Cell Tissue Res (1994) 276:387 397

Cell&Tissue Research 9 Springer-Verlag1994

Membrane dynamics during migration of placental cells through trophectodermal tight junctions in sheep and goats F.B.P. Wooding 1, G. Morgan 1, M.R. Brandon 2, S. Camous 3 1AFRC Babraham Institute, Department of Cellular Physiology, Babraham, Cambridge CB2 4AT, UK 2 Centre for Animal Biotechnology, Melbourne University, Australia 3 Station de Physiologie Animale, INRA, F-78352 Jouy en Josas, Cedex, France Received: 26 August 1993 /Accepted: 28 September 1993

Abstract. Binucleate cells in ruminant trophectodermal epithelium are unique in that they form part of the tight junction as they migrate across it, maintaining the ionic barrier seal to the internal milieu of the fetus. Such participation imposes considerable constraints on the cell migration because membrane cannot flow through a tight junction. We report quantitative ultrastructural immunocytochemical evidence for vesicle membrane insertion into the binucleate cell plasmalemma which allows the cells to form a pseudopodium past the tight junction. This pseudopodium increases continuously in area by vesicle insertion and develops a close apposition to the plasmalemma of the fetomaternal syncytium which constitutes the fetomaternal boundary in the placenta of the sheep and goat. Eventually the apposed membranes of the binucleate cell pseudopodium and the syncytium fuse by vesiculation and the cytoplasm and nuclei of the binucleate cell merge into the fetomaternal syncytium. The binucleate cell plasmalemma remaining on the trophectodermal side of the tight junction is blebbed off into, and phagocytosed by, the uninucleate trophectoderreal cells between which the binucleate cell passed. This process permits the delivery of the binucleate cell granules to the maternal side of the placenta but none of the fetal molecules expressed on the plasma membrane of the binucleate cells are exposed to potential maternal immunological rejection. Key words: Placenta - Trophectoderm - Binucleate cells - Tight junctions - Cell migration - Sheep Goats

Introduction Epithelial cells are sealed together at their apices by a tight junction which limits paracellular transport to a variable degree depending on the number and pattern of the junctional strands (Claude and Goodenough 1973; Madara, 1989; Mandel et al. 1993). In the sheep, the Correspondence to: F.B.P. Wooding

structure of the tight junction in the fetal placental trophectodermal epithelium is complex, indicating a very restricted paracellular flow (Morgan and Wooding 1983). However, throughout pregnancy the characteristic binucleate cells which develop in the ruminant trophectoderm have been shown to migrate through the tight junction while maintaining its structure (Wooding 1982, 1992). These are the only cells known to the authors which have been shown to migrate through a tight junction in this way when traversing an epithelium. Studies of white blood cells crossing the endothelium, by far the most common transepithelial cell traffic migration, indicate that the blood cells are more likely to burrow through the endothelial cell body (Cho and DeBruyn 1981, 1986) or pass through discontinuities in the endothelial tight junction (Bundgaard 1984; Cramer 1991). There is no direct pictorial evidence for a white blood cell forming one half of a tight junction or opening the usually discontinuous junction between endothelial cells. This unique ability of the trophectodermal binucleate cell, when mature, to form a tight junction with neighbouring uninucleate cells ensures maintenance of the internal milieu of the fetus while allowing the content of the characteristic binucleate cell granules to pass to the maternal side of the placenta. However, the presence of such a junction imposes considerable constraints on the cell migration since the plasmamembrane of a cell cannot flow through a tight junction (VanMeer and Simons 1986): the binucleate cell is anchored by the structure: This paper reports a quantitative ultrastructural immunocytochemical study which was designed to investigate the interrelationships of the membranes of the binucleate cell during this migration. A preliminary account of the results was presented at a meeting of the Anatomical Society (Wooding and Brandon 1993).

