Segregation during cleavage in the mammalian embryo? A critical comparison of whole-mount/CLSM and section immunohistochemistry casts doubts on segregation of axis-relevant leptin domains in the rabbit.

Segregation of certain cytoplasmic molecules during cleavage and blastocyst formation that was previously reported to occur in the human and the mouse (Antczak and Van Blerkom Mol Hum Reprod 3:1067-1086, 1997; Antczak and Van Blerkom Hum Reprod 14:429-447, 1999) has been reinvestigated in the rabbit model. Additional methodology was used and two approaches were compared: (1) whole-mount immunohistochemistry followed by confocal laser scanning microscopy (WM-IHC/CLSM) versus (2) IHC performed on histological sections of resin-embedded material (S-IHC). This study concentrates on leptin and cytoskeletal proteins (actin and cytokeratins). With S-IHC, leptin was localized predominantly on the surface of blastomeres which is facing the perivitelline space, and in the extracellular embryonic coats, without any polar asymmetry being detectable along (presumptive) embryonic axes. A polar distribution of leptin with a pattern that could be interpreted as predictive of the prospective embryonic-abembryonic axis was seen only with WM-IHC/CLSM, not with S-IHC, although the latter gave excellent resolution. With both techniques, no differences between blastomeres were detected with respect to actin and cytokeratin patterns, an increased expression of cytokeratin in trophoblast cells occurring no earlier than at blastocyst formation. Artifacts that can occur with the two methodological approaches are critically discussed, as is the possible significance of the findings for theories on the differentiation of trophoblast versus embryoblast and on axis formation in early mammalian development. It is concluded that these data call for cautioning when studying distribution patterns of diffusible molecules with WM-IHC/CLSM technology, whereas patterns obtained with S-IHC are more reliable. Specifically these data cast doubts on previous claims that leptin IHC would allow to monitor cytoplasmic domain segregation occurring during cleavage as an element of early embryonic pattern/axis formation.

Influences of extracellular matrix and of conditioned media on differentiation and invasiveness of trophoblast stem cells.

Embryo implantation in the human and rodents relies on the trophoblast's ability to invade into the uterine stroma, partly depending on proteinases degrading components of basement membrane and underlying extracellular matrix (ECM). We have utilized mouse trophoblast stem (TS) cells (Science, 1998, 282:2072) to study trophoblast invasion and trophoblast-ECM interactions in vitro. On plastic in fibroblast-conditioned medium containing fibroblast growth factor (FGF)-4 and heparin, the cells remain proliferative but display increased differentiation in media without these components. Marker gene expression (Eomes, Pl-1, Tpbp) and invasion assays showed that TS cells exhibit increased invasive capacity when differentiating into giant cells and spongiotrophoblasts in unconditioned media without FGF-4 and heparin. Concomitantly, an up-regulation of matrix metalloproteinases (MMP)-9 and -14 was observed. Culture on gels of the basement membrane-like Matrigel resulted in striking changes in morphology and gene expression. Differentiating TS cells invaded into this ECM in a three-dimensional culture, while in turn ECM contact enhanced differentiation of TS cells and up-regulated the expression of MMP-9 and its tissue inhibitor (TIMP)-3. These findings implicate that the TS cell culture system used in this study can be utilized as a model for studying the regulation of trophoblast-ECM interactions, differentiation, and invasion in vitro.

Potentiality of embryonic stem cells: an ethical problem even with alternative stem cell sources.

The recent discussions about alternative sources of human embryonic stem cells (White Paper of the US President's Council on Bioethics, 2005), while stirring new interest in the developmental potential of the various abnormal embryos or constructs proposed as such sources, also raise questions about the potential of the derived embryonic stem cells. The data on the developmental potential of embryonic stem cells that seem relevant for ethical considerations and aspects of patentability are discussed. Particular attention is paid to the meaning of "totipotency, omnipotency and pluripotency" as illustrated by a comparison of the developmental potential of three-dimensional clusters of blastomeres (morula), embryonic stem cells, somatic or (adult) stem cells or other somatic (non-stem) cells. This paper focuses on embryoid bodies and on direct cloning by tetraploid complementation. Usage and patenting of these cells cannot be considered to be ethically sound as long as totipotency and tetraploid complementability of embryonic stem cells are not excluded for the specific cell line in question. Testing this poses an ethical problem in itself and needs to be discussed in the future.

