Journal of Neurocytology 4, 453-468

(I975)

Complex tight junctions of epithelial and of endothelial cells in early foetal brain K. M O L L G A R D

1 and N. R. S A U N D E R S

~'

1The Laboratory of Electron Microscopy, Anatomy Department A, University of Copenhagen, Universitetsparken 1, 2100 Kobenhavn O, Denmark 2Department of Physiology, University College London, Gower Street, London WC1E 6BT, U.K.

Received 3oth December 1974; revised 7 March 1975; accepted I2 March 1975

Summary T h e morphology of epithelial and of endothelial intercellular junctions in human foetal (9-I5 weeks gestation) and sheep foetal (50, 60 and 125 days gestation, term 147 days) brain has been studied using the freeze-fracture technique and thin section transmission electronmicroscopy. Freeze-fracture replicas of the choroid plexus of both early human and sheep foetuses showed that the choroidal ependymal cells are linked at the ventricular surface by tight junctions. Freeze-fracture replicas of foetal cortical endothelial cell junctions showed that they are still more complex than those of choroidal epithelial cells, in all specimens so far examined. I n some 60 day sheep foetuses the dye Alcian blue, which binds to plasma albumin and which is electrondense when treated with osmium tetroxide, was injected intravenously a few minutes prior to fixation. T h e dye penetrated from blood into brain extracellular space and c.s.f, but apparently not by an intercellular route. T h e dye was found in a tubular system (endoplasmic reticulum) in both choroidal epithelial and cortical endothelial cells. T h e possibility that protein penetrates into the foetal brain and c.s.f, by a transcellular route is discussed. T h e possible significance of these findings in relation to previous ideas and studies of the development of blood-brain barrier mechanisms is also considered.

Introduction

The concept of a blood-brain barrier arose from investigations which demonstrated that certain dyes injected intravenously stained most tissues but not the brain (Ehrlich, 1885). Later it was shown that these dyes were bound to plasma proteins, so that the barrier to the dyes was actually a barrier to dye-protein complexes (Tschirgi, 195o). More recently (see Davson, 1967) the concept has been extended to include not only mechanisms which prevent entry of materials into the brain and c.s.f, but also to mechanisms which control the composition of the brain's extracellular environment. A description of the morphological 9 1975 Chapman and Hall Ltd. Printed in Great Britain

454

MOLLG~RD and

SAUNDERS

localization of the barrier to circulating protein was given by Reese and Karnovsky (1967) and by Brightman and Reese (I969) using horseradish peroxidase, a protein which can be localized by electron microscopy. Endothelial cells of normal brain capillaries are characterized by the presence of only a small number of pinocytotic vesicles, by the absence of fenestrations and by the presence of tight junctions between neighbouring endothelial cells (Reese and Karnovsky, 1967). Reese and Karnovsky (I967) and Brightman and Reese (I969) have suggested that these junctions are the structural basis for the blood-brain barrier to protein. Using horseradish peroxidase and colloidal lanthanum, Brightman and Reese (1969) also demonstrated that mature choroidal epithelial cells are linked at their ventricular surface by tight junctions which were suggested to be the site of the blood-c.s.f, barrier. It has frequently been suggested that the mechanisms which control the brain's extracellular environment are immature in the developing brain (e.g. Barcroft, I938; Bakay, I953; Lee, I97I). As pointed out elsewhere (Evans, Reynolds, Reynolds, Saunders and Segal, I974), evidence for immaturity of these mechanisms is sparse and much of it is open to alternative explanations. Some dye studies have been interpreted as showing that the bloodbrain barrier to dye or protein-bound dye is incomplete in the foetus or new born of some species, for example the experiments of Behnsen (1927) and of N~t~inen (I947). However, as will be considered in the discussion, the investigations of Gr6ntoft (I954) in previable human foetuses showed that dye does not penetrate into even very immature brains unless the foetus is asphyxiated for some time. In spite of this evidence, many authors have continued to suggest that the blood-brain barrier to dye or protein-bound dye is immature in the foetus and that this immaturity might be due to a lack of development of tight junctions in cerebral vessels of foetal animals. The appearance of tight junctions in the developing brain has been investigated in the present study, using the freeze-fracture technique. This method (Moor and M~hlethaler, I963; Branton, I966) has revealed that tight junctions consist of a number of strands presumed to be sealing elements and which are thought to correspond to the points of membrane fusion in thin sections of tight junctions (Kreutziger, I968; Staehelin, I973). The arrangement of the strands has been shown to vary with different epithelial tissues (Friend and Gilula, I972) and a correlation between transepithelial resistance and the number of strands has been proposed (Claude and Goodenough, I973). This paper presents evidence that choroidal epithelial and cerebral endothelial cells in the immature brain already possess very complex tight junctions. These observations confirm that dye or protein-bound dye would not be expected to penetrate into even very immature brain via an intercellular route. Some preliminary evidence is presented for the penetration of the dye Alcian blue across endothelial and choroidal epithelial cells in a tubular system which might be a transcellular pathway for dye or protein penetration. Materials and methods

