactory layer of the olfactory bulb. I t was small and not identifiable by light microscopy; its lumen measured

Fig. 8 and 9. Type IV capillaries

Fig.8. Leptomeningcal space of the olfactory bulb of a 170 mm fetus. Vascular sprout con- sisting of numerous endothelial cells, two nuclei of which are seen. Note the many junctional complexes (arrows) and small intercellular clefts representing the primordial lumen. The basement membrane appeared thin, interrupted or absent in this type of capillaries. Uranyl

acetate and lead citrate. Scale: 5 ~z • 7500

Fig. 9. Cerebellar molecular layer of a 200 mm fetus. Vascular sprout with a pericyte (P). The lumen (L) is wider. Note the very thin and interrupted basement membrane (arrows). Uranyl

acetate treatment of the block. Scale: 5 ~ • 10000

1 As defined by M. Simionescu, N. Simionescu, and G. E. Palade: Morphometric data on the endothelium of blood capillaries, J. Cell Biol. 60, 128--152 (1974).

Development of Human Cerebral Cupill~ries 235

Figs. 8 ~nd 9

236 J.-J. ttauw et al.

from 0.5--3 ~z in the greatest width. I t contained no blood cells. Several endothelial cells were present; their nuclei were large and their long axes were directed towards the lumen; their c}~oplasm was abundant. The endothelial cells showed several kinds of cytoplasmic organelles such as fairly numerous ribosomes, smooth and rough endoplasmic reticulum, elements of the Golgi apparatus, several pinocytotic vesicles and abundant mitochondria. These endothelial cells were joined by some- what tortuous, extended junctions which appeared similar to those of the other types of capillaries described above. Pericytes and a fairly dense, amorphous continuous basement membrane, sometimes contiguous with that of the glial processes, were observed.

D. Type I V Capillaries

Up to now, this type of capillary was observed only in the leptomeningeal cleft of the olfactory bulb in a fetus measuring 170 mm and in the cerebellar inter- mediate and molecular layers in embryos and fetuses from 50--200 mm.

These formations, unrecognizable by light microscopy, with tortuous non- patent lumina, were composed of two or more endothelial cells joined by pin-point or extended junctions. They had irregular nuclei that often lay parallel to one another. Their cytoplasm was dense and contained numerous free ribosomes and mitochondria, a few rough endoplasmic reticulum cisternae and pinocytotic vesicles. 5,To mitosis was observed. The interendothelial junctions of these capil- laries seemed to be either of the complex type, comparable to that already described, or made up of a simple localized reduction in the width of the extra- cellular space, which was marked by condensation of the adjacent cytoplasm without an observable quintuple-layered zone. Type IV capillaries did not have a well individualized basement membrane, but sometimes showed limited, irregular condensation of an amorphous substance near the plasma membrane. Dark adjacent cells with neither junctions nor tubules, well developed rough and smooth endoplasmic reticulum and fairly abundant mitochondria were observed and appeared to correspond to pericytes (Fig. 9).

E. Type V Capillaries

In this last type, some complex formations composed of several pericytes and endothelial cells were observed. The endothelial cells were joined together by multiple junctions of the complex type, i.e., with quintuple-layered zones, leading to a complex luminal pattern. Although the capillary structure was

Fig. 10. Type V capillary. Leptomeningeal space of the olfactory bulb of a 170 mm fetus. Complex sprout consisting of numerous endothelial cells (E) and pcricytcs (P) (see text).

Uranyl acetate and lead citrate. Scale: 1 ~ X 11700

Fig. 11 and 12. Basement membrane of Type I capillaries

Fig. 11. Olfactory bulb of a 72 mm fetus. Continuous basement membrane (arrow). L lumen; E endothelial cell; P pericyte. Uranyl acetate treatment of the block. Scale: 0.5 fz • 55000

[Fig. 12. Intermediate layer of the cerebellum of a 50 mm embryo. Discontinuous basement membrane. Stage I of B~r and Wolff (1972) (arrows). L lumen; E endothelial cell. Uranyl

acetate and lead citrate. Scale: 0.5 [l • 43000

Development of Human Cerebral Capillaries 237

Figs. 10--12

238 J.-J. Huuw et al.

complex, the cellular organization appeared similar to that of the Type IV capil- laries.

This "Type V" capillary was observed only in a fetus measuring 170 mm, in the leptomeningeal cleft of the olfactory bulb.

Discussion

Capillaries that we grouped as Type I were not observed in man by Dahl (1963) or Pappas and Purpura (1964). Some of Bauer and Vester's high magnification electronmicrographs (1970), particularly their Figs. 6, 8 and, perhaps, 17 and 18, obtained from fetuses measuring 200, 90, 100 and (perhaps) 110 mm respectively, may correspond to this capillary type but these authors did not distinguish Type I from Type II capillaries.

I f we accept the hypothesis that the development of the basement membrane is one of the criteria of vascular maturation (Bar and Wolff, 1972), the vessels that we grouped in Type I may correspond to two different stages:

1. By virtue of their diameter, the flatness of the endothelium, and their presence from the earliest stages studied (25 ram), Type I capillaries with dis- continuous basement membrane (Figs. 1 and 8) may correspond to primary vessels as described by Breipohl (1968) and Bar and Wolff (1972). However, they do not contain pores or fenestrations, which are characteristically present in primary vessels in the early stages, but which disappear during maturation (Bar and Wolff, 1972). Thus, they may be primary vessels that have lost their sinusoid charac- teristics.

2. Type I capillaries with continuous basement membrane (Fig.7) have never been demonstrated. They may be more mature vessels that develop either from Type I capillaries without continuous basement membrane or from Type I I capil- laries. Nevertheless vessels at this immature stage must be considered as capil- laries, the site of metabolic exchange (Majno, 1965).

Type I I capillaries are easier to identify. They undoubtedly correspond to the immature capillaries already described in men (Dahl, 1963; Pappas and Purpura, 1964; Bauer and Vester, 1970) and animals: chicken (Gruner, 1961; Wechsler, 1965; Wechsler and Meller, 1967; Delorme et al., 1968); mouse (Breipohl, 1968); rat (Donahue and Pappas, 1961; Shimoda, 1963; Tani and Ishii, 1963; Phelps, 1972) ; dog (Ichev, 1970) ; eat (Pappas and Purpura, 1964) ; rabbit (Dona.hue, i964). These immature capillaries will develop into mature capillaries (Bier and Wolff, 1972). In man, studies dealing with specimens obtained from fetuses of advanced gestation (premature infants) did not demonstrate any immature capillary in the cerebral cortex but reported the presence of a thin and irregular capillary basement membrane in the youngest subjects of 25 and 28 weeks gestational age (Gruner, 1970).

The apparently rare Type I I I capillaries are more difficult to classify. Perhaps, one eleetronmierograph obtained by Bauer and Vester (1970) from an embryo measuring 50 mm illustrates them, but as the figure (12) is only partial, their presence cannot be confirmed. Does this type of capillary merit being isolated? Is it only an artifact, a matter of the particular section or a seldom-encountered normal structure? I f it were the first of these three possibilities, then, because of

Development of Human Cerebral Capillaries 239

the number of endothelial cells and of the presence of a continuous basement membrane, we must conclude that it is an artifactual shrinkage of a Type I or I I capillary. The characteristics of the adjacent brain tissue (without enlargement of the extracellular space) seem to contradict this interpretation. I t may also be simply a tangential section of the extremity of the loop of a Type I I capillary. Although vascular loops can be observed in the developing nervous system in vertebrates as well as in invertebrates (Craigie, 1955), their presence could not be confirmed, particularly by electron microscopy in most of the higher vertebrates (B/~r and Wolff, 1972).

Type IV capillaries clearly correspond to the endothelial sprouts observed in the central nervous system of human embryos measuring 90 and 100 mm (Figs.2 and 3 of Bauer and Vester, 1970) and in animals (Caley and Maxwell, 1970; B/~r and Wolff, 1972). I t is presumed that these develop from primary sinusoid vessels and become immature capillaries.

Should Type V capillaries (complex sprouts) be distinguished from Type IV capillaries (more simple sprouts)? They could simply demonstrate different parts, Type IV characteristics corresponding to sections closer to the extremity, as suggested by B/~r and Wolff (1972). I t could also correspond to the arising or bifur- cation zones.

I t sometimes seems difficult to formally distinguish Type IV capillaries con- taining rather wide lumina (1--2 ~x) from immature capillaries (Type II). This difficulty results from the somewhat artificial classification adopted: the second type seems derived from the first; thus, the presence of intermediate forms cannot be surprising.

The interendothelial j,~tnctions were seen in all sections examined. We found that seamless capillaries were rare or questionable, contrary to the observations of Wolff and B/~r (1972) in the rat. On the other hand, we saw no pores or fenes- trations, phenomena which seem to correspond to earlier stages of development than the ones studied in our specimens. This observation is linked with those made by other authors who have studied human embryonic cerebral capillaries.