Endometrial receptivity: selective adhesion competence of rabbit uterine epithelium for trophoblast but not for various tumor cells.

A modification of an established in vitro model for embryo implantation was used to probe the receptive uterine epithelium for any specificity of interaction with various invasive cells other than trophoblast. Endometrial explants consisting of stroma and epithelium taken from pseudopregnant rabbits were cultured in the presence of progesterone in order to regenerate a complete epithelial lining while maintaining the receptive state. Such precultured fragments were brought into contact with multicellular spheroids of different invasive tumor cell lines from different species. In contrast to the trophoblast of the rabbit blastocyst (previous publication), none of the tumor cell lines was able to adhere to intact epithelium of endometrial fragments nor to penetrate it. The uterine epithelium was also an insurmountable barrier for tumor cell spheroids confronted with the epithelium of fresh complex explants consisting of endometrium and myometrium or for spheroids introduced into the uterine lumen of pregnant/pseudopregnant rabbits at the periimplantation phase. However, all tumor cells were able to adhere to and mostly also to invade into the endometrial stroma when it was exposed artificially, i.e. when the epithelium was removed. These results suggest that the receptivity of rabbit uterine epithelium shows a remarkable selectivity with respect to cell type (trophoblast) and species (rabbit, not human, mouse, or rat).

The regulatory function of the uterine epithelium for trophoblast attachment: experimental approaches.

Embryo implantation in the mammalian uterus is initiated by the formation of a direct cell-to-cell contact between the trophoblast of the blastocyst and the uterine epithelium. This process is far from trivial since apical plasma membranes of epithelial cells are normally non-adhesive. The uterine epithelium has the remarkable ability to enter, under steroid hormone control, a specific state ("receptivity") at which it can down-regulate this repellent property and can finally become apically adhesive for trophoblast (probably aided additionally by local paracrine signals). Experimental data from recent years are beginning to shed some light on the involved cell biological/molecular events. They will be discussed on the basis of concepts concerning the regulation of epithelial cell polarity and with side views on epithelial-mesenchymal transformation. Recently developed experimental in-vitro systems have allowed to detect a remarkable degree of selectivity in the interaction of trophoblast and uterine epithelium, in contrast to stroma invasion. A new approach enables us to determine actual adhesive forces between living trophoblast and uterine epithelial cells with a special modification of the atomic force microscope (force spectroscopy). The potential use of such an approach is discussed.

Calcium influx in human uterine epithelial RL95-2 cells triggers adhesiveness for trophoblast-like cells. Model studies on signalling events during embryo implantation.

RL95-2 is a human uterine epithelial cell line that exhibits adhesion competence on its apical surface for trophoblast-like JAR cells. Using confocal microscopy and an adhesion assay we have found that changes in intracellular free calcium ([Ca(2+)](i)) in RL95-2 cells are involved in binding of JAR spheroids. Impact of spheroids upon, and movement of spheroids across, monolayers of RL95-2 cells produced a transient increase in [Ca(2+)](i). Pretreatment of RL95-2 cells with the Ca(2+) channel inhibitor, diltiazem, reduced the [Ca(2+)](i) increase. Interestingly, resting of JAR spheroids on RL95-2 cells caused no detectable alterations in [Ca(2+)](i) although cell-cell bonds were formed during prolonged contact. However, separation of established bonds did produce an increase in [Ca(2+)](i) which could be reduced by the Ca(2+) channel blocker, SKF-96365, but not by diltiazem. SKF-96365 also reduced adhesion of JAR spheroids to RL95-2 cells. In all experiments, the increase in [Ca(2+)](i) was due to influx from the external medium, as it could be blocked both by removing extracellular Ca(2+) and by nickel. These results suggest that the plasma membrane of uterine RL95-2 cells contains two types of Ca(2+) channels that are involved in trophoblast adhesion, i.e. diltiazem-sensitive channels contributing to initiation of JAR cell binding and SKF-96365-sensitive channels participating in a feedback loop that controls the balance of bonds.