Materials and methods At defined stages of pregnancy Clun sheep (29, 41, 44, 70, 114, 140 and 142 days post coitum (d.p.c.)) and Saanen goats (90, 95 and 130

388 days) were killed with an overdose of sodium pentobarbitone. The uterus was removed immediately, the uterine arteries cannulated and perfused at room temperature with 1% (w/v) glutaraldehyde plus 3 % (w/v) paraformaldehyde in 0.1 M sodium cacodylate buffer, pH 7.2, containing 5% (w/v) sucrose. Five minutes after perfusion was started, fixative was also injected into the uterine lumen or perfused down the umbilical arteries. The placentomes were then cut into thin "matchsticks" each of which ran the full depth from fetal to maternal stroma. The matchsticks were then placed in fresh fixative for a further period of i h and stored in cacodylate buffer at 4~ Material from each animal was processed for structural studies through osmium and ethanol into Araldite and for immunocytochemistry through ethanol into Araldite, glycol methacrylate, or Lowicryl K4 M resin, at -20~ (Wells 1985). Thin sections of both nonosmicated Araldite and K4 M blocks were picked up on naked 300 mesh nickel grids and the immunogold procedure used to localize the SBU-3, PSP-60 or placental lactogen antigens. The antibodies to these antigens have been characterized previously, (SBU-3; Gogolin-Ewens et al. 1986: ovine placental lactogen; Chan et al. 1976: PSP-60; Camous et al. 1991). Grids were floated on drops of reagents on parafilm at room temperature. The initial incubation was on 0.1 M phosphate buffer, pH 7.4, plus 1% bovine serum albumin and 0.05% Thimerosal (DBSBT buffer) which was used for all dilutions of antibody and gold reagents. Grids were transferred to drops of mouse monoclonal SBU-3 antibody or nonimmune serum at 1:10 or 1:200 dilution and incubated in a wet box at 4~ for 16-18 h, jetwashed with DBS, placed on a drop of rabbit antimouse IgG (Dako, High Wycombe, UK) diluted 1 : 1000 for 30 min, washed again with DBS and then floated on 10 nm gold colloid which had been coated with goat anti-rabbit IgG (Biocell, Cardiff, UK) diluted 1 : 25 for 30 min. Polyclonal rabbit anti PL and PSP-60 were used in a similar way without the rabbit antimouse IgG step. After immunostaining the grids were washed with DBS and glass distilled water then stained with saturated uranyl acetate in 50% (v/v) ethanol and lead citrate, or with 1% (w/v) phosphotungstic acid (PTA) (Wooding et al. 1980). The sections were examined by use of a JEOL 100 C microscope operated at 80 kV. The frequency of gold particles per lam of plasmalemma was determined by counting in the electron microscope. A wire circle of known diameter was used as a length marker and was superimposed over the plasmalemma at one start point, for example one tight junction at the end of the binucleate cell migration front. The number of gold particles touching or within 20 nm of the membrane trace was then counted. The image was then shifted along one diameter and the next count made. This procedure was then repeated until the end point, in this case the other tight junction was reached. This process provides the number of diameters between start and finish and the total number of gold particles counted. Given the magnification the number of gold particles per gm can then be calculated. This method eliminates the need for numerous micrographs and works well for straight or gently curving membranes but not for the microvillar junction which is too complex. Here the membrane length and number of gold particles were measured on micrographs using a digital pad tracer. For freeze-fracturing, the glutaraldehyde fixed material was soaked in 10% glycerol in phosphate-buffered saline (PBS) for 2 h and immersed in 25% glycerol in PBS overnight prior to placing on gold support stubs and freezing in Isceon 22 (monochlorodifluoromethane) cooled by liquid N2. The frozen specimens were stored in liquid N 2 until they were freeze-fractured in Balzers B.A.F. 300 and replicated without etching with platinum and carbon at - 100~ and 104 Torr. The replicas were cleaned by immersion in 35% chromic acid overnight, rinsed in water, and then transferred to bleach for 2 h. The replica fragments were finally rinsed thoroughly in glass-distilled water and mounted on carbon-coated, celloidin-covered copper slotted (2 x 1 mm) discs.

