Expression of syndecan, a putative low affinity fibroblast growth factor receptor, in the early mouse embryo

Syndecan is a heparan sulfate- and chondroitin sulfate- bearing integral membrane proteoglycan originally isolated from mouse mammary epithelial cells. Synde- can’s binding properties suggest that it can serve as a receptor for extracellular matrix components and heparin-binding growth factors, while localization studies and transfection experiments indicate that it plays important roles both in the maintenance of adult tissue architecture and in organogenesis (see Bernfield and Sanderson, 1990, for review). Despite its predominance on epithelia in adult tissues (Hayashi et al. 1987), syndecan binds in vitro to extracellular matrix components such as fibrillar collagens, fibronectin and thrombospondin, which are usually associated with interstitial matrices (Koda et al. 1985; Saunders and Bernfield, 1988; Sun et al. 1989). This has led to the proposal that syndecan serves as an anchor to bind epithelial sheets to their subjacent stroma (Bernfield and Sanderson, 1990). Experimental evidence shows that syndecan does play an important role in maintenance of epithelial morphology. Decreasing the level of syndecan expression in cultured epithelial monolayers causes them to become fusiform and fibroblastic (Saunders et al. 19896; Kato and Bernfield, 1990), while increasing the level of syndecan expression in cultures of transformed, epithelial cells that are fibroblastic causes them to reacquire some epithelial characteristics (Jalkanen et al. 1990; Leppa et al. 1991). In addition, syndecan expression undergoes striking changes during epithelial–mesenchymal transitions that are a part of normal fetal development: it is turned on in mesenchymal cells that will become epithelial, and turned off in epithelial cells that become mesenchymal (Thesleff et al. 1988; Vainio et al. 1989; Fitchett et al. 1990; Trautman et al. 1991).

Recent studies have shown that syndecan also participates in cellular interactions with basic fibroblast growth factor (bFGF). Syndecan binds to bFGF in vitro (Bernfield and Sanderson, 1990; Krufka et al. 1990) and is homologous to a low-affinity FGF receptor identified by ligand-affinity cloning (Keifer et al. 1990). The binding of FGF to cell surface heparan sulfate proteoglycans is critical to its biological activity. Treating cells with heparitinase to remove cell surface heparan sulfate GAG, or with chlorate, to inhibit their sulfation, prevents binding of bFGF to a high-affinity receptor (A.C. Rapraeger, personal communication). In fibroblasts, this prevents FGF-stimulated growth, and in MM14 muscle cells, causes initiation of differentiation (A.C. Rapraeger, personal communication). Expression of a high-affinity FGF receptor in various mutant CHO cell lines results in significant high-affinity binding only in cells that also express cell surface heparan sulfate (Yayon et al. 1991). The role of heparan sulfate proteoglycans was found to be more than stabilization of bFGF, and may involve changing the conformation of FGF in a way that allows it to interact with high-affinity receptors (Yayon et al. 1991). These results indicate that the ability of cells to respond to FGF can be regulated by changes in expression of cell surface proteoglycans (Klagsbrun, 1990).

The apparent diversity of syndecan’s functional properties may reflect alterations in its composition. Syndecan contains both heparan sulfate and chondroitin sulfate glycosaminoglycan (GAG) chains (Rapraeger et al. 1985) whose size and number vary between cell types (Sanderson and Bernfield, 1988; Kato and Bernfield, 1989). Changes in the number of associated heparan sulfate GAG chains or in their sulfation could alter the affinity of syndecan for bFGF, which binds specifically to highly sulfated heparan sulfate (Yayon et al. 1991). The ratio of heparan sulfate to chondroitin sulfate chains bound to the syndecan core protein is modulated by cells in response to the growth factor TGF-β (Rasmussen and Rapraeger, 1988) suggesting one mechanism by which syndecan’s activity or specificity could be modified. Thus, syndecan has the potential to promote signal transduction by one class of growth factors (FGFs), and to be structurally, and perhaps functionally, regulated by another class of growth factors (TGF-βs).