Adhesion of trophoblast to uterine epithelium as related to the state of trophoblast differentiation: in vitro studies using cell lines.

At the initial phase of embryo implantation, the trophoblast must have acquired competence for adhesion to the uterine epithelium, a condition whose cell biological basis is far from understood. In the present study, trophoblast-type cells (BeWo, JAr, and Jeg-3 choriocarcinoma cell lines) were treated with retinoic acid, methotrexate, dibutyryl-cAMP, or phorbol-12-myristate-13-acetate in order to modulate their ability to adhere to uterine epithelial cells (RL95-2). In an established model, multicellular spheroids of choriocarcinoma cells were transferred onto the surface of monolayer cultures of RL95-2 cells followed by a centrifugal force-based adhesion assay. In controls, about 45% of BeWo and JAr cell spheroids and 75% of Jeg-3 spheroids adhered to uterine monolayers within 30 min. Pretreatment of spheroids with either of the agents stimulated differentiation as indicated by the rate of chorionic gonadotropin secretion, but consistently reduced the adhesion to the endometrial monolayer in all three choriocarcinoma cell lines. While previous investigations had shown that invasiveness of trophoblast cells (into extracellular matrix) does not seem to be linked to the differentiation program in a simple manner, the present data suggest that such an (inverse) link may indeed exist with respect to the ability to initiate an adhesive interaction with the uterine epithelium. These observations support the view that epithelial cell interactions as typical for the initial phase of embryo implantation are regulated in a way that is clearly different from cell-matrix interactions governing later phases of trophoblast invasion into the endometrial stroma.

Structural dynamics and function of early embryonic coats.

The extracellular embryonic coats (embryo/blastocyst coverings) that surround early mammalian embryos are most often still referred to as zona pellucida. Accumulating evidence from a number of earlier and recent studies clearly indicates that this is an oversimplification which cannot be defended anymore, at least in many species. Structural modifications of the coats occur during cleavage and blastocyst stages; these are most obvious in a number of species with the central type of implantation and, related to this, a high degree of blastocyst expansion, notably the rabbit and the horse. In this contribution, formation and transformation of the various layers of coats (zona pellucida, mucoprotein layer, neozona and gloiolemma in the rabbit, capsule in the horse) will be reviewed, as will be comparable structures seen, e.g., in the fur seal (subzonal layer) and the baboon. These phenomena will be discussed in the context of structurally more subtle changes found in other species, including those with small blastocysts and other types of implantation, in particular with biochemical modifications which may be physiologically quite important. The molecular mechanisms of deposition of coats and their transformation and shedding/dissolution will be briefly addressed. The possible functional significance of the coats and their transformation will be discussed (mechanical, morphogenetic and immunological role, molecular transport control, blastocyst positioning, implantation, trap and reservoir function for signalling molecules).

Interactions between trophoblast and uterine epithelium: monitoring of adhesive forces.