Results

Freeze-fracture studies d e m o n s t r a t e that the binucleate cell forms part of the t r o p h e c t o d e r m a l tight j u n c t i o n when migrating across the microvillar j u n c t i o n (Fig. 1 and see sketch on Table 1). G o l d labelling is m o s t easily seen and c o u n t e d on P T A - s t a i n e d n o n - o s m i c a t e d material (Figs. 2-10). This produces a selective staining emphasizing the m e m b r a n e s of the microvillar junction, the binucleate cell Golgi transcisternae and the granules they p r o d u c e (Fig. 2). N u clei and connective tissue are lightly stained; bulk cytoplasm is sufficiently contrasted to allow recognition of cell boundaries for the p u r p o s e of gold label counting (Fig. 2). All sheep and goat placentomes examined from 4 1 140 d.p.c, showed preferential localisation of the SBU-3 a n t i b o d y to that area of the binucleate cell plasma m e m brane past the tight j u n c t i o n and a p p o s e d to the fetomaternal syncytial layer (Figs. 3-6). This is subsequently referred to as the binucleate cell migration front. Labelling of the Oolgi b o d y and granules was optimal at 1/1001/200 dilution of SBU-3 but to p r o d u c e consistent labelling of the migration front the a n t i b o d y h a d to be used at 1/10 dilution. W h e n used at this dilution the migrating binucleate cells often could be separated into two p o p u lations, one with a m u c h higher label on the migrating front than the other. H o w e v e r this high label was coupled with a correspondingly high level on one or m o r e of the other m e m b r a n e profiles. The migrating front normally showed five to seven times m o r e label than any other m e m b r a n e (Table 1; Figs. 3-9). As we have reported previously the a n t i b o d y is also localised over the characteristic binucleate cell granules and the cisternae of the Golgi b o d y from which they are p r o d u c e d (Figs. 3, 7, 9). The localisation on the binucleate cell migration front can be d e m o n s t r a t e d on Araldite, glycolmethacrylate and Lowicryl K 4 M sections with highest label on K 4 M but also the highest b a c k g r o u n d . I m m u n o d e t e c t i o n is not sensitive to the concentration (between 1 and 4%) of glutaraldehyde or formaldehyde e m p l o y e d for the initial perfusion but use of o s m i u m completely abolishes all SBU-3 reactivity. Other antibodies which label the binucleate cell granules and Golgi b o d y such as rabbit antiovine placental lactogen [PL] or rabbit antibovine p r e g n a n c y serum protein (PSP-60) (Figs. 7 a, 9 a) do not label the migration front (Fig. 8 a). Treatment of n o n o s m i c a t e d sections with p h o s p h o tungstic acid (PTA) identifies a p o p u l a t i o n of preferentially stained small vesicles and tubules which also label with SBU-3 a n t i b o d y but not with placental lactogen or PSP-60 antibodies (Figs. 3-9). These labelled m e m b r a n e s are occasionally closely associated with the Golgi apparatus but more predictably with the migration front. Skeins of these labelled vesicles and tubules sometimes bridge the gap between the P T A stained Golgi cisternae and the migration front of the binucleate cell (Fig. 9 b). D u r i n g the early part of the binucleate cell migration t h r o u g h the t r o p h e c t o d e r m a l tight junction the Golgi

389

Fig. 1. Freeze-fracture image of a binucleate cell (B) partway through the tight junction (arrows) it forms with adjacent uninucleate cells in the trophectodermal epithelium (7). Binucleate cells are readily identified by their size, characteristic granule content and, when migrating, by the absence of microvilli on the migration front

(below arrows) Note the continuity and complexity of the broad band of tight junction ridges (arrows). This cell is at an equivalent stage of migration to the larger binucleate cell (B2) in Fig. 2. S Syncytium. Ewe, 110 days pregnant, x 11 000. Bar: 1 gm

390 Table 1. Intensity of immunogold labelling for SBU-3 antigen (gold particles per Bm transected membrane) over the plasmalemma of migrating trophectodermal binucleate and other cells at the fetomaternal interface of the placenta Days pregnant

Plasmalemma region (see sketch below) A

29 ewe 41 ewe 90 goat 114 ewe 130 goat 141 ewe 142 ewe

High Low High Low High Low

142 ewe

8• 12 34• 10• 29• 9• 10• 13• 14• 5• 17•

2 4 6 3 4 4 4 2 5

B

C

D

E

F

G

1• 1 1• 2• 4• 1• 2• 1• 2• 1• 2•

1• 4 6• 2• 2• 2• 2• 3• 2• 1• 2•

2• 4 6• 2• 1• 2• 2• 1• 5• 1• 2•

1• 1 2• 2• 3• l•

1• 2 1• 1• 2• 1•

1• 1 1• 1• 1• 1•

1•

1•

1•

Number of binucleate cells assessed 4 2 6 5 16 4 9 4 6 6 5

1•

Plastic type

A A E E A A,E A E A A E

Frequencies are given • SEM; A, acrylic (K4M, GMA); E, epoxy Labelling for other antibodies such as anti PSP-60 or anti ovine placental lactogen was never more than 2_+ 1 over any region

FETAL

' ] [ ,~

i k

~/'/~ |

]

~

//o

~'

o ,,\

|

!