Some of the earliest events in mammalian embryogenesis include the organization of epithelia, such as the trophectoderm and primitive endoderm (Fleming and Johnson, 1988), and early inductive interactions that may involve members of the FGF family of growth factors (Slack et al. 1990). To help determine what role syndecan might play in early development, we examined its distribution in mouse embryos between the 2cell and late gastrula stages, using monoclonal and polyclonal antibodies that recognize the core protein of the proteoglycan (Jalkanen et al. 1985; Hayashi et al. 1987; Jalkanen et al. 1988). We find that syndecan is expressed from the 4-cell stage on, and has distribution patterns in the postimplantation embryo that suggest roles in stabilization of epithelia, in placental development, and in early axial patterning of the mesoderm.

Protein A–Sepharose and CNBr-activated Sepharose 4B were obtained from Pharmacia, Piscataway, NJ. Trypsin was obtained from Difco, Detroit, MI. Diaminobenzidine, pancreatin, polyethylene glycol 1450 and 3350, the protease inhibitors benzamidine, phenylmethylsulfonylfluoride, A-ethylmaleimide and pepstatin, and Tween 20 were from Sigma Chemicals, St Louis, MO. Dulbecco’s modified Eagle’s medium (DME) was from the UCSF Cell Culture Facility. Diethylene glycol distearate and conical micromolds were from Polysciences, Warrington, PA. Normal donkey serum, biotinylated secondary antibodies to rabbit IgG, and streptavidin–Texas Red were from Amersham, Arlington Heights, IL. Biotinylated secondary antibodies to rat and to rabbit IgG and horseradish peroxidase detection kit were obtained from Vector Laboratories, Burlingame, CA. Protein relative molecular mass standards were obtained from Bethesda Research Laboratories, Gaithersburg, MD, and included myosin (205 × 10³M_r), phosphorylase b (97.4×10³M_r), bovine serum albumin (68×l0³A/_r), ovalbumin (43×lO³M_r), and carbonic anhydrase (29×l0³M_r). Gene-Trans was from Plasco, Inc., Woburn, MA. Heparitinase and chondroitin sulfate ABC lyase were from Miles Laboratories, Naperville, IL.

The rat monoclonal antibody (mAb) to syndecan, 281-2, was purified from hybridoma culture supernatant (Jalkanen et al. 1985), and the rabbit polyclonal anti-syndecan IgG (Jalkanen et al. 1988) was purified from the antiserum using protein A–Sepharose. Control antibodies included the rat monoclonal antibody MEL-14, which recognizes the murine peripheral lymph node lymphocyte homing receptor (Gallatin et al. 1983; and is isotype matched to 281-2 (a gift of Dr Irving Weissman, Stanford University, Stanford, CA), and non- immune rabbit IgG purified from serum using protein A–Sepharose.

Embryos were flushed at the 2-cell stage from superovulated female ICR mice (12 weeks old; Harlan Sprague-Dawley Inc.) and cultured as described previously (Sutherland et al. 1988). Unattached hatched blastocysts were obtained by culturing embryos at the unhatched, expanded blastocyst stage in serum-free Eagle’s medium (Spindle and Pedersen, 1973; Spindle, 1980) in organ culture dishes for 48 h. Ectopiacental cone (EPC) expiants were made by removing 7.5-day embryos from their implantation sites, dissecting away Reichert’s membrane, separating the embryo and ectoplacen-tal cone, and then incubating the latter tissue in 0.5% trypsin−2.5 % pancreatin for 15 min at 4°C to isolate the diploid core from the surrounding giant cells. After pipetting through a fine-bore pipette, the isolated cores were cultured for 2 – 5 days in DME H-21 medium containing 10% fetal bovine serum, during which time they differentiate and form monolayers of giant cells.

Preimplantation embryos were fixed in Carnoy’s fixative (6:3:1 of ethanol, chloroform and glacial acetic acid) for 30 min, and dehydrated and embedded in polyethylene glycol (PEG) according to the methods of Watson and Kidder (1988). The PEG-embedded embryos were sectioned at 1 um and mounted on slides coated with 0.1% poly-L-lysine. For staining, the sections were rehydrated, rinsed with 0.1% Tween 20 in phosphate-buffered saline (PBS–TW), and then incubated in 1 % normal donkey serum in PBS–TW. Primary antibody incubation was for 1 h at room temperature and was followed by an overnight rinse in PBS–TW at 4 °C. The sections were then incubated in biotinylated secondary antibody followed by streptavidin–Texas Red. The slides were examined and photographed using Kodak Tri-X film.