The material examined comprises visual cortex and choroid plexus from human foetuses of 6o-I6o rnm crown-rump length (CRL) (approximately 9-I5 weeks gestation) and from foetal sheep of three gestational ages (50, 6o and x25 days; term I47 days).

Foetal brain epithelial and endothelial cell tight junctions

455

HUMAN MATERIAL H u m a n material was obtained by caesarean sections in connection with legal abortion. Immediately following the operation the brain was taken out and immersed in ice-cold 2.5 % glutaraldehyde in o. IM cacodylate buffer (pH 7.4). Small pieces of choroid plexus and blocks of visual cortex were prepared after I h of fixation and then kept for another 2 h in the fixative. Following I h of infiltration with 3o% buffered glycerol at room temperature, the specimens of cortex were mounted and oriented so that a transverse cleavage perpendicular to the cortical surface could be achieved. SHEEP MATERIAL Ewes (Clun Forest) were anaesthetized with 5% thiopentone and 1% chloralose and the foetuses delivered by caesarean section as described in Bradbury, Crowder, Desai, Reynolds, Reynolds and Saunders (I972). Perfusion and immersion fixation were performed with a modified Karnovsky's fixative (del Cerro, personal communication): 3 % glutaraldehyde and 3 % paraformaldehyde in 0.2 M cacodylate buffer is diluted with 2 volumes of o.I M cacodylate. T h e mixture has an osmolarity of about 65o mosmol and the pH is adjusted to 7-4; it is referred to as 1% Karnovsky. Two brains of 125 days gestation were fixed in 3 % Kamovsky. Small pieces of choroid plexus and blocks of visual cortex were prepared as described for h u m a n foetuses. FREEZE-FRACTURE Following infiltration with glycerol the specimens were frozen in liquid Freon 22 cooled by liquid nitrogen. Freeze-fracturing followed by platinum-carbon shadowing was performed with a Balzer apparatus (BAF 3oI) (Moor and Mfihlethaler, I963; Branton, I966). Specimens were fractured at a stage temperature of -- I I5~ Prior to the shadowing some of the specimens were etched for between 80 sec and 5 min at -- Ioo~ Replicas were cleaned in dimethylformamid, bleach and chromic acid. It is assumed on the basis of previous evidence (Branton, I966; Pinto da Silva and Branton, I97o) that freeze-fracturing actually splits membranes, producing two complementary fracture faces: the outside surface of the cytoplasmic half of the membrane (the A face) and the inside surface of the external half of the membrane (the B face). The tight junction consists of a meshwork of anastomosing and branching strands which appear as ridges on the A face and as grooves on the B face. The term strand will be used for either ridges or grooves. T H I N SECTIONING Prior to fixation I ml of 0.5% Alcian blue in Krebs solution was injected via a placental cotyledenary vein over 2-5 m i n in two 6o day sheep foetuses. 5-Io m i n later the brains were fixed by perfusion with 1% Kamovsky over 15 min; they were then removed and placed in 1% Kamovsky fixative overnight. T h e specimens were washed (24-48 h) in buffer and then postfixed for 2 h at 4~ in 2% osmium tetroxide in o.i M cacodylate buffer; they were stained en bloc in o.5% aqueous uranyl acetate for I h, dehydrated in increasing concentrations of ethanol, transferred to propylene oxide and embedded in Epon.