!-

I

r

FETAL TROPHECTODERM

I

MATERNAL body is still producing characteristic large dense granules from the trans-Golgi network cisternae (which label with PL, PSP-60 and SBU-3 (Figs. 3, 7 b, 9 b), and these are clearly quite different from the small labelled vesicles (which only label with SBU-3) (Figs. 3-9). Some vesicles when cut in cross section indicate that the label is associated with the membrane but most are too small to show this (Figs. 4, 5). Since the matrix of the binucleate cell granules is so heavily labelled it is not possible to be certain that their membrane is specifically marked; most profiles suggest that it is not. Similarly sized PTA-stained small vesicles are often seen in uninucleate trophectodermal cells or syncytium close to the microvillar junction, such vesicles are SBU-3 negative (Figs. 3, 4). Label on the migration front is clearly localised to the plasma membrane, and stops abruptly at the tight junction (Figs. 3-6). Label on the binucleate cell membrane

below the tight junction and on other plasma membranes of the trophectoderm and syncytium rarely exceeds background levels (Table 1; Figs. 3-6). When the microvillous junction between trophectoderm and syncytium is art• separated during preparation label is seen only on the binucleate cell migration front and is absent from the area of plasmalemma of the syncytium to which the front was apposed (results not shown). Treatment of the sections with a saturated solution of sodium metaperiodate prior to immunocytochemistry abolishes all SBU-3 immunoreactivity, but has no effect on placental lactogen or PSP-60 labelling. Once a broad migration front has been established it consists of binucleate cell plasmalemma closely apposed to the plasmalemma of the syncytium. The two membranes are separated uniformly by a 15-20 nm gap similar to that which separates the membranes of the mi-

391

Fig. 2. Glycol methacrylate (GMA) section of nonosmicated placentome stained with phosphotungstic acid (PTA). The microvillar junction (J) between fetal trophectoderm (7} and fetomaternal syncytium (S), the binucleate cell granules (arrowheads) and Golgi bodies (G) are densely stained as are lysosomes (L) in the uninucleate trophectodermaI cells. The nuclei (N) and maternal connective tis-

sue (asterisks) are also visible. The young binucleate cells (B1) develop within the uninucleate trophectoderm and when mature (i.e., fully granulated, B2) migrate through the trophectodermal tight junction (position indicated by the arrows) by forming a migration front (A) past that tight junction. Ewe, 114 days pregnant, x 4 800. Bar: 1 gm

crovillar junction (Figs. 14, 15). The apposed membranes then fuse together locally initially producing discontinuous gaps in the migration front (Fig. 10) and eventually disappear altogether. The contents of the binucleate cell merge into the syncytiotrophoblast cytoplasm and the granules translocate to the area close to the maternal connective tissue. This leaves what was originally binucleate cell plasmalemma as a tight junction bounded patch of membrane in the syncytiotrophoblast plasmalemma which forms the microvillar junction (Fig. 12). This patch of membrane, including the tight junction, is then phagocytosed by the apposed trophectoderm cells, sometimes producing bizarre scroll-like membrane formations (Figs. 11, 12). The microvillar junction reforms between the syncytiotrophoblast plasmalemma and trophectoderm as the patch is removed. Very little ultrastructural detail can be visualised in

the cytoplasm behind the migration front of the binucleate cell after the glutaraldehyde-acrylic resin-phosphotungstic acid sequence of section preparation. This is misleading as examination of equivalent sections processed by the conventional glutaraldehyde-osmium-Araldite resin-uranyl and lead sequence demonstrates (Figs. 13 15). There is normally a centriole and a full complement of mitochondria, rough endoplasmic reticulum and cytoplasmic filaments as well as the Golgi body granules and vesicles which are the only organelles outlined by phosphotungstic acid (Figs. 13, 14). Several sorts of small vesicles are seen in osmicated material but there seems to be a concentration close to the migrating front of one clear round type (Figs. 13-15). Such distributions are equivalent to the PTA staining vesicles positive for SBU-3. As yet we have no ultrastructural evidence for budding of such vesicles from the Golgi,