At embryo implantation, it is postulated that the initial contact between blastocyst and maternal tissues is by adhesion of the trophoblast to the uterine epithelium. This cell-to-cell interaction is thought to be critical for implantation, although the actual adhesive forces have never been determined. In the present study, the atomic force microscope (AFM) was used to study the adhesion between human uterine epithelial cell lines (HEC-1-A; RL95-2) and human trophoblast-type cells (JAR). Specific interaction forces of these epithelia via their apical cell poles were determined on the basis of approach-and-separation cycles. For this purpose, the AFM tip was functionalized with JAR cells, then brought to the surface of uterine epithelial monolayers and was kept in contact for different periods of time (ms, 1, 10, 20, 40 min). The approach force curves displayed repulsive interactions for both HEC-1-A and RL95-2 cells. However, RL95-2 cells (with a smooth surface structure and a thin glycocalyx) showed lower values of the repulsive regime than HEC-1-A cells (with a rough surface structure and a thick glycocalyx). After having overcome repulsive interactions, the initial contact was followed by adhesive interactions. For contact times of 20 and 40 min, RL95-2 cells, but not HEC-1-A cells, showed specific JAR binding, i.e. the separation force curves displayed repeated rupture events in the range of 1-3 nN with a distance between 7-15 microm and, thereafter, a final rupture event at a distance of up to 45 microm. These features point to the formation of strong cell-to-cell bonds. Collectively, these studies provide the first definition of interaction forces between the trophoblast and the uterine epithelium, and are consistent with the hypothesis that an RL95-2-like architecture of uterine epithelial cells, i.e. an non-polarized phenotype, is essential for apical adhesiveness for the human trophoblast.

Identification of differentially expressed genes in human trophoblast cells by differential-display RT-PCR.

Molecular mechanisms controlling human trophoblast invasiveness are still poorly understood. In the present investigation, mRNA patterns of trophoblast cells isolated from first trimester (high invasiveness) and term placentae (no/low invasiveness) were compared by differential-display reverse transcriptase polymerase chain reaction (DDRT-PCR) revealing differential expression of numerous genes. Of 18 differentially expressed DDRT-PCR products analysed, 11 were unknown, four showed homologies with expressed sequence tag sequences, and three others were homologous to integrin-beta1, ATP-synthetase U6 (both showing higher expression in first trimester) or to aldose-reductase (higher expression at term), respectively. One of the unknown transcripts (PBK1, accession number: AJ007398) was cloned from a first trimester placenta cDNA library and was characterized. The 1908-bp gene fragment contains an open reading frame of 1551 bp and an Alu-sequence in the 3' non-coding region. According to Northern blot analysis on JAr choriocarcinoma cells, the fragment is close to full-length cDNA. By in situ hybridization, PBK1 was detected only in first trimester but not term placentae in the proximal parts of cell islands and in closely adjacent villous cytotrophoblast. This expression pattern suggests that the newly identified molecule, PBK1, could be involved in the regulation of proliferation/ differentiation and potentially in invasion of trophoblast cells.

Differentiation markers and invasiveness: discordant regulation in normal trophoblast and choriocarcinoma cells.

In tumor cells, malignant (invasive) behavior and differentiation tend to be correlated inversely, although it is not clear to what extent this can be generalized and whether it may also apply to normal invasive cell types. We have modulated differentiation of normal trophoblast cells from first trimester or term placenta as well as choriocarcinoma cells (BeWo, Jeg-3, and JAr) with retinoic acid (RA), methotrexate (MTX), dibutyryl-cAMP (dbcAMP), or phorbol-[12-myristoyl-13-acetyl]-diester (PMA). The secretion of the differentiation marker chorionic gonadotrophin was stimulated by nearly all substances in all cell types. The activity of cellular sterylsulfatase showed a tendency to be increased (decreased by RA and dbcAMP in normal trophoblast; not detected in JAr). Invasiveness was decreased by all effectors in normal trophoblast (both types) and in BeWo. In Jeg-3 and JAr, however, PMA treatment (in JAr also RA treatment) increased invasion rates. These results suggest that only in normal trophoblast and in BeWo (but not in other choriocarcinoma cells, i.e., Jeg-3 and JAr) invasiveness and differentiation tend to be correlated inversely. When extrapolating to the various subpopulations of cells within a tumor, induction of differentiation-as intended in certain strategies for tumor therapy ("differentiation therapy")-may have the unwanted effect of stimulating invasiveness in certain subpopulations of tumor cells.

Adhesiveness of the free surface of a human endometrial monolayer for trophoblast as related to actin cytoskeleton.