Results FREEZE-FRACTURE AND FREEZE-ETCH APPEARANCE OF IMMATURE CHOROIDAL EPENDYMAL CELLS

Freeze fracturing Freeze-fracturing of choroidal ependymal cells of both human and sheep foetuses shows that

456

MOLLGARD and SAUNDERS

they are linked at the ventricular surface by tight junctions. The meshwork of anastomosing and branching strands which appear as ridges on the A face and as grooves on the B face (Figs. I and 2) seems to form a beltlike structure around each choroidal epithelial cell, since it is consistently observed whenever the apical membrane is exposed. The appearance of the tight junctions is quite different from area to area even within the same region of one replica. For example, Figs. I and 2 are electronmicrographs obtained from two adjacent pieces of the same sample of choroid plexus. All investigated junctions (about 60) had a minimum number of 2- 3 complete strands and 6-8 strands were often observed (cf. Bohr and Mellg~rd, 1974). The area of junctional membrane which in the B face is enclosed by a continuous groove is convex; in the A face it is enclosed by a continuous ridge and appears concave. This means that corresponding areas of junctional membrane of adjacent cells bulge away from each other except in the strand area, where ~hey are probably kept in contact. Circular invaginations of the lateral cell membrane are often seen (Fig. I, curved arrows). In the apical part of the membrane these invaginations are frequently found in close association with some of the blindly-ending strands of sealing elements which are connected to the basal side of the apical tight junctional formation (Fig. I).

Freeze-etching Extensive areas of cytoplasmic membrane can be visualized after deep etching (Fig. 3). An apparently continuous network of membrane-limited tubules extends from invaginations at the base of microvilli of the apical surface towards the lateral cell membrane. THIN SECTIONS OF CHOROIDAL EPENDYMAL CELLS FOLLOWING INTRAVENOUS INJECTION OF ALCIAN BLUE The choroidal cells are rich in smooth and granular endoplasmic reticulum, particularly near the lateral cell membrane where a very close relationship exists between reticulum and cell membrane (Figs. 4 and 5). Portions of the smooth reticulum extend towards the apical cell membrane where they seem to terminate at the base of the microvilli (open arrows in Figs. 5 and 6). Following intravenous injection of the marker substance Alcian blue, a particulate precipitate can be observed inside an extensive tubular system belonging to the endoplasmic reticulum (Figs. 6 and 7). Sometimes the intracellular system of tubules can be followed up to the base of the microvilli where fusion of an area of the external cell membrane and of the intracellular tubular system forms an electron dense structure apparently separating the c.s.f, and the tubular system (small arrow in Fig. 6).

Fig, I, Freeze-fracture replica of human foetal choroidal epithelial cell. The fracture plane has exposed a large part of the lateral cell membrane (the B face: B). The apical tight junctional meshwork consists of at least 2-3 complete strands running roughly parallel to the apical cell surface. Curved arrows point to circular membrane invaginations which are often associated with blindly-ending strands. Some of these strands are indicated by black arrows. CRL I4I mm. • 35 ooo.

458

MOLLGARD

and

SAUNDERS

Mitochondria, vesicular profiles, which might be cross-sectioned tubules, and coated vesicles are found in close association with the system of membrane-limited tubules. A precipitate on the surface in contact with c.s.f. (the apical cell membrane) is occasionally seen (Fig. 6). F R E E Z E - F R A C T U R E A P P E A R A N C E OF E N D O T H E L I A L

CE L L MEMBRANES IN FOETAL

NEOCORTEX

The only replicas used were those where it was possible to distinguish clearly the cortical plate and the free surface of the cerebral wall. Thus far only six replicas of endothelial cell junctions have been obtained, partly because of the difficulty of preparing replicas of foetal brain. Four of the replicas were from human foetuses of IO and 13 weeks gestation and two were from a relatively more mature sheep foetus of I25 days gestation. The cortical plate of IO-I 3 week human foetus appears as a dense band of cells lying just below the molecular (or marginal) layer. In the I25 day sheep foetus the cell density is less. Blood vessels can be followed for quite long distances as they penetrate the cortical wall from a plexus on the free surface. By using a cleavage plane perpendicular to the cortical surface it is possible to obtain longitudinal fractures which expose extensive areas of membrane. The endothelial cells are arranged in such a way that the long axis of the tight junctions between them lies approximately along the length of the vessel, whereas their short axis runs between the inside (luminal) and outside (abluminal) surfaces of the vessel. Very complex tight junctions connect the endothelial cells in the area of membrane apposition (Figs. 8 and 9). The tight junctional meshwork seems to form a continuous longitudinal band at the region of endothelial cell overlap since strands are consistently observed whenever this region is exposed.