392

Figs. 3--6. GMA-PTA sections labelled with SBU3 showing the increase in area of the migration front (A) between migrating binucleate cells (B) and the syncytium (S). The tight junction at the edge of the migration front is not distinguishable at these magnifications, so its position is indicated by the small double arrowheads. The boundary between uninucleate and binncleate cells is also faint and is shown by single small arrowheads. Note that the microvillar junction (J) immediately adjacent to the migration front is not significantly labelled, nor are the small vesicles (small arrows) in the uninucleate trophectodermal cells (7). On 3 the binucleate cell (B) Golgi body (G) trans cisternae are heavily stained and gold labelled for SBU-3 as are the granules (white asterisks) which form from there, There is also a population of small labelled vesicles (some with predominantly peripheral label (large arrowheads) in the cytoplasm, mostly concentrated at the binucleate cell migration front (A) which is also labelled. Some of the small labelled vesicles on 4 and 6 are in aggregates (large arrows). Ewe 114, days pregnant. 3 x 12 000, 4 x 13 800, 5 x 12000,6 x 13000. Bar: l g m

and they are frequent on both trans and cis sides of that organelle. After p r o l o n g e d o s m i u m fixation (more than 48 h) characteristic aggregates of similar vesicles and tubules are revealed close to the migrating front in positions very similar to those of the P T A stained m e m b r a n e s (Fig. 14).

Discussion This quantitative ultrastructural i m m u n o c y t o c h e m i c a l study clearly shows that new m e m b r a n e is used for the formation and expansion of the migrating front of the sheep and goat fetal binucleate cell as the cell contents pass t h r o u g h the t r o p h e c t o d e r m a l tight junction. A

Figs. 7-9. Adjacent (a,b) G M A : PTA sections of binucleate cells (B), 7 a, 8 a and 9 a labelled with PSP-60 antibody and 7 b, 8 b and 9 b labelled with SBU-3 antibody. Both antibodies label the Golgi cisternae (G) and granules (7 and 9) but the small tubules and vesicles in the cytoplasm are labelled only by SBU-3. The migration front (A, 8) is also labelled only by SBU-3. S Syncytium. Ewe, 114 days pregnant. 7 x 24 000, bar: 0.5 gm, 8 x 19000, bar:0.5~tm;9 x 10 100, bar: l~tm

394

Fig. 10. Unlabelled Lowicryl K4M-PTA section of a binucleate cell (B) half way across the microvillar junction (J). The migration front between binucleate cell and syncytium (S) is discontinuous at the arrows. The plasma membranes between the binucleate cell cytoplasm and trophectoderm (7) are faint (small arrowheads) but continuous, limited at both ends by the tight junction (double small arrowheads). Ewe 70 days pregnant, x 6 700. Bar: 1 gm

source of new m e m b r a n e is necessary because the tight junction which seals the uninucleate trophectodermal epithelium has to be maintained to protect the internal milieu of the fetus from the influences of the 'foreign' maternal environment. Our previous studies have shown that in all ruminant placentas examined all of the characteristic trophectodermal binucleate cells are destined to migrate past the junction to fuse with maternal uterine epithelial cell derivatives to form fetomaternal hybrid tissue throughout pregnancy (Reviewed in Wooding 1982, 1992; Wooding et al. 1993). Since the junction is a membrane anchorage site across which little or no m e m b r a n e or molecular flow can occur (VanMeer and Simons 1986), considerable constraints are imposed on the mechanisms of this cell migration. Use of the SBU-3 antibody demonstrates that binucleate cells produce a population of small vesicles and tubules which fuse only into the plasm a l e m m a past the tight junction once the cells have become part of the tight junction. This allows the formation of a continuously expanding pseudopodium (defined as the migration front) apposed to and indenting the maternal syncytiotrophoblast. This insertion of new m e m b r a n e into the migration front is analogous to the general system suggested for cell migration in unconstrained situations. Observations in individual living keratinocytes (Kucik et al. 1990) and immunofluorescence studies on fibroblasts (Singer and Kupfer 1986) indicate that there is little or no m e m b r a n e flow to allow directed m o v e m e n t but rather an insertion of new m e m b r a n e at the migration front or pseudopodium controlled by cytoskeletal reorganisation. Such a reorganisation is also seen in the binucleate cell. The centriole and Golgi apparatus are relocated to a position between the nucleus and the migrating front. Once the migrating front has formed past the tight junction it fre-