Adhesiveness of the apical (free) plasma membrane of uterine epithelial cells for trophoblast is essential for the process of human embryo implantation. As epithelial cells are normally repellent, i.e. apically non-adhesive, we argue that a remodelling of the epithelial organization from a polarized to a non-polarized phenotype might prepare the apical pole for cell-cell adhesion during the so-called receptive phase. To identify details of apical adhesiveness we examined human epithelial RL95-2 cells (RL cells) which, in contrast to other cell lines, allow trophoblast to adhere to their apical plasma membrane. To determine whether the cytoskeletal structure is functionally critical for adhesiveness for trophoblast, RL cells were treated with actin depolymerizing cytochalasin D, i.e. 0.4 microM for 120 min. Changes in adhesiveness for trophoblast were monitored with a centrifugal force-based adhesion assay. Moreover, ultrastructural features, organization of the actin network and expression of integrins, i.e. alpha 6, beta 1, beta 4, were studied using electron microscopy, confocal laser scanning microscopy and cell surface immunogold-labelling techniques. Changes in transmission of mechanical signals via integrins into uterine cells were examined using a magnetic drag force device, thereby monitoring intracellular calcium responses. The results suggest that adhesiveness of the free surface of RL cells for human trophoblast requires an intact but non-polarized actin cytoskeleton, apically localized integrins linked to actin, and calcium signalling originating at the free surface.

A hypermedia tutorial for cross-sectional anatomy: HyperMed.

Modern imaging techniques like computer tomography (CT) and nuclear magnetic resonance (MR) imaging have become essential in clinical diagnostics and also in teaching gross anatomy to medical students. As a consequence, special classes in (cross)-sectional anatomy are being added to the curriculum in many anatomical institutions. Since institutional budgets often do not allow extensive supervision beyond the very limited time frame of traditional courses in gross anatomy, a computer-based hypermedia tutorial (HyperMed) was created and integrated into the teaching program of the Institute of Anatomy at Essen University. HyperMed offers two components, one for authors (e.g. teachers who can customize the contents of the program) and a second for users (e.g. students). In the present version, digital cross-sectional human images have been edited. The relevant anatomical structures in these images have been marked, named, and linked to additional information and figures (in particular schematic figures and CT images). Users can obtain information at different levels: (1) index-based retrieval, (2) navigational retrieval (on inspecting cross-sectional images the user is asked to identify structures) and (3) a history list enabling users to go back to any previous point of navigation. HyperMed was first tested in the winter terms 1995/1996 and 1996/1997 during classes on cross-sectional anatomy which are a supplement to the traditional dissection course of the Institute of Anatomy, University Essen. It was well received by the students who found it a helpful adjunct to learning cross-sectional anatomy.

Neoglycoprotein-binding sites (endogenous lectins) in the Fallopian tube, uterus and blastocyst of the rabbit during the preimplantation phase and implantation.

Regulation of the initial phase of embryo implantation may involve the recognition interplay of glycoconjugates and respective receptors such as endogenous lectins on both cellular surfaces. Whereas changes in glycoconjugate composition have been detected in preparation for embryo implantation and described in detail, knowledge on endogenous lectins has remained scant. Affinity probes (carrier-immobilized carbohydrate structures as ligand part on a histochemically inert backbone) are used in the present investigation in order to gain further insights in this area. Cryostat sections of rabbit Fallopian tubes and uteri in nonpregnant and early pregnant [tubes: 3 days post coitum (d p.c.); uteri: 3, 5, 7 and 9 d p.c.] states were studied for binding patterns of a series of biotinylated (neo)glycoproteins. A high density of binding sites was detected with beta-galactosides (with decreasing intensity: beta-D-galactose-BSA, asialofetuin with its triantennary glycan chains, lactose-BSA). Considerably less binding (but with the same pattern) was obtained with beta-N-acetyl-D-glucosaminide-BSA and is interpreted to originate from a cross-reactivity of such sites which may bind physiologically to Gal-beta1,3/4-GlcNAc sequences. In contrast, no evidence for the presence of binding molecules with specificities for a-D-mannose-BSA, maltose-BSA, N-acetyl-galactosaminide-BSA and N-acetyl-D-neuraminic acid-BSA was obtained in these tissues under the same conditions. The epithelium of the Fallopian tube showed a high density of beta-galactoside-binding sites at the apical cell poles (including the cytoplasm and membrane region) already in the nonpregnant state. At 3 d p.c., a strong reaction in all epithelial cells of the isthmus and a marked decrease in the ampulla were noted. The putative lectin(s) appear(s) to be synthesized and secreted by the tubal epithelium. A physiological role in forming the mucoprotein layer of the blastocyst coverings by precipitating the appropriate mucin-type molecules can be considered. Within the endometrium, the beta-galactoside-binding molecules were almost exclusively localized at the apical cell pole of epithelial cells, whereas there was hardly any binding in the epithelial cytoplasm or in the endometrial stroma. The reaction was very weak in the non-pregnant state but increased considerably until 5 d p.c., starting in the luminal-most parts of the epithelium. While the reaction was rather homogeneous at the surface of the luminal epithelium at 5 d p.c., the degree of heterogeneity increased stepwise from 7 to 9 d p.c. In the implantation chamber, the density of these beta-galactoside-specific 'receptors' was further enhanced in particular at the epithelial surface of the placental folds. In contrast, the reaction was less intense at the antimesometrial uterine epithelium and in interblastocyst segments of the uterus, and it remained weak in the middle and deep crypts. The trophoblast showed a high density of galactoside-binding sites at its surface, and less in the cytoplasm. Neoglycoprotein binding to the blastocyst coverings observed at 7 d p.c. was strong in particular at the outer and inner surfaces. Physical factors (e.g. differential texture at surfaces) are discussed to influence the staining patterns of these extracellular coverings. Nevertheless, the observations made on the tubal and the uterine mucosa suggest that the putative lectin(s) detected here is (are) secreted by these epithelia and could be involved in the structural organization of the various layers of the blastocyst coverings with their remarkable content of oligosaccharide chains. This effect on topological aspects of the zona pellucida equivalents may be important for the interplay between trophoblast and uterine epithelium and the cascade leading to implantation initiation.

Adhesiveness of the apical surface of uterine epithelial cells: the role of junctional complex integrity.

Embryo implantation necessitates that the apical plasma membrane of uterine epithelial cells acquires adhesiveness. Recent studies have indicated that modulation of a major element of the epithelial phenotype, i.e. apical-basal cell polarity, might be critical in this respect. Here, we analyze polar characteristics of nonadhesive vs. adhesive uterine epithelial cell lines focusing on cytoskeletal-junctional interactions that may play a role in regulating adhesiveness of the apical plasma membrane. HEC-1-A is a human uterine epithelial cell line exhibiting nonadhesive properties of its apical surface for trophoblast, whereas RL95-2 represent another such cell line exhibiting adhesive properties enabling trophoblast attachment. Homotypic intercellular contacts and functionally related proteins, i.e. ZO-1, E-cadherin, alpha-catenin, beta-catenin, plakoglobin, and desmoplakin 1, were examined by transmission electron microscopy, immunocytochemistry, confocal laser scanning microscopy, and immunoprecipitation techniques. In addition, details of actin filament architecture were studied after phalloidin labeling. While nonadhesive HEC-1-A exhibited the well-known pattern of cell-to-cell contacts of polarized epithelial cells, adhesive RL95-2 showed a lack of ZO-1 expression, tracer leakiness of the paracellular pathway, and atypical features in adherens junctions: E-cadherin, alpha-catenin and plakoglobin were colocalized in all plasma membrane domains and beta-catenin was localized in lateral membrane domains. Immunoprecipitations showed in both cell lines the presence of two different E-cadherin-catenin complexes, one composed of E-cadherin, alpha-catenin and beta-catenin, and the other of E-cadherin, alpha-catenin and plakoglobin. Concerning RL95-2 these data indicate that E-cadherin/plakoglobin complexes are randomly distributed, whereas E-cadherin/beta-catenin complexes are laterally localized in these cells. Additionally, the actin-based cytoskeleton of RL95-2 lacked a polar organization. With respect to the intermediate filament-desmosome system, both cell types expressed desmoplakin I, but the vast majority of RL95-2 lacked well-formed desmosomes as demonstrated by electron microscopy. It is concluded that modulation of tight junctions and/or remodelling of adherens junctions, e.g. differential distribution of E-cadherin/plakoglobin complexes and E-cadherin/beta-catenin complexes, are correlated with the development of apical adhesiveness of human uterine epithelial cells. This model system should allow to test experimentally whether this correlation is due to any causal function in the development of epithelial cell polarity.