Junctional configuration The junctional strands run roughly parallel to each other as well as to the free luminal border of the endothelial cell. Sometimes the strands are heavily interconnected in a very complex manner, giving the junction a mosaic-like appearance. Where the total luminal part of the endothelial cell membrane is exposed the junctional meshwork can be seen to form a well defined continuous band (Fig. 8). No blindly-ending strands of sealing elements are connected to the abluminal part of the tight junction (Fig. 9) and no completely dissociated strands have been observed.

Fig. 2. Freeze-fracture view of membrane cleavage faces at the apical tight junctional formation of human foetal choroid plexus. The tight junction consists of anastomozing ridges on the A face (A) and grooves on the B face (B). This replica and the replica used in Fig. i were obtained from the same foetus. Note the very complex pattern of junctional strands compared with the strands shown in Fig. i. The area of junctional membrane enclosed by strands appears concave in the A face and convex in the B face. C indicates where the plane of fracture passes through the cytoplasm of the cell. Human foetus, CRL r4r mm. • 58 ooo. Fig. 3. Freeze-etch replica of human foetal choroidal epithelial cell. After 5 rain of etching an extensive tubular network of membranes is apparent (-~) in the apical cytoplasm of choroidal ependymal cells. Human foetus, CRL x4r ram. • 72 ooo.

F i g s . 4 a n d 5. Electronmicrographs of thin sections of choroidal epithelial cells from a 6o day sheep foetus. T h e tubular system (TS) seems to make very close contact with the lateral cell membrane (LM) which is obliquely cut in these sections. Open arrow indicates a point of contact between the tubular system and the apical cell membrane. T h e junctional complex (JC) separates the apical cell membrane and c.s.f. (L) from the lateral cell membrane and the lateral interspace. Magnification • 97 ooo.

Strand number I n t h e v e r y r a r e cases (two u n t i l n o w ) w h e r e a l a r g e area o f j u n c t i o n a l as well as n o n j u n c t i o n a l m e m b r a n e is e x p o s e d , in t h e d i r e c t i o n l u m i n a l - a b l u m i n a l , t h e s t r a n d n u m b e r can F i g s . 6 a n d 7. These electronmicrographs were obtained from the choroid plexus of another 60 day sheep foetus which received Alcian blue i.v. about Io min prior to fixation. A particulate precipitate (P) is observed inside some parts of the endoplasmic reticulum. I n Fig. 6 a precipitate can also be seen on the apical cell membrane (curved arrows). T h e tubular system makes very close contact with the apical cell membrane, the outer surface of which is exposed to the c.s.f. (open arrows, Fig. 6). An electron dense structure (small arrow, Fig. 6) seems to separate the c.s.f, and the tubular system. In Fig. 7 a similar precipitate-containing tubular system can be seen (arrows). T h e system appears to make very close contact with the lateral cell membrane (LM) and the lateral intercellular.space which is separated from the c.s.f (L) by a junctional complex (JC). Magnification • 5o ooo.

462

MDLLGARD and SAUNDERS

be calculated in the manner of Claude and Goodenough (I973) who counted the minimum number of interconnected junctional strands interposed between luminal and lateral membrane surfaces in epithelial tissues. In the case of endothelial cell tight junctions the number of junctional strands is on average about Io (Fig. 9) but even in cases where only a fraction of the junctional membrane (from luminal to abluminal side) is exposed an average number of 4-5 is found. T H I N S E C T I O N S F R O M N E O C O R T E X OF 6 0 DAY S H E E P F O E T U S E S F O L L O W I N G I N T R A V E N O U S I N J E C T I O N OF A L C I A N BLUE

Following Alcian blue injections a dense precipitate is found widely distributed in the extracellular space (Fig. IO). A similar precipitate is sometimes seen in some small tubules in endothelial cells. No precipitate is observed inside the tight junctions of these endothelial cells. A more extensive description of these findings will be published elsewhere. Discussion