Figs. ll, 12. Unlabelled sections (11 glutaraldehyde-osmiumAraldite, 12 GMA-PTA) of recently migrated binucleate cells (B). Whorls and blebs of residual plasma membrane (asterisks) are being taken up by the trophectoderm (7). The position of the original

binucleate-uninucleate tight junction, which is continuous with the microvillar junction (J), is indicated by the double arrowheads. 11 Ewe 29 days pregnant, x 4 600. Bar: 1 gm. 12 Ewe 114 days pregnant, x 10 100. Bar: 0.5 gm

395

Figs. 13, 14. Unlabelled glutaraldehyde, osmium, araldite (G,O,A) sections of tissue from the same cotyledon as used for Figs. 2-6. This shows the wealth of ultrastructural detail present in this material, but invisible on GMA-PTA sections. The binucleate cell (B) migration front (A) is bounded by the tight junction (double small arrowheads in 13) and has numerous small clear vesicles (small arrows) and tubules close to it, occasionally in aggregates (curved

arrows in 14). The binucleate cell Golgi body (G) and centriole (open arrows) are characteristically located between the nucleus and the migration front (A). There are numerous mitochondria (M) and rough endoplasmic reticulum cisternae (R) in the binucleate cell cytoplasm. S Syncytium. Ewe 114 days pregnant. 13 x 12 000. Bar: lgm. 14 x 16 600. Bar: 0.5gm

Fig. 15. High magnification of a binucleate cell (B) migration front on a G, O, A section. The small clear vesicles (1O are usually separated from the plasmalemma (arrowheads) by filamentous material (F), but some (small arrow) appear close enough for imminent fusion. S Syncytium. Ewe 105 days pregnant, x 74 000. Bar: O.1 ~m quently develops the shape and characteristic fibrillar content (after conventional fixation) of a typical pseud o p o d i u m (Fig 3, W o o d i n g et al. 1980; Fig. 10, W o o d i n g 1982) and a definite n a r r o w fibrillar zone forms immediately under the p l a s m a l e m m a of the remainder of the

binucleate cell still s u r r o u n d e d by trophectoderm. The vesicle p o p u l a t i o n which forms the migration front in binucleate cells is uniquely identified by its size; its staining with p h o s p h o t u n g s t i c acid (on n o n o s m i c a t e d sections); its SBU-3 i m m u n o c y t o c h e m i c a l label; and its po-

396

sition, associated predominantly with the migration front. The monoclonal antibody SBU-3 was originally raised against a preparation of microvilli from the trophectoderm of the fetal sheep (Gogolin-Ewens et al. 1986). It has previously been used as a specific marker for the contents of the characteristic ruminant binucleate cell granules with no indication of any membrane localisation (Wooding 1992). Under the conditions used here, the SBU-3 antibody labels not only the content of the granules and the Golgi cisternae from which the granules form but also the membrane of the small vesicle population and the plasmalemma of the migrating front. The SBU-3 label on the small vesicles is specific for binucleate cells, since the similarly sized vesicles near the microvillar junction, usually present in adjacent uninucleate cells do not label. Other immunocytochemical markers for the contents of the binucleate cell Golgi cisternae and granules (ovine placental lactogen, bovine pregnancy serum protein-60) label neither the small vesicles nor the migration front. This demonstrates that the small vesicles are not just tiny granules but a unique population with a different protein composition to that of the granules. It also suggests that the SBU-3 antibody is recognising two different molecules, one stored within the granules and one bound to membrane. Since the immunoreactivity of both can be abolished by treatment of the section with periodate oxidation, the epitopes recognised are probably carbohydrate. Characterisation of the antibody (Gogolin-Ewens et al. 1986; Atkinson et al. 1993) indicates that the carbohydrate epitope was present on a series of glycoproteins with a range of molecular weights from 30-200 kD which makes our suggestion of at least two different molecules plausible. This dual labelling makes it impossible to determine the origin of the small vesicles unequivocally. They frequently are seen as if streaming down from the Golgi region to the migrating front, but do not have as clearly defined an origin as the granules budding from the fenestrated trans-Golgi network. The vesicles do stain with phosphotungstic acid which suggests an origin from the trans-Golgi cisternae but there are so many small vesicles in this area that no definite conclusions can be reached concerning their origin. There is an interesting parallel with this dual localisation in human salivary gland secretion (Takano et al. 1991). A monoclonal antibody against a carbohydrate epitope on the secretory granule protein, salivary agglutinin, labels both secretion granules and a small vesicle population plus the plasma membrane. Takano and his colleagues suggest that this small vesicle population may mediate constitutive secretion and consider the plasma membrane label indicates a different protein with the same carbohydrate epitope since a second monoclonal antibody against the protein sequence of agglutinin does not label the plasma membrane. We have no monoclonal antibody available against the peptide sequence of the SBU-3 antigen so we cannot further characterise our putative membrane protein but the binucleate and salivary gland cell systems clearly demonstrate analogous separation of functions between granule and vesicle populations.

