Acellularity of starfish embryonic mesenchyme cells as shown in vitro

In early embryogenesis of many multicellular organisms, mesenchyme cells differentiate from epithelial cells and migrate through the extracellular matrix (ECM) of the blastocoel as individual cells before participating in morphogenetic activities. Mesenchyme cells of vertebrate embryos have been extensively studied; e.g. their transformation from epithelial cells (Greenburg and Hay, 1986, 1988), the mechanism of their migration (Thiery, 1984), interaction with epithelial cells (Haffen et al. 1987; Cunha, 1984; Dhouailly and Sengel, 1983) and the ECM (Tomasek et al. 1982) etc. However, studies of invertebrate mesenchyme cells are still very limited.

The primary mesenchyme cell of the sea urchin embryo, which specifically functions in the formation of spicules, is probably the most extensively studied mesenchyme cell of invertebrate origin. In contrast, the secondary mesenchyme cell, which has more general functions in embryogenesis such as a role in gastrulation (Dan and Okazaki, 1956; Gustafson and Kinnander, 1956; Spiegel and Burger, 1982), has received less attention. The single type of mesenchyme cell found in most animal bodies composed of three germ layers seem to correspond with the secondary mesenchyme cell.

In the starfish embryo, the mesenchyme cells form at the tip of the embryonic archenteron and migrate into the blastocoel (Chia, 1977) in a manner similar to the secondary mesenchyme cells of the sea urchin. These cells eventually take part in the formation of the mouth and coelomic pouches (Crawford and Abed, 1978, 1983) and form ‘fibrous networks’ on the ectodermal wall and on the surface of the larval gut (Chia, 1977).

In order to elucidate the nature and function of these cells more clearly, we have observed them morphologically, dynamically and ultrastructurally under culture conditions. They form a large syncytium at the beginning of the culture, instead of migrating singly as found in vivo. The large syncytium, however, eventually spread out into a network of small syncytial fragments interconnected by fine cell processes. Our results indicate that the mesenchyme cells are fundamentally acellular in nature, with no cell boundaries between the cell bodies or the cell processes. We also show that the network is three-dimensional in nature with crossing points. We have also studied living embryos to see if the network pattern of the mesenchyme cells in vitro is also present in vivo. The acellularity of these cells is discussed in relation to their physiological role in the embryogenesis.

Developing embryos of the starfish, Asterina pectinifera, were obtained by the method described previously (Dan-Sohkawa, 1976), in which the eggs treated with 1-methyladenine (1-MA) (Kanatani, 1969) were inseminated with diluted dry sperm and allowed to develop in the artificial sea water, Jamarin U (Jamarin Lab., Osaka, Japan), at 20°C. Embryos were collected at various stages of development by hand centrifugation and were submitted to further treatments. The nomenclature of the developmental stages of the starfish follows that presented previously (Dan-Sohkawa et al. 1980).

Archenteron preparation medium (APM): 1.2 M glycine in distilled water (DW) supplemented with 1/100 volume of Jamarin U (Dan-Sohkawa and Kaneko, 1989).

Culture medium: Jamarin U supplemented with 4 % newborn bovine serum (M. A. Bioproduct Lab.).

Detaching medium: Ca²⁺/Mg²⁺-free Jamarin (Jamarin Lab., Osaka, Japan).

Archenteron complexes (archenterons and varying portions of the ECM surrounding them) were prepared by the method reported previously (Dan-Sohkawa and Kaneko, 1989), in which 10 ml of APM was added to a centrifuge tube containing 0.4 ml of packed embryos at the mouth formation stage. After 5 min of treatment, the medium was changed to 4 ml of the same medium, in which embryos were submitted to 10 to 15 gentle strokes of a broad-mouthed pipette (Komagome’s pipette; tip inside diameter 1.5−2 mm) to remove the ectoderm. Archenteron complexes were separated from dissociated ectodermal cells and fragments by hand centrifugation, washed twice with 10 ml of Jamarin U and suspended in 3 ml of the culture medium. Various amounts of this suspension, ranging from one drop to 1ml, was inoculated in final 1 to 1.5 ml of the culture medium, and cultured at 20°C on to different substrata, as explained below for each case.