The junctions appeared to be complexes comprising quintuple-layered seg- ments, for some at least, in all types of capillaries. Nevertheless, it is appropriate to doubt the true tightness of the junctions observed, even if they did contain a quintuple-layered contact. They could be physiologically permeable transitory localized fusions as demonstrated by Delorme (1972) in the chick embryo. Some electron micrographs that we observed support this hypothesis (compare Figs. 2 and 3). They are, in any event, similar to those of Delorme (1972). Furthermore, no tilting of the specimen was performed. Such a study, coupled with freeze etching and analysis of tracers diffusion is needed for the discrimination between focal tight junctions and gap junctions as seen in the early mesenehymal structures (Revel et al., 1973).

The growth of Basement membrane does not appear to parallel the age of the embryo. I t is already present, thick and continuous in some capillaries of embryos measuring 25 mm. Sometimes, on the contrary, it is missing in older fetuses. I t should be noted that the basement membrane is less thick in sprouts (where in fact it is not constant) and more developed in Type I I capillaries (in which it is some- times continuous). On the other hand, the basement membrane seems to be of

240 J.J . Hauw et al.

13

I

Mature capillary

Fig. 13. Schematic drawing summarizing the specific features of the various types of capillaries observed in the growing human brain. Arrow-heads indicate the developing basement mem- brane. The various pathways of the development such as proposed by Bar and Wolff (1972) in the cerebral cortex of rats (solid arrows) and those hypothetized in man (interrupted arrows) are shown. The mature capillary not seen in developing brains of our studied cases is included

for a better understanding. For details, see text

variable thickness in Type I capillaries, sometimes being discontinuous, some- times continuous. These results do not depend on the staining technique used and do not seem to parallel the continuity or discontinuity of the network of pericapil- lary processes. The most attractive hypothesis to interpret these data is that the basement membrane appears progressively, not in relation to the age of the embryo, but in relation to the age of the capillary (B~r and Wolff, 1972).

Pericytes appear early, as soon as the endothelial sprout develops (Type IV and V). Since the basement membrane is not always seen at this stage, pericytes cannot be identified by their relation to the basement membrane (i.e., in one of its folds). On the other hand, their position adjacent to the endothelium, the absence of junctions in between, their strong electron density of cytoplasm and the paucity of microtubules appear characteristic.

In conclusion, the study of human cerebral capillary maturation derives its interest from the slow development of the central nervous system in man. This probably explains the presence of structures found in our study that have not been

Development of Human Cerebral Capillaries 24I

d iscovered up to now in animals , n a m e l y Type I I I capillaries, and Type I capil- laries wi th cont inuous b a s e m e n t membrane . These var ious format ions as shown in this s t u d y could cor respond to several , pe rhaps concomi tan t , p a t h w a y s of vascu la r deve lopmen t (d i ag rammed in Fig. 13); the basemen t m e m b r a n e a p p e a r e d to grow wi thou t re la t ion to the ges ta t iona l age of the embryo , b u t wi th re la t ion to the m a t u r i t y of the vessel. I n the per iod of deve lopmen t t h a t we s tud ied (i.e., 25- -200 m m C-R), excep t in the mos t i m m a t u r e vessels, in te rendothe l i a l junc t ions seemed to be cons t i tu t ed b y complexes conta in ing qu in tup le - l aye red zones. Junc t ions wi thou t qu in tup le - l aye red zones were observed only in some eases of Type I capi l lar ies t h a t d id no t show cont inuous basemen t m e m b r a n e and in a few capil lar ies of Type IV. One fact should f inal ly be emphas ized : as a l r e a dy observed in animals , pa r t i cu l a r ly b y Caley and Maxwell (1970), several cap i l l a ry s t ruc tures of poss ib ly different pe rmeab i l i t y coexis t Over a long per iod of t ime. This coexis tence seems to be especia l ly m a r k e d in man, whose per iod of m a t u r a t i o n is long, and m a y p l a y a p a r t in the d e v e l o p m e n t of the b lood-bra in barr ier .

The authors are indebted to Professors Hervet, Gautray and Lanvin and Doctors Bout, Josso and Papiernick for their help with obtaining specimens and Mrs. Simonneau, Misses Asselineau, Sallaberry and Tong for their accurate technical assistance. This work was sup- ported by funds from the D~l~gation G~n~rale ~ la Recherche Scientifique et Technique (Con- vention 71.7.2919) and I.N.S.E.R.M.