The simplicity of the structures visible on sections of phosphotungstic acid-stained nonosmicated material is very misleading. Conventional glutaraldehyde- and osmium-fixed Araldite-embedded preparations of equivalent material show that the region between nucleus and/ or Golgi body and the migrating front has a full complement of mitochondria, endoplasmic reticulum and cytoplasmic filaments as well as numerous small vesicles. Often the filaments exclude most of the other organelles to form a characteristic pseudopodium at the migrating front but the small vesicles can usually be found in the interstices of these filaments or at the edge of the migrating front. Since osmication abolishes all SBU-3 immunoreactivity it is not possible to establish directly which vesicles are equivalent to those immunolabelled with SBU-3 on nonosmicated sections. However a population of small clear vesicles, with no apparent coating, can be recognised on the osmicated sections with the same distribution and size as the immunolabelled small vesicle population; but again no obvious precursor/ product relationship with the Golgi body can be identified. They are however very similar in size and appearance to the small vesicle population suggested by Takano et al. (1991) to carry constitutive membrane proteins to the plasma membrane in salivary gland cells. The labelled migrating front plasmalemma does not persist, it is designed to fuse with the plasmalemma of the fetomaternal syncytium to which it is apposed (Wooding 1992). Judging by the difficulty of finding examples of fusion the process is very rapid and no indication of the fate of the vesiculated membranes has been found. This fusion allows the cell contents of the binucleate cell to merge into the fetomaternal syncytial plaque and the granules to move to the maternal boundary for exocytosis. This leaves the original "rear end" of the binucleate cell bounded by the tight junction (past which the cell contents have now passed) as part of the microvillar junction. The residual membrane and tight junction are then lost by blebbing into and phagocytosis by the trophectoderm. The microvillar junction between the syncytium and the trophectodermal uninucleate cells reforms as the binucleate cell residual "tail" is removed. This ability to pass through, while maintaining the structure of, a complex tight junction appears to be unique. Other cells crossing epithelia either pass through the epithelial cells away from the tight junction (horse placental girdle cells, Allen et al. 1973; lymphocytes, Cho and De Bruyn 1981) or pass through endothelium whose tight junctions are normally discontinuous (Bundgaard 1984) or have become so by inflammation (Simionescu 1980). The process of migration is also quite different in cells migrating out of the epithelium in which they form, which is a closer analogy to the binucleate cell system. In the case of spermatocyte maturation (Russell 1978) or the loss of cells from the gut epithelium, a new tight junction forms below the migrating cell which never forms part of the junction. One advantage of the binucleate cell system is that it allows the trophectoderm to maintain the internal physiological milieu of the fetus independent of the mother. This protects the fetus from possible deleterious immunological and physiological fluctuations in the ma-

397 t e r n a l b l o o d . T h e p r o c e s s also p e r m i t s the delivery of fetal p r o d u c t s of the b i n u c l e a t e cell to the m a t e r n a l side of the p l a c e n t a w i t h o u t e x p o s i n g a n y of the fetal m o l e c u l e s e x p r e s s e d on the p l a s m a m e m b r a n e of t h a t cell to p o t e n t i a l m a t e r n a l i m m u n o l o g i c a l rejection.