The network cultured for 24 h in a plastic dish (35 mm in diameter; Falcon no. 3801) was exposed to 0.5 % of methylene blue in culture medium for 15 min, in order to stain the cytoplasmic granules of the constituent mesenchyme cells. The stained cells were detached from the surface by incubating them for 10 min with the detaching medium and two successive washes with the same medium. The suspension was centrifuged at 1500 revs min^-1 for 5 min to collect the stained cells. After removal of the supernatant, they were reinoculated over the network of unstained cells.

The morphology and behavior of the network was observed with a phase-contrast microscope (Nikon TMD). The behavior of the stained granules in cocultures of stained and unstained cells, was traced under the bright-field optics (Goshima et al. 1984).

In order to observe the nuclei of the cultured mesenchyme cells, the archenteron complexes were inoculated on a glass coverslip (18×18 mm) and the resultant network was cultured for 24 h. The network was treated for 30 min with final 2 μgml^-1 of Hoechst 382065 (Calbiochem Lab.) added to the culture medium, and washed briefly with Jamarin U. It was then mounted on a slide glass and the nuclei were located under a phase-constrast microscope coupled with fluorescence equipment (Nikon TMD).

For TEM observations, the network cultured on a plastic dish for 24 h was washed and fixed with 3 % glutaraldehyde in 0.1 M cacodylate buffer (pH 7.2) for 30 min at 4°C. The network was washed for several times in the same buffer, postfixed with 1 % OsCO4 in DW for 30 min at 4°C, and washed with DW at 4°C. The network was then stained en bloc for 30 min with 2% aqueous uranyl acetate at 20 °C, followed by another wash with 0.1 % sodium acetate to remove remaining uranyl acetate. After dehydration by passage through a graded ethanol series, the network was removed from the plastic dishes with propylene oxide and centrifuged in the same solvent to form a pellet, which was finally embedded in Supper’s low viscosity resin, and ultrathin sections cut. The ultrathin sections were stained with 3% uranyl acetate in 30% ethanol followed by treatment with Reynold’s lead citrate and observed with a H-600 transmission electron microscope (Hitachi).

For SEM observations, the archenteron complexes were inoculated on tissue culture coverslips (13 mm in diameter, Thermanox, Miles Lab. Inc.) and the resultant networks were cultured for 24h. They were washed with Jamarin U to remove the culture medium, fixed with 2% OsO₄ in Jamarin U for 15 min at 20°C and rinsed with DW for 15 min. After dehydration through an ethanol series, the specimens were critical-point dried (Hitachi HCP-2), sputter-coated with gold (Eiko IB-3), and examined by a scanning electron microscope (JEOL T-300).

In order to search for a mesenchymal network in vivo, embryos were anesthesized with a small amount of 4 % paraformaldehyde added from the sides of the glass coverslip immediately before taking photographs and were observed under a differential interference microscope (Olympus IMT-2-21 RFN).

Formation and dynamics of mesenchymal network in vitro Immediately after preparation, the mesenchyme cells are scattered singly in the ECM, which surrounds the archenteron (Fig. 1). The ECM becomes deflated within an hour and the archenteron attaches to the surface of the substratum. The mesenchyme cells, on the other hand, escape from the ECM and form a large syncytium instead of migrating singly on the substratum. Within one hour following attachment, the syncytium can be seen as a large lamellipodium migrating from underneath the archenteron (Fig. 2A). It steadily migrates away from the archenteron and breaks into smaller fragments (Fig. 2B-D). By 24h, a typical network is formed around the archenteron (Fig. 2E).

The network is composed of two parts, the cell body and the cell process (Fig. 3A). The cell body is elongated often along the paths lamellipodia and filopodia. The majority of the cell bodies contain one nucleus, whereas some contain 2-6 and still others none (Fig. 3B). Fine cell processes connect two cell bodies or cell bodies and lamellipodia. Some of the processes are branched, some are free ended and still others have flattened parts (arrowheads in Fig. 3A).

Fig. 4 depicts an optic field photographed at various intervals, showing that the network changes its pattern actively with time. This change is mainly caused by an active movement of the cell bodies. New cell processes, however, are continually formed in two ways. (1) When a cell body breaks into two smaller fragments, a new cell process remains between the two (arrows in Fig. 4A-F). (2) When two independent cell processes are extended and come in contact, a new cell process is formed by fusion of the two at their distal ends (arrowheads in Fig. 4D-E, Fig. 6B,C).