Pregnant female mice, obtained as described above, were housed until the appropriate day of gestation. For 5.5-day embryos, the uterus was cut into sections, each of which contained an implantation site, and fixed without further dissection. For 6.5-day, 7.5-day and 8.5-day embryos, implantation sites were removed from the uterus, and the embryos were dissected out. All embryos were fixed for 45 min in Carnoy’s fixative, dehydrated and embedded either in paraffin or in diethylene glycol distearate (DGD; Capeo and McGaughey, 1986; Valdimarsson and Huebner, 1989). DGD sections (1 urn) were mounted on poly-L-lysine-coated slides, rehydrated, processed and examined for immunofluorescence as described above for polyethylene glycol sections. Paraffin sections were cut at 5μ m, mounted on slides coated with chrome alum-gelatin, deparaffinized in xylene and rinsed in 100% ethanol. Endogenous peroxidase activity was quenched by incubation in 1 % hydrogen peroxide in methanol for 30 min at room temperature. The paraffin sections were incubated in 1 % normal donkey serum in PBS followed by an overnight incubation in primary antibody at 4°C. The sections were then incubated with biotinylated secondary antibody, followed by streptavidin–horseradish peroxidase. Finally they were reacted with diaminobenzidine to develop color, the intensity of which was enhanced by incubation in a 0.5% solution of CuSO₄ in 0.9% NaCl. The resulting slides were photographed using Kodak Technical Pan film and a blue filter.

Hatched blastocysts were cultured overnight in serum-free, low-sulfate Eagle’s medium (with MgCl₂ substituted for MgSO₄) containing 200μCi ml^-1 of [³⁵S]H₂SO₄ (ICN Radiochemicals; specific activity 43 Ci mg^-1 S). The labeled embryos were then collected in a microfuge tube, frozen in liquid nitrogen and stored at − 80°C until use.

For labeling, ectoplacental cone expiants were cultured overnight on glass coverslips (12 mm diameter) in low-sulfate Eagle’s medium containing 10% fetal calf serum and 200 μCi ml^-1 of [³⁵S]H₂SO₄, which were then frozen in liquid nitrogen and stored at −80°C until use.

Syndecan was isolated from labeled or unlabeled blastocysts and outgrowths as described previously for other tissues (Sanderson and Bernfield, 1988). Briefly, syndecan was extracted from embryos in 1ml extraction buffer (10 HIM Tris, pH7.4, containing 150mM NaCl, 1% Triton X-100, 0.5M KCl, 1 MM phenylmethylsulfonylfluoride, 5 DIM /V-ethyl malei- mide, 5IHM benzamidine–HCl, and l0 μgml^-1 pepstatin) and isolated using mAb 281-2 conjugated to Sepharose 4B beads. The beads with bound proteoglycan were pelleted, rinsed and resuspended either in digestion buffer (for enzymatic digestion of the GAG chains on unlabeled samples) or in Laemmli sample buffer (for SDS–PAGE of ³⁵S-labeled samples). GAG chains were removed from the core protein by incubating the bead-bound syndecan proteoglycan in a mixture of heparitinase and chondroitinase-ABC for Ih at 37 °C as described previously (Sanderson and Bernfield, 1988). Whole proteoglycan or isolated core protein samples were run on either 3.8–15% or 3.5–20% polyacrylamide gradient gels. For whole proteoglycan, the gels were fixed, dried and exposed to X-ray film. Core protein samples were electrophoretically transferred to Gene-Trans, a cationic nylon filter, which was incubated with ¹²⁵I-labeled mAb 281-2 and then exposed to film.

Syndecan was first detected, using polyclonal anti- syndecan antibodies, in 4-cell embryos. From the 4-cell stage to the 16-cell morula, it was found intracellularly and on the external surface of the blastomeres, but not in areas of cell–cell contact (Fig. 1A–C). By the late morula stage (16–32 cells), syndecan was detected in a few areas of intercellular contact, often at the interface between outer and inner cells (Fig. 1D). In the unhatched blastocyst (120 h post-hCG), staining for syndecan was seen in regions of intercellular contact throughout the embryo, as well as on the external surfaces of the trophectoderm (Fig. 1E).