Bohr and Mollgfird (I974) have described features of developing tight junctions in freezefracture replicas of early human foetal choroid plexus (6o-I5O mm CRL) namely, blindly ending strands, isolated strands and wide loops of strands. These observations are confirmed in the present study. However the junctions appear well enough developed to be tight to protein and probably also to smaller molecules. The endothelial cell junctions in early foetal brain seem to be even more mature and complex since, in contrast to the choroidal epithelial cells in the same feotus, none of the aforementioned features of developing tight junctions were seen associated with the endothelial cells. The freeze-fracture appearance of endothelial tight junctions was similar in the I25 day sheep foetus (which is relatively mature) and in the IO week human foetus (which is very immature); this suggests that endothelial cell tight junctions develop very early in foetal life. The freeze-fracture appearance of cerebral endothelial cell tight junctions in adult mice has recently been briefly described (Connell and Mercer, 1974), but no comments were made concerning strand number or completeness of the junction; so far no data are available about these important features in adult cerebral capillaries. Delorme (I972) has described the development of tight junctions in chick embryo telencephalic endothelial cells. The junctions were said to be completely developed as early as the Ioth day of incubation. However these observations involved only the examination of thin sections and in some experiments (Delorme, Gayet and Grignon, 197o) intravascular horseradish peroxidase. There is therefore some doubt of the significance of these observations since, as pointed out by Revel, Yip and Yang (I973), it is very difficult to distinguish gap junctions and tight junctions in sectioned material whereas the freezefracture technique allows an almost unequivocal identification of junctional types. Fig. 8. Freeze-fracture replica of a capillary from the upper part of the cortical plate of a human foetus (CRL i41 mm). The fracture plane has exposed a tight junction along a considerable length of the vessel (straight arrows). Note that the junctional membrane is fractured also in the lower left corner of the picture. The complex junctional meshwork seems to form a continuous band (curved arrows). >< 35 ooo.

464

MOLLG,~RD and

SAUNDERS

Claude and Goodenough (r973) have classified the tight junctions of epithelial tissues on the basis of the number of strands present. If we can extrapolate to endothelial cell junctions, then the endothelial cell junctions of even the earliest foetal brain vessels examined would be classified as 'tight' to 'very tight' tight junctions. These constitute an absolute barrier to protein penetration and probably also do not permit any appreciable passive ion diffusion (Fr6mter and Diamond, I97Z). There are numerous published estimations of the level of protein in c.s.f, of human newborn infants (e.g. Schultze and Heremans, I966). In most cases the protein concentration was higher than in adults and some authors have attributed this raised level to a leak of protein from plasma into c.s.f. (e.g. Arnhold and Zetterstr6m, i958). We have been able to find only one reference to c.s.f, protein in the human foetus by Klosovskii (1963) who reports the finding of Purin (I958) that in human foetal c.s.f, as early as eight weeks gestational age the protein level was some twenty times higher than that of the adult. Klosovskii also attributed this raised level to an increased permeability of the cerebral vessels to plasma proteins. We are able to confirm that the level of protein in foetal c.s.f, is indeed very high in immature foetuses: ten to twenty times the adult level at about one third of the way through gestation in sheep and pigs (Durbin, Dziegielewska, Evans, Malinowska, Mollg~rd, Reynolds, Reynolds, Reynolds and Saunders, I975). This high level of protein is present inthe sheep foetus at a time when the tight iunctions of the choroid plexus have a freeze-fracture appearance which would suggest that they would prevent penetration of protein from plasma to c.s.f. T h e stage of brain maturity of the least mature human foetus studied (60 m m crown-rump length) is probably equivalent to a sheep foetus of about 40 days gestation (see Astr6m, I967). Thus the presence of very tight junctions in the endothelial cells of human foetuses of a still earlier stage than that at which we find a high level of protein in the foetal sheep c.s.f, would suggest that this high level of protein cannot have come from the plasma via the intercellular pathway between the endothelial cells. This agrees with the finding of Gri3ntoft (I954) that the dye Trypan blue did not penetrate from the cerebral blood vessels into the brain of very immature human foetuses, providing that the dye was given within IO rain of clamping the umbilical cord. At longer times after cord clamping the dye did penetrate into the brain. Probably this result was due to a change in vascular permeability of Trypan blue associated with either hypoxia or more likely hypercapnia (cf. Cameron, Davson and Segal, I969), both of which would develop after cord clamping. Since neither dye nor protein appears to be able to penetrate into brain and c.s.f, via an intercellular route, the high c.s.f. protein level in the foetus cannot be explained by an intercellular leak of plasma proteins into the c.s.f. Alcian blue (MW I39o) was probably protein bound under the conditions used in the present study (Mollg~rd and Sorensen, I974). The finding of a rapid penetration of Alcian blue into brain extracellular space and its presence in an intracellular tubular system in choroidal epithelial and in cortical endothelial cells suggests that some of the protein in c.s.f, may come from the plasma via an intracellular route. This is supported by preliminary results of immunoelectrophoretic studies of the proteins in foetal and adult sheep c.s.f, and plasma. T h e c.s.f./plasma ratios of the proteins present in both fluids do not appear to be what would be expected from passive transfer (Feldman, Platt, Saunders and Soutter, unpublished observations). If a mechanism for transcellular transport of protein does exist

Foetal brain epithelial and endothelial cell tight junctions

465

Fig. 9. Freeze-fracture replica of a capillary from the upper part of the cortex of a I25 day sheep foetus. The short axis (luminal and abluminal) of an endothelial cell (E) tight junction (TJ) is exposed. Complete strands can be followed (~), but no blindly-ending strands are connected to the abluminal part of the tight junction. • 47 ooo.

in foetal brain, then early development of tight junctions might be important in allowing such a mechanism to operate selectively rather than as a non-selective passive diffusion pathway which would occur between cells in the absence of tight junctions. A transcellular mechanism appears to be involved in transport of protein across the gut. Wissig and Graney (1968) and Knutton, Limbrick and Robertson (1974) have described a continuous network of membranelimited tubules originating at the base of microvilli which appears to be involved in the transport of protein in the cells lining the ileum of suckling rats. Another possibility is that some of the foetal c.s.f, protein comes from the brain side of the blood-brain barrier and may be of functional significance in the development of the nervous system. A blood-brain barrier mechanism which has been demonstrated to develop in the sheep foetus is that which limits the entry of sucrose from blood into brain and c.s.f. (Evans et aL,

466

MOLLGARD and

SAUNDERS

I974). Labelled sucrose penetrates from blood into brain at a rate which is at least ten times faster in immature, 5o day foetuses than in more mature I23 day foetuses. Much of the marked restriction on sucrose penetration which develops, does so between 5~ and 7~ days gestation. Given the tight appearance of the intercellular junctions in both capillary endothelium and choroid plexus epithelium, of the immature foetuses described in this paper, it seems unlikely that changes in tight junction structure could account for the observed change in sucrose permeability. The foetal development of ion gradients between c.s.f, and plasma has been described in a number of species (see Bradbury et al., I972 for references). In the sheep foetus there is a gradient for magnesium at least as early as 45 days gestation; over the next IO-I5 days a gradient for chloride develops (Bradbury et al., I972). The presence of a gradient implies that there is some active mechanism, presumably in the choroid plexus epithelium or cerebral vessel endothelium or both, which establishes the gradient. It also implies that there is some restriction on exchange of the ion for which there is a gradient, between c.s.f, and plasma. The appearance of a gradient for magnesium very early in gestation correlates well with the similarly early appearance of tight junctions in choroid plexus epithelium and capillary endothelium. The later development of gradients to other ions, for example chloride and still later calcium (see Bradbury et al., 1972) suggests that the successive appearance of gradients for different ions is due to the establishment of effective ion pumps between c.s.f. and plasma at different times during gestation. The present morphological findings strongly support the idea that an important barrier to diffusion of materials between the blood and the brain is situated at the level of the endothelium. This has previously been suggested on the basis of both physiological experiments (e.g. Crone, 1965; Davson, I967) and morphological observations (e.g. Reese and Karnovsky, I967; Brightman and Reese, 1969). This paper thus presents preliminary morphological evidence that tight junctions in both choroidal epithelial cells and in brain endothelial cells develop so early in gestation that lack of development of tight junctions in the foetal brain is most unlikely to account for the few established cases of immature blood-brain barrier mechanisms in the foetus. Finally, some preliminary evidence is given for intracellular transport of protein via the endoplasmic reticulum in brain endothelial and choroidal epithelial cells.