References Allen WR, Hamilton DW, Moor RM (1973) The origin of equine endometrial cups II Invasion of the endometrium by trophoblast. Anat Rec 177:485 502 Atkinson YH, Gogolin-Ewens K J, Hounsell EF, Davies M J, Brandon MR, Seamark RF (1993) Characterisation of placentation specific binucleate cell glycoproteins possessing a novel carbohydrate: evidence for a new family of pregnancy associated molecules. J Biol Chem (in press) Bundgaard M (1984) The three dimensional organisation of tight junctions in a capillary endothelium as revealed by serial section electron microscopy. J Ultrastruct Mol Struct Res 88:1-17 Camous S, Coste V, Guillomot M, Martal J (1991) Demonstration of an ovine conceptus protein related to the bovine serum pregnancy protein (PSP-60) of Mr 60000. J Reprod Fertil [Suppl] 43 : 301 Chan JSD, Robertson HA, Friesen HG (1976) The purification and characterisation of ovine placental lactogen. Endocrinology 95:65-75 Cho Y, DeBruyn PPH (1981) Transcellular migration of lymphocytes through the walls of endothelial venules in the lymph node. J Ultrastruct Mol Struct Res 74:259-266 Cho Y, DeBruyn PPH (1986) Internal structure of the postcapillary high endothelial venules of rodent lymph nodes and Peyer's patches and the transendothelial lymphocyte passage. Am J Anat 177:481~490 Claude P, Goodenough DA (1973) Fracture faces of zona occludens from tight and leaky epithelia. J Cell Biol 58:390400 Cramer EB (1991) The ability of leucocytes to cross tight junctions. In: Cereijido M (ed) Tight junctions. CRC Press, Florida, pp 321-336 Gogolin-Ewens K J, Lee CS, Mercer WR, Moseby AM, Brandon MR (1986) Characterisation of a sheep trophoblast derived antigen first appearing at implantation. Placenta 7:243-255

Kucik DF, Elson EL, Sheetz MP (1990) Cell migration does not produce membrane flow. J Cell Biol 111 : 1617-1622 Madara JL (1989) Loosening tight junctions, lessons from the intestine. J Clin Invest 83:1089-1094 Mandel LJ, Bacallao R, Zampighi G (1993) Uncoupling of the molecular fence and paracellular gate functions in epithelial tight junctions. Nature 361:552-555 Morgan G, Wooding FBP (1983) Cell migration in the ruminant placenta. A freeze fracture study. J Ultrastruct Mol Struct Res 83:148-160 Russell LD (1978) The blood-testis barrier and its formation relative to spermatocyte maturation in the adult rat: a lanthanum tracer study. Anat Rec 190:99-t12 Simionescu M (1980) Structural and functional differentiation of microvascular endothelium. Ciba Found Syrup 71:39-60 Singer SJ, Kupfer A (1986) The directed migration of eukaryotic cells. Ann Rev Cell Biol 2:337-368 Takano K, Bogert M, Malamud D, Lally E, Hand AR (1991) Differential distribution of salivary agglutinin and amylase in the Golgi body and secretory granules of human salivary gland acinar cells. Anat Rec 230: 307-318 VanMeer G, Simons K (1986) The function of tight junctions in maintaining differences in lipid composition between the apical and the basolateraI surface domains of MDCK cells. Eur J Mol Biol 5:1455 1464 Wells B (1985) Low temperature box and tissue handling device for embedding biological tissue for immunostaining in electron microscopy. Micron Micros Acta 16:49 53 Wooding FBP (1982) The role of the binucleate cell in ruminant placental structure. J Reprod Fertil [Suppl] 31:31-39 Wooding FBP (1992) The synepitheliochorial placenta of ruminants: binucleate cell fusions and hormone production. Placenta 13:101-113 Wooding FBP, Brandon MR (1993) Placental cell migration through tight junctions. J Anat 183:187 188 Wooding FBP, Chambers SG, Perry JS, George M, Heap RB (1980) Migration of binucleate cells in the sheep placenta during normal pregnancy. Anat Embryol (Ber) 158:361-370 Wooding FBP, Hobbs T, Morgan G, Heap RB, Flint APF (1993) Dynamics of growth in sheep and goat synepitheliochorial placentomes: an autoradiographic study. J Reprod Fertil 98:275283