A major reorganization in syndecan distribution occurred with the segregation of primitive endoderm from the inner cell mass at the late blastocyst stage. Syndecan became restricted to the interface of the primitive ectoderm and primitive endoderm with some discontinuous patches on the blastocoele surface of the trophoblast cells, and was no longer detectable on the external surface of the trophoblast (Fig. 1G). Intracellular staining for syndecan was detected in the primitive ectoderm (Fig. 1H,I). The monoclonal antibody 281-2 did not detect syndecan in preimplantation embryos prepared for immunohistochemistry under the conditions used, suggesting that its epitope is inaccessible at these early stages.

In the embryonic ectoderm of the postimplantation embryo (5.5–8.5 days), syndecan was expressed on the basolateral surfaces of all cells at all stages examined; no regional differences in staining were detected (Figs 2, 3, and 4). Syndecan was not detectable in the visceral endoderm (Figs 2, 3, and 4), but was expressed by the squamous definitive embryonic endoderm that replaces the visceral endoderm in the embryonic region during gastrulation (Fig. 4). As in the embryonic ectoderm, syndecan was distributed on the basal and lateral cell surfaces of the definitive embryonic endoderm and no regional differences in its expression were detected.

Posterior and lateral mesoderm stained more intensely for syndecan than did anterior and mid-dorsal mesoderm. Syndecan was expressed by cells in the primitive streak, and in lateral and distal mesoderm in the 7.5-day embryo (Fig. 4A and B). In contrast, syndecan staining was barely detectable in the anterior mesoderm of the head fold (Fig. 4C) and in the head process and its derivative, the notochordal plate (Fig. 5). The head process originates at the anterior tip of the primitive streak and gives rise to the notochordal plate during primitive streak regression (Tam and Meier, 1982; Poelmann, 1981). The same asymmetry of syndecan expression was seen in the 8.5-day embryo; syndecan was strongly expressed in the primitive streak, in presomitic mesoderm and in fully formed somites, but was undetectable in the mesenchyme underlying the anterior neural ectoderm (Fig. 6). The change in expression was fairly abrupt, and occurred at the level of the otic sulcus (Fig. 6A). The asymmetry in syndecan staining in the mesoderm was confined to the embryonic region; staining for syndecan was uniformly intense in all areas of the extraembryonic mesoderm (Fig. 8A).

Both primary and secondary trophoblast giant cells stained intensely for syndecan at all stages examined (Figs 7 and 8). The secondary giant cells form in the ectopiacental cone, where they differentiate from a central core of replicating diploid cells (reviewed in Rossant, 1986). Most, but not all, cells in the center of the core showed no staining for syndecan, while an annulus of cells in the zone between the central core cells and the more peripheral differentiated giant cells showed the most intense staining (Figs 7 and 8). The extent of staining of central core cells appeared to be age-related. In general, there was little or no staining in the central core cells of younger embryos (6.5 days) and increased staining in central core cells of older embryos (7.5–8.5 days), perhaps reflecting the gradual terminal differentiation of the cells in the core.

Syndecan was not detected in extraembryonic ectoderm cells at 6.5 days (Fig. 7). By 7.5 days, it was detectable in those extraembryonic ectoderm cells that contributed to the chorion as well as at the interface of the two layers of the chorion (Fig. 8C). Syndecan staining was detected at the interface of the embryonic ectoderm and overlying visceral endoderm at 7.5 days. Since no staining was detected in visceral (extraembryonic) endoderm at any stage, the syndecan present in this location is likely produced by the extraembryonic ectoderm and localized at the interface of the two cell layers.

The striking temporal changes in the pattern and intensity of syndecan staining that occurred during periimplantation development suggested that temporal and spatial modulation of syndecan structure might be occurring. We examined this possibility by determining the relative molecular mass of both the syndecan core protein and the intact proteoglycan in hatched blastocysts and in isolated ectoplacental cone.