Acknowledgements The authors wish to thank Mr Bjarne Lauritzen for his expert technical assistance in the freeze-fracture experiments. We should like to thank our colleagues Mr C. A. N. Evans, Dr J. M. Reynolds and Mrs M. L. Reynolds for assistance in obtaining the foetal sheep specimens used in this study, and Dr M. P. del Cerro for his advice concerning the fixation of foetal animal brains. We would also like to thank Professor C. Crone, Dr H. Davson, Professor Sir Andrew Huxley and Professor H. H. Ussing for their helpful comments upon the manuscript. This work was supported by grants from the MRC (UK) and by Grant No. 512-1326 from Statins laegevidenskabelige Forskningsrad.

F o e t a l b r a i n e p i t h e l i a l a n d e n d o t h e l i a l cell t i g h t j u n c t i o n s

467

References ARNHOLD, R. G. and Z E T T E R S T R O M , R. (1958) Proteins in the c.s.f, in newborn. An electrophoretic study including haemolyfic disease of the newborn. Pediatrics 2I, 279-87. .~STRC)M, K.-E. (1967) On the early development of the isocortex in fetal sheep. Progress in Brain Research 26, 1-59. BAKAY, L. (1953) Studies on blood-brain barrier with radioactive phosphorus. I I I . Embryonic development of the barrier. Archives of Neurology and Psychiatry 7o, 30-9. BARCROFT, J. (1938) The Brain and Its Environment. p. I. New Haven: Yale University Press. BEHNSEN, G. (1927) (Sber die Farbstoffspeicherung im Zentralnervensystem der Weissen Maus in verschiedenen Alterszustiinden. Zeitschrift f~r Zellforschung und mikrosk opische Anatomic 4, 515-72. BOHR, V. and MOLLGARD, K. (1974) Tight junctions in human fetal choroid plexus visualized by freeze-etching. Brain Research 81, 314-18. B R A D B U R Y , M. W. B., C R O W D E R , J., DESAI, S., R E Y N O L D S , J. M., REYNOLDS~ M. and SAUNDERS, N. R. (1972) Electrolytes and water in the brain and cerebrospinal fluid of the foetal sheep and guinea-pig. Journal of Physiology 227, 591-61o. BRANTON, D. (1966) Fracture faces of frozen membranes. Proceedings of the National Academy of Sciences (Washington) 55, 1048-56B R I G H T M A N , M. W. and REESE, T. S. (1969) Junctions between intimately apposed cell membranes in the vertebrate brain. Journal of Cell Biology 4o, 648-77. C A M E R O N , I. R., DAVSON, H. and SEGAL, M. B. (1969) The effect of hypercapnia on the blood-brain barrier to sucrose in the rabbit. Yale Journal of Biology and Medicine 42, 241-7. CLAUDE, P. and GOODENOUGH, D. (1973) Fracture faces of zonulae occludentes from 'tight' and 'leaky' epithelia. Journal of Cell Biology 58, 39o-4oo. CONNELL, C. J. and MERCER, K. L. (1974) Freeze-fracture appearance of the capillary endothelium in the cerebral cortex of mouse brain. American Journal of Anatomy I4O , 595-9. CRONE, C. (1965) T h e permeability of brain capillaries to non-electrolytes. Acta Physiologica Scandinavica 64, 4o7-17 . D AVSO N, H. (I 967) Physiology of the Cerebrospinal Fluid. London: Churchill. DELORME, F. (1972) Diff6renciation ultrastructurale des jonctions intercellulaires de l'endothelium des capillaires t616ncephaliques chez l'embryon de poulet. Zeitschrift fi~r Zellforschung und mikroskopische Anatomie 133 , 571-82. DELORME, P., GAYET, J. and GRIGNON, ~. (1970) Ultrastructural study on transcapillary exchanges in the developing telencephalon of the chicken. Brain Research 22, 269-83. D U R B I N , G. M., D Z I E G I E L E W S K A , K. M.~ EVANS, C. A. N., M A L I N O W S K A , D. H., M O L L G A R D , K,, R E Y N O L D S , E. O. R., R E Y N O L D S , J. M., REYNOLDS~ M. L. a n d SAUNDERS, N. R. (1975) Penetration