To elucidate further the syncytial nature of the mesenchyme cells, movements of cytoplasmic inclusions were traced. Fig. 5 shows cells with their cytoplasmic granules stained with methylene blue inoculated over a sparse network of unstained cells. Fig. 6 illustrates schematically the movement of the upper three cells shown in Fig. 5 along with that of the stained granules and the nuclei. The cells exchange their cytoplasmic inclusions as they repeat fusion and separation. Actually, three mononucleate cells containing the stained granules and one cell containing unstained granules were produced from two mononucleate stained cells and a binucleate unstained cell during the 68 min span of observation. It is also noteworthy that a nucleus enwrapped in a small mass of cytoplasm can be translocated through the cell process (Fig. 6C-G). An anucleate fragment sustained locomotive activity (Fig. 6F,G). Although solitary epithelial cells containing stained granules, probably of archenteron origin, were sometimes found to contaminate these cultures, they were easily distinguished by their rounded shape and the presence of a cilium. Fusion and separation never took place between the epithelial cells and the mesenchyme cells.

The syncytial nature of the mesenchyme cell in the network was further confirmed by TEM. Fig. 7 shows a typical region of the network where two cell bodies are connected by a cell process. No cell boundary is detectable between the two nuclei in such areas. The lack of cell boundaries is confirmed in a wider area by SEM (Figs 8, 9).

Another characteristic of the network revealed ultrastructurally is its three-dimensional nature. A typical example is shown by SEM in an area where the cell bodies are crowded (Fig. 8). Many overlapping sites of cell bodies and cell processes are seen. In order to determine whether or not the three-dimensional configuration of the network depends on the degree of crowding of the cells, we prepared networks of various condensations by changing the density of the archenteron complexes. Unexpectedly, the three-dimensional configuration depended on the density of the initial archenteron complexes rather than the degree of crowding of the cells themselves. That is, even when the crowding of the cells is similar (Fig. 9), a greater number of overlapping sites are present in a network resulting from the culture of a larger number of the archenteron complexes (Fig. 9B, 33 overlapping sites, more ECM), as compared to that obtained from a smaller number of them (Fig. 9A, 5 overlapping sites, less ECM).

In normal embryos, the mesenchyme cells are relatively spherical at the mouth formation stage (Fig. 10A), bearing only short or inconspicuous processes (Fig. 10B,C). As the development proceeds, they increase in number to approximately 110 at the bipinnaria stage (Fig. 10D) (Kominami, 1985) and the cells transform to an irregular shape. They extend prominent, branched processes, which spread out into small lamel-lipodia at some places (Fig. 10E,F). In some regions of the blastocoel, 2 to 10 cells are seen to form a local network (Fig. 10G,H).

When embryonic mesenchyme cells of the starfish are cultured by inoculating archenteron complexes separated from the ectoderm in sea water, they behave in an acellular manner. The mesenchyme cells escape from the ECM, form a large syncytium on the substratum, which eventually fragments into smaller and smaller pieces until it becomes a network of small acellular fragments interconnected by fine cell processes after 24h of culture (Fig. 2). The pattern of the network changes constantly, driven mainly by the movement of the cell bodies (Fig. 4). Cytoplasmic granules and nuclei are seen to translocate between the fragments during this movement (Figs 5, 6). The lack of cell boundaries between the fragments is confirmed by TEM observations (Fig. 7). The highly acellular nature of these cells, in vitro, is shown most dramatically in the formation and absorption of anucleate fragments at any site of the network. This event is also easily observed in a sparse culture whose major component is mononucleate (Fig. 6G,H). The acellularity of these cells seems to depend on the highly flexible nature of their plasma membrane. We know, however, nothing about this membrane, at present.

The degree of acellularity of these cells is considered much stronger than that of the primary mesenchyme cells of the sea urchin. That is, the cell bodies of the latter are always clearly distinguishable from their processes, as are beads on a string, both in vivo (Akasaka et al. 1980; Gibbons et al. 1969; Okazaki, 1960, 1965) and in vitro (Decker et al. 1987). We consider that the mesenchyme cells observed by Karp and Solursh (1985),in vitro were secondary not primary mesenchyme cells for the following three reasons. (1) Their cells are morphologically similar to ours rather than to the primary mesenchyme cells. (2) Their manner of fusion is similar i.e. the cell processes and cell bodies fuse. This mode of fusion is not reported