Intermembrane distances at the feto-maternal interface in epitheliochorial placentation

: Introduction In an epitheliochorial placenta, the apical membranes of trophoblast cells and of uterine epithelial cells are in contact to each other (feto-maternal contact). In addition, there are also folds in which the trophoblast membrane is in contact with itself (feto-fetal contact) and areas where apical uterine epithelial membrane is in contact with itself (materno-maternal contact). Methods We use transmission electron microscopy of placental samples from pigs. (n = 3), cows (n = 2), sheep (n = 2), goat (n = 2) and roe deer (n = 1) to study the intermembrane distance in these three contact types. Results The measured intermembrane distances vary between 8 and 25 nm. One common feature is that the distance at feto-fetal contact sites is about 6–10 nm wider than at materno-maternal sites and feto-maternal sites show intermediate values. Discussion This finding suggests that the membrane distance at feto-maternal contact sites is determined by heterophilic binding of larger fetal to smaller maternal binding molecules. Homophilic binding of smaller maternal or larger fetal molecules lead to the smaller or wider intermembrane distances at materno-maternal or feto-fetal contact sites respectively. The observation that this similar pattern of membrane distances is present in pigs and in ruminants suggest that an evolutionary mechanism is involved in determining the intermembrane distance in epitheliochorial placentas. Introduction: In an epitheliochorial placenta, the apical membranes of trophoblast cells and of uterine epithelial cells are in contact to each other (feto-maternal contact). In addition, there are also folds in which the trophoblast membrane is in contact with itself (feto-fetal contact) and areas where apical uterine epithelial membrane is in contact with itself (materno-maternal contact). Methods: We use transmission electron microscopy of placental samples from pigs. (n = 3), cows (n = 2), sheep (n = 2), goat (n = 2) and roe deer (n = 1) to study the intermembrane distance in these three contact types. Results: The measured intermembrane distances vary between 8 and 25 nm. One common feature is that the distance at feto-fetal contact sites is about 6 – 10 nm wider than at materno-maternal sites and feto-maternal sites show intermediate values. Discussion: This finding suggests that the membrane distance at feto-maternal contact sites is determined by heterophilic binding of larger fetal to smaller maternal binding molecules. Homophilic binding of smaller maternal or larger fetal molecules lead to the smaller or wider intermembrane distances at materno-maternal or feto-fetal contact sites respectively. The observation that this similar pattern of membrane distances is present in pigs and in ruminants suggest that an evolutionary mechanism is involved in determining the intermembrane distance in epitheliochorial placentas. which show that the inter-cellular membrane distances at areas of fetal-fetal, fetal-maternal and maternal-maternal membrane contacts differ. These data


Introduction
The epitheliochorial placenta is characterised by the contact of the trophoblast to the apical surface of the maternal uterine epithelium. In higher mammals an epitheliochorial placenta evolved in several clades of laurasiatheria (cetartiodactyla, perissodactyla and pholidota) and in strepsirrhine primates [1]. Thus, major groups of domesticated animals (e.g. camelids, ruminants, equids, pigs) have an epitheliochorial placenta. In the evolution of the placenta in higher mammals, this non-invasive epitheliochorial placenta evolved from an invasive placenta of the endotheliochorial [2] or haemochorial [3] type.
During the development of the epitheliochorial placenta, the connection between the trophoblast and the uterine epithelium passes through stages of apposition, adhesion and further into the formation of the definitive placenta, which is typically characterised by membrane attachment between interdigitating fetal and maternal microvilli [1,4].
Many studies deal with the molecules that are involved in feto-maternal adhesion in the initial stages of pregnancy (reviewed by Refs. [5][6][7]). Roles for adhesion molecules from different groups (e.g. integrins, cadherins) have been identified for these early stages, but the molecular basis of the attachment in later stages is much less well defined.
Recently, we characterised ultrastructural features of the mature epitheliochorial placenta in several ruminant species, in domestic pigs and in one horse [8]. It was shown that in addition to the contact of fetal to maternal membranes there are also folds in which apical fetal cell membrane is attached to fetal membrane and areas in which apical maternal cell membrane is in contact to maternal membrane. It was also revealed that in the definitive mature epitheliochorial placenta the fetal membrane surface is larger than the maternal surface [8].
In the present study, we present ultrastructural data which show that the inter-cellular membrane distances at areas of fetal-fetal, fetalmaternal and maternal-maternal membrane contacts differ. These data give important information about the connection between fetal and maternal surfaces in the mature epitheliochorial placenta.