Epithelial cell polarity and embryo implantation in mammals.

At embryo implantation we are confronted with the fact that uterine and trophoblast epithelium make contact via their apical cell membranes. This epithelium-epithelium adhesion leading to definitive attachment of the embryo to the uterine wall, however, is far from being trivial and has been called a cell biological paradox. It has been proposed that some of the molecular events involved in epithelium-to-mesenchyme transformation might play a role in the interaction between uterine cells and trophoblast. As a mechanism to achieve uterine epithelium adhesiveness for trophoblast it is postulated that uterine cells partially modulate their epithelial phenotype. Data from recent in vitro experiments give support to this hypothesis and suggest that loss of apical-basal cell polarity might prepare the apical cell pole of uterine epithelium for cell-to-cell contact with trophoblast in vivo.

The role of matrix contact and of cell-cell interactions in choriocarcinoma cell differentiation.

Cell differentiation is supported much better by gels of extracellular matrix than by the same matrix provided as a rigid substrate. Many cell types including normal and malignant trophoblast cells, however, form multicellular multilayered aggregates on matrix gels with increased cell-to-cell contacts as compared to regular monolayers on rigid matrix substrates. In such cultures, it remained open, so far, whether stimulated expression of differentiation markers is caused by enhanced cell-to-cell communication or is displayed only by cells in direct contact to the gel. Therefore, choriocarcinoma cells (BeWo) were grown as aggregates: (a) on gels of the basement membrane-like Matrigel, (b) on plastic coated with poly-HEMA, or (c) as aggregates (spheroids) in suspension culture. Production of the differentiation marker chorionic gonadotropin was stimulated significantly in aggregates attached to gels of Matrigel or to the poly-HEMA substrate but not in suspended spheroids. With respect to cell-cell communications, however, expression of E-cadherin mRNA was not altered in any type of aggregates, as compared to control cultures on plastic. The expression of connexin43 mRNA (not of connexin26) was increased only in suspended spheroids, while microinjection of the fluorescent dye Lucifer Yellow suggested that cell communication via gap junctions was absent from cells grown as monolayers and was not induced in any type of aggregate. When cells were grown on gels of Matrigel, the relevance of direct cellular contact to the substrate for differentiation was analyzed by immunohistochemistry. Trophoblastic differentiation markers (chorionic gonadotropin, placental lactogen, placenta-type alkaline phosphatase, and pregnancy-specific glycoprotein beta 1) as well as the proliferation marker Ki-67 were not preferentially expressed in cells that were in contact with the gel. Similar random distributions of all these markers were also observed in spheroids cultured in suspension. The distributions of several matrix molecules and of different integrins were comparable between aggregates on matrix gels and those in suspension culture. According to these data, cell-cell communication appears to play a subordinate role for cytodifferentiation in cell aggregates on matrix gels, so that substrate anchorage and physical properties of the substrate may be the decisive factors. Interestingly, however, direct contact to the substrate does not seem to be essential for the stimulation of differentiation in cells on matrix gels. The results are discussed in the context of the "tensegrity"-model for cell-matrix interactions in which proper mechanical properties of the substrate are important for the regulation of cell differentiation by allowing a balanced integrity of external and cell-internal tensile forces.