The core protein immunoisolated from hatched blastocysts migrated as two bands of about 75×l0³M_r and 46×10³M_r (Fig. 9A, lane 1) while the core protein from 72 h embryo outgrowths ran as a single band at 75×l0³M_r (Fig. 9A, lane 2). Syndecan core protein isolated from a variety of adult epithelial tissues migrates as a single band of 69×l0³M_r (Sanderson and Bernfield, 1988). Since syndecan is known to be modified by oligosaccharides as well as by glycosaminoglycans, the difference in the apparent molecular mass between the embryonic and adult forms may be due to variation in the composition of these oligosaccharides (Weitzhandler et al. 1988). The 46×l0³M_r band detected in hatched blastocysts likely represents the core protein of the cleaved ectodomain of syndecan, which lacks the cytoplasmic and transmembrane domains (Jalkanen et al. 1987).

The intact proteoglycan from hatched blastocysts ran as a continuous region of radioactivity extending from about 70×l0³M_r to the top of the gel (>300×l0³M_T), with some additional material not entering the gel (lane 2). In contrast, syndecan precipitated from isolated, cultured ectoplacental cone ran as a more discrete region of radioactivity extending from 200–260×l0³M_r(lane 4). These results indicate that syndecan structure is modulated during peri-implantation development.

Syndecan has multiple distinct activities in vitro: it acts as a receptor for several interstitial matrix molecules (Koda et al. 1985; Saunders and Bernfield, 1988; Sun et al. 1989), it plays a role in organization of cells into epithelial sheets (Saunders et al. 1989a; Jalkanen et al. 1990), it is nearly identical to a hamster low-affinity FGF receptor recently identified by ligand-affinity cloning (Keifer et al. 1990), and binds bFGF (Bernfield and Sanderson, 1990; Krufka et al. 1990). Determining which of these properties are important for particular developmental events is speculative, but can be approached through a comparison of syndecan expression patterns with known syndecan functions.

The distribution of syndecan in the late blastocyst and in the embryonic ectoderm of the early postimplantation embryo is consistent with a role in cell–cell or cell–matrix interactions that may stabilize epithelial morphology. Starting with endoderm segregation at the late blastocyst stage, syndecan becomes localized to the interface between embryonic ectoderm and visceral endoderm. This pattern is identical to that of the first basement membrane of the embryo, as shown by the distributions of fibronectin, laminin and collagen type IV (Zetter and Martin, 1978; Wartiovaara et al. 1979; Leivo et al. 1980; Wu et al. 1983). The primitive endoderm cells are the source of the matrix, while our data showing the presence of syndecan staining on the cells of the embryonic ectoderm, but not the visceral endoderm, suggest that the ectoderm is the source of syndecan (Figs 1–4). How syndecan redistributes at the time that the primitive endoderm segregates is not known, but may involve selective shedding of the syndecan ectodomain from the apical plasma membrane and retention of intact syndecan at the basal surface. Selective shedding of syndecan has been shown to account for its absence on the apical surfaces and its accumulation at the basolateral surfaces of confluent epithelial monolayers (Rapraeger et al. 1986; Jalkanen et al. 1987). In support of this idea, the syndecan core protein immunoprecipitated from late blastocysts (just after endoderm segregation) contains two components: a 75 × 10³M_r polypeptide whose size is consistent with that of the full-size core protein, and a 46×l0³M_r polypeptide whose size is similar to that of the core protein of the ectodomain alone (Jalkanen et al. 1987).

In considering to what extent syndecan promotes and maintains epithelial organization in embryonic and adult tissues, it is important to note that its presence is not required for the epithelialization process in general, and its presence is not restricted to epithelial cells during development. For example, there are no changes in the distribution of syndecan during compaction at the 8-cell stage, nor does it become localized to the basolateral cell surfaces of the trophectoderm, the first epithelium formed in the embryo. Furthermore, syndecan is not detectable in the epithelial visceral endoderm layer, and is expressed in nonepithelial ceils such as mesoderm and trophoblast giant cells (discussed below). Therefore, its roles must be related to specific functions that are undertaken by the cells in which it is expressed.