of labelled protein from blood into foetal brain and c.s.f. Proceedings of the Physiological Society, Copenhagen (April) J . Physiol., in press. EHRLICH, P. (1885) Das Sauerstoff Bediirfnis des Organismus. Eine fabenanalytische Studie, pp. 69-72, Berlin: Hirschwald. EVANS, C. A. N., R E Y N O L D S , J. M., R E Y N O L D S , M. L.~ SAUNDERS~ N. R. and SEGAL, M. B. (1974) T h e development of a blood-brain barrier mechanism in foetal sheep. Journal of Physiology 238, 371-86. FRIEND, D. S. and GILULA, N. B. (1972) Variations in tight and gap junctions in mammalian tissues. Journal of Cell Biology 53, 758-76FROMTER, E. and DIAMOND, I. (1972) Route of passive ion permeation in epithelia. Nature New Biology 235, 9-13. GRONTOFT, O. (1954) Intracranial haemorrhage and blood-brain barrier problems in the newborn. Acta Pathologica et Microbiologica Scandinavica Supplementum C pp. 1-1o2. I~REUTZlGER, G. O. (1968) Freeze-etching of intercellular junctions of mouse liver. I n Proceedings of the 26th Meeting of the Electronmicroscopical Society of America pp. 234-235. Baton Rouge: La Claitor's Publishing Division.

468

MOLLGARD and

SAUNDERS

KLOSOVSKII, B. N. (1963) The Development of the Brain and its Disturbance by Harmful Factors (Translated by Haigh, B.), p. 8. Oxford: Pergamon. LEE, ~. C. (I97I) Evolution in the concept of the blood-brain barrier phenomenon. Progress in Neuropathology I, 84-I45. K N U T T O N , S., L I M B R I C K , A. R. and ROBERTSON~ )'. D. (1974) Regular structures in membranes. I. Membranes in the endocytic complex of ileal epithelial cells. Journal of Cell Biology 62, 679-94. MOOR, H. and MUHLETHALER, K. (1968) Fine structure in frozen-etched yeast cells. Journal of Cell Biology 17, 6o9-28. MOLLGARD, K. and SORENSEN, S. C. (1974) The permeability of cerebral capillaries to a tracer molecule, Alcian blue, with a molecular weight of 139o. I n The Pathology of Cerebral Microcirculation (edited by j. CERVOS-NAVARRO), pp. 119-121 , Berlin: Walter de Gruyter and Co. NAATANEN, E. (1947) Ueber die Speicherung der intraven6s injuzierten Methylenblaul6sung in Zentralnervensystem des menschlichen Fetus. Acta Societatis medicorum fennicae 'Duodecim' 24, I--II.

PINTO DA SILVA, P. and BRANTON~ D. (197 O) Membrane splitting in freeze-etching. JournalofCell Biology 45, 598-6~ REESE, T. A. and KARNOVSKY, M. ]'. (1967) Fine structural localization of a blood-brain barrier to exogenous peroxidase. Journal of Cell Biology 34, 2o7-17 9 REVEL, J. P., YIP, P. and CHANG, L L. (1973) Cell junctions in the early chick e m b r y o - - A freeze etch study. Developmental Biology 35, 3o2-17. SCHULTZE, H. E. and HEREMANS, J. F. (1966) Molecular Biology of Human Proteins, with special reference to plasma proteins. 17ol. 1, Nature and metabolism of extracellular proteins. Amsterdam: Elsevier. STAEHELIN, L. A. (1973) Further observations on the fine structure of freeze-cleaved tight junctions. Journal of Cell Science I3, 763-86. TSCHIR6I, R. D. (1950) Protein complexes and the impermeability of the blood-brain barrier to dyes. AmericanJournal of Physiology I63, 756P. WlSSI6, s. L. and 6RANEY, D. O. (I968) Membrane modifications in the apical endocytic complex of ileal epithelial cells. Journal of Cell Biology 39, 564-79.