References

Ainsworth, S. K., Karnowsky, M. J.: An ultrastructural staining method for ferritin in thin sections. J. Histochem. Cytoehem. 20, 225--229 (1972)

Biir, T., Wolff, J. l~. : The formation of capillary basement membranes during internal vaseu- larization of the rat's cerebral cortex. Z. Zellforsch. 133, 231--248 (1972)

Bauer, K. F., Vester, G.: Das elektronenmikroskopische Bild der Itirneapillaren menschlieher Fetch. l%rtschr. Neurol. Psyehiat. 38, 270--318 (I970)

Breipohl, W.: Entwicklung und gli6se Beziehungen der Basalmembran der Kapillaren des M~insegehirns. Aeta anat. (Basel) 69, 393--408 (1968)

Caley, D. W., Maxwell, D. S.: Development of the blood vessels and extraeellular spaces during postnatal maturation of rat cerebral cortex. J. eomp. Neurol. 138, 31--48 (1970)

Craigie, E. A. : Vascular patterns of the developing nervous system. In: Biochemistry of the developing nervous system, pp. 28--49. New York: Academic Press 1955

Dahl, V. : The ultrastructure of capillaries in cerebral tissue of human embryos. A preliminary report. Dan. reed. Bull. 10, 196--199 (1963)

Delorme, P.: Diff6reneiation ultrastrueturale des jonctions intercellulaires de l'endoth@lium des eapillaires t61enc6phaliques chez l'embryon de poulet. Z. Zellforsch. 133, 571--582 (1972)

Delorme, P., Grignon, G., Gayet, J . : Ultrastructure des capillaires dans le t61enc@hale du poulet au eours de l'embryog6n6se et de la eroissanee postnatale. Z. Zellforsch. 87, 592--602 (1968)

Donahue, S. : A relationship between fine structure and function of blood vessels in the central nervous system of rabbit fetuses. Amer. J. Anat. 115, 17--26 (1964)

Donahue, S., Pappas, G. D. : The fine structure of capillaries in the cerebral cortex of the rat at various stages of development. Amer. J. Anat. 108, 331--347 (1961)

Gruner, J. E. : Les basales dans le n6vraxe. Histopathologie fine. C.R. Soc. Biol. (Paris) 1,56, 166--170 (1961)

Gruner, J. E. : The maturation of human cerebral cortex in electron microscopy study of post- mortem punctures in premature infants. Biol. Neonat. (Basel) 16, 243--255 (1970)

Ichev, K. : Ultrastructural parallels and differences between vessels in young individuals and newly-formed ones in adults. C.R. Acad. bulg. Sci. 28, 1015--1018 (1970)

242 J . J . Hauw et al.

Karnovsky, M. J. : The ultrastructural basis of capillary permeability studied with peroxydase as a tracer. J. Cell Biol. 85, 213--236 (1967)

Majno, G.: Ultrastructure of the vascular membrane. In: Handbook of Physiology, sect. 2, vol. I II . Washington: Amer. Physiol. Soc. Ed. 1965

Pappas, G. D., Purpura, D. P. : Electron microscopy of immature human and feline neocortex. Progr. Brain Res. 4, 176--186 (1964)

Phelps, C. H. : The development of glio-vascular relationships in the rat spinal cord. Z. Zell- forsch. 128, 555--563 (1972)

Revel, J. P., Yip, P., Chang, L. L. : Cell junctions in the early chick embryo. A freeze etch study. Develop. Biol. 85, 302--317 (1973)

Shimoda, A.: An electron microscope study of the developing rat brain, concerned with the morphological basis of the blood-brain barrier. Acta path. jap. 18, 95--105 (1963)

Tani, E., Ishii, S. : Ontogenic studies on the rat brain capillaries in relation to the human brain tumor vessels. Aeta neuropath. (Berl.) 2, 253--270 (1963)

Wechsler, W. : Die Entwicklung der Gef~Be und perivascul~ren Gewebsr~ume im Zentral- nervensystem yon Hiihnern (Elektronenmikroskopiseher Beitrag zur Kenntnis der mor- phologischen Grundlagen der Blut-Hirnschranke wi~hrend der Ontogenese). Z. Anat. Entwickl.-Gesch. 124, 367--395 (1965)

Wechsler, W., iVIeller, K.: Electron microscopy of neuronal and glial differentiation in the developing brain of the chick. Progr. Brain Res. 26, 93--144 (1967)

Wolff, J. R., B~r, T. : "Seamless" endothelia in brain capillaries during development of the rat 's cerebral cortex. Brain Res. 41, 17--24 (1972)

Dr. J . -J . Hauw Laboratoire de Neuropathologie Charles Foix C.H.U. Piti6-SalpStriSre 47 Bd. de l 'H6pital F-75013 Paris France