Syndecan is expressed strongly on mesoderm cells even after they have left the primitive streak, suggesting that syndecan has an important function in early mesoderm. The asymmetry of syndecan distribution in the mesoderm suggests a role in anteroposterior patterning. Syndecan expression is weak on the most dorsal and anterior mesodermal structures of the gastrulating mouse embryo (head process, notoplate and anterior head mesoderm), and strong on the more posterior and lateral structures (primitive streak, somites). The potential significance of these data can be appreciated in the light of several other findings. Syndecan binds bFGF (Bernfield and Sanderson, 1990; Krufka et al. 1990), is nearly identical to a low-affinity receptor for FGF recently cloned from hamster (Keifer et al. 1990) and, as a low-affinity receptor, can regulate the biological effects of bFGF (Yayon et al. 1991; Bernfield and Hooper, 1991; A.C. Rapraeger, personal com munication). There is evidence in other developmental systems that members of the FGF family of growth factors play an important role in specifying the fate of prospective mesoderm (Smith, 1989; Melton and Whitman, 1989; Mitrani et al. 1990). For example, in the amphibian Xenopus laevis, treatment of animal cap expiants with bFGF results in the induction of posterior/lateral types of mesodermal structures (Slack et al. 1987; Kimelman and Kirschner, 1987), whereas treatment with activin or TGF-β2 results in the induction of anterior/dorsal mesodermal structures (Smith, 1987; Rosa et al. 1988; reviewed in Smith, 1989). The timing and mechanism of mesoderm specification have not been established in the mouse embryo; however, it is known that at least one member of the FGF family, int-2, is expressed during gastru- lation in the posterior primitive streak (Wilkinson et al. 1988). Taken together, these data are consistent with the hypothesis that the asymmetric distribution of syndecan observed in the mouse gastrula could spatially restrict the activity of member(s) of the FGF family involved in generating an anterior-posterior axis in mouse mesoderm.

The patterns of syndecan expression in the ectopiacental cone and trophoblast giant cells are consistent with its proposed role as a receptor for heparin-binding growth factors (Bernfield and Sanderson, 1990; Keifer et al. 1990; Klagsbrun, 1990). The high level of syndecan expression by trophoblast cells may reflect a role for such growth factors in trophoblast function during implantation. Trophoblast giant cells are invasive, migrating through the uterine epithelium to establish the embryo in the stroma (Billington, 1971; Welsh and Enders, 1987), where they then form an intricate network of bloodspaces, and ultimately, the fetal portion of the mature chorioallantoic placenta (Rossant and Croy, 1985; Rossant, 1986). Trophoblast cell function thus involves motility, protease secretion and neovascularization, all of which are promoted by bFGF in cultured cells (Rifkin and Moscatelli, 1989).

In summary, it is likely that syndecan has several functions in the early development of the mouse embryo. Its distribution in the late blastocyst and the embryonic ectoderm of the postimplantation embryo is consistent with a role in matrix attachment and epithelial organization, whereas its distribution in the embryonic mesoderm and on trophoblast giant cells is consistent with its proposed role as a receptor for FGF (Keifer et al. 1990; Klagsbrun, 1990; Bernfield and Hooper, 1991). Our results, showing a difference in the relative molecular mass of syndecan between blastocysts and trophoblast giant cells, suggest that the spatial and temporal regulation of syndecan structure may reflect these potentially disparate functions. Previous studies have demonstrated that the relative molecular mass of the intact syndecan proteoglycan differs in simple and stratified epithelia (Sanderson and Bernfield, 1988), and changes during B-cell differentiation (Sanderson et al. 1989) and with epithelial– mesenchymal interactions (Boutin et al. 1988). These size differences are due to differences in the number and length of heparan sulfate and chondroitin sulfate chains attached to the core protein (Sanderson and Bernfield, 1988). Thus, variations in GAG composition may determine whether the syndecan present in different locations functions in matrix attachment, epithelial organization and/or growth factor binding. The data presented here on the pattern of syndecan expression during peri-implantation development provide a basis for future experiments designed to determine more directly its function in the early embryo.

We thank Dr Irving Weissman for the generous gifts of antibodies; Dr Ray Keller for use of his microscope and for critical comments on the manuscript and Dr Gerry Kidder for the suggestion of DGD as an embedding medium. We also thank Dr Barry Gumbiner, Dr Mike Frohman and Dr Ann Poznanski for critical comments on the manuscript. Mercedes Jo ves provided expert technical assistance. The work was supported by NIH grants HD22593, CA28735, HD06763 and HD06703, and by the Arthritis Foundation.