Transmission electron microscopy (TEM)
Placental specimens from cattle (gd 140 and 213) were obtained at routine slaughtering at commercial slaughterhouses. Samples of sheep (gd 120 and 143), goat (gd 100 and 145) and roe deer (fetal crown rump length 8 cm) placenta were epon embedded archival blocks, provided by Dr. Peter Wooding, University of Cambridge, UK. Blocks of porcine placenta (gd 42, 76 and 108) were provided by Dr. Vibeke Dantzer, University of Copenhagen, Denmark. All samples were taken in compliance with the national animal welfare regulation at the time and location of sampling. Specimens from domestic animals were perfusion fixed through the fetal vasculature. The tissues were postfixed with osmium tetroxide and embedded in epon.
Semithin sections (1 μm) were cut with an ultramicrotome (Ultracut E, Reichert, Vienna, Austria), mounted on glass slides (Menzel-Gläser, Braunschweig, Germany), stained with toluidine blue and coverslipped. These sections were used to identify well-preserved tissue regions. Resin blocks with such regions were trimmed and ultrathin sections (70 nm) were cut with the ultramicrotome, mounted on grids and stained with uranyl acetate and lead citrate. A TEM (CM12, Philips, Eindhoven, Holland) was used to identify areas with a well preserved feto-maternal interface. Digital photographs of the feto-maternal interface were taken with a CCD camera (Orius SC1000, Gatan, Pleasanton, California, USA) at a magnification of ×110 000.

Measurement of membrane distances
Images were opened with ImageJ (Version 1.0). Regions with distinguishable fetal and maternal membranes were selected. This distinction was based on the knowledge of the ultrastructure of the fetomaternal junction in the species studied [8]. Areas in which adjacent membranes were cut perpendicularly (the pairs osmiophilic leaflets of each membrane were clearly visible) measurements were made between the outer osmiophilic leaflet of the membranes. In each specimen at least 10 measurements were made for feto-fetal, feto-maternal and materno-maternal contact.

Transmission electron microscopy
A close attachment of the membranes at the feto-maternal interface was found only in specimens with a very good ultrastructural preservation. In such specimens (Fig. 1) no large gaps of the intercellular space were observed. A complex system of interdigitating fetal and maternal microvilli and folds was found. In other regions with a suboptimal fixation, fetal and maternal tissues were typically separated at this interface. In well preserved regions, adjacent membranes run in parallel at specific distances (Figs. 2 and 3).
In one goat (gd 100) fetal and maternal membranes showed obvious staining differences (Fig. 2 A and B). In this case the outer extracellular leaflet of the fetal membrane was covered by a layer of electron dense material. On the maternal membrane this layer was absent or much less obvious.
In one pig placenta (gd 42) a double line in the middle of the intercellular space could be seen in a feto-fetal contact area (Fig. 2C).
In all specimens a difference between the three distances (feto-fetal, feto-maternal, materno-maternal) was observed (Figs. 3 and 4). The distance was widest at feto-fetal contact areas, intermediate at fetomaternal and smallest at materno-maternal contact areas (Fig. 5).

Lectin staining
Strong binding of the lectin to the feto-maternal interface was observed with the RCA, SBA, WGA and PNA lectins. This was observed in regions where fetal and maternal membranes were attached to each other and also in regions where they were separated. In such regions where fetal and maternal surfaces were separated, differences in the staining intensity were observed (Fig. 6). While in the RCA and WGA stained specimens the staining of both surfaces did not show obvious differences in intensity, staining with PNA and SBA was stronger on the

Discussion
The main finding of our study is that the distances between the cell membranes differ in feto-fetal, feto-maternal and in materno-maternal contact areas. These distances could only be measured in sections of very well-fixed tissue. Only locations in which the membranes were cut perpendicularly were measured. Deviations from this perpendicular orientation would have slightly increased the measured values due to geometric distortion. These preconditions prevented random-systematic sampling of a stereological study. Nevertheless, we think that the measurements are representative, since they were found consistently in a (low) number of specimens per species and also across species. Our distance values underestimate the real values since there is a shrinking factor, which is due to tissue fixation and processing. The linear shrinkage of tissue processing used for our samples is likely to be about 10-15% [10,11].
From a cell biological perspective the attachment of the trophoblast to the uterine epithelium is a very unusual condition [12]. In this case adhesion molecules, which are typically expressed at the baso-lateral membrane, are employed for intercellular adhesion at apical surfaces. A partial depolarisation of the trophoblast and the uterine epithelium is a prerequisite for adhesion. Structural changes of the receptive uterine epithelium include a temporary disappearance of apical microvilli, which later regrow in a modified shape [13].
The molecular interactions that initiate the contact between the blastocyst arriving in the uterus and the initial contact between trophoblast and uterine epithelium have been studied in some detail [5]. The distance between trophoblast and uterine epithelium at these early stages is about 20 nm in mice [14]. In species with an invasive endotheliochorial or hemochorial placenta, this apical contact is later replaced by contact of the trophoblast to deeper layers of endometrial tissue or to maternal blood. In the epitheliochorial placenta the apical contact between trophoblast und uterine epithelium remains intact and is further developed into the definitive mature placenta. But what actually keeps the membranes in contact in the developed epitheliochorial placenta is not well characterised.
Our study does not provide the identity of the molecules involved in feto-maternal attachment, but it gives some information about them. The difference in membrane distance indicates that the molecules that determine the distance are larger at the fetal surface and about 3-4 nm smaller at the maternal surface. Thus, homophilic interaction could lead to the small distance in materno-maternal contact and the about 6-8 nm larger distance in feto-fetal contact. The intermediate distance in fetomaternal contact areas might result from heterophilic interaction between the fetal and maternal molecules.
The feto-maternal interface is strongly glycosylated and these glycans may be involved in membrane attachment [15]. Our lectin staining shows that there is a difference between fetal and maternal surfaces. The stronger fetal staining intensity of SBA and PNA lectin suggest that specific N-acetylgalactosamine and galactose residues, respectively, differ between these surfaces. Further investigation would be needed to identify the structure of these glycans.
It has been suggested that integrins and osteopontin might be involved in the feto-maternal attachment, not only in early pregnancy but also in the definitive placenta [16]. In that model extracellular osteopontin serves as a bridging molecule between integrin molecules of the trophoblast and the uterine epithelium. Different types of integrins and different activation stages could lead to differences in the intercellular distance as observed in our study. Integrins occur in different conformations, which can be upright or bent. In the bent conformation the ligand binding site is close to the cell membrane which would allow the relatively small intermembrane distances measured in our study. In the sheep placentome, osteopontin was detected apical between trophoblast and maternal caruncular epithelium [17] and thus may act as a bridging molecule as suggested [16]. In pigs a more diffuse intracellular localisation of osteopontin was observed in the definitive placenta [16,18]. This may indicate that osteopontin bridging between fetal and maternal apical integrin molecules is not a general mechanism of feto-maternal attachment in epitheliochorial placentas.
One further possibility is that cadherins are part of this attachment mechanism. The intercellular distances that we measured are in the range of those described in the cadherin mediated zonula adherens (15-25 nm) [1,19]. The two lines observed in the porcine placenta ( Fig. 2C) are very similar to the observations in adherens junction, visualised by cryo-electron microscopy [20]. Cadherins are expressed at the feto-maternal junction in cattle [21][22][23][24], sheep [25] and pigs [26]. Beta-catenin, which is involved in the linkage of actin to the intracellular domain of cadherins, is also expressed in bovine feto-maternal junction [23]. The ability of cadherins to undergo heterophilic and homophilic binding [27] would also permit intercellular distance differences, as observed in our study.
In summary, we show that the membrane distances in the epitheliochorial placenta differ between feto-fetal, feto-maternal and materno-maternal contact sites. Feto-fetal distances are 6-8 nm wider than materno-maternal distances and feto-maternal values are intermediate. This suggests size differences between the molecules that determine these distances. The fact that these distance differences occur in ruminants and in the domestic pig may indicate that conserved molecular mechanisms are involved in membrane attachments in the epitheliochorial placentation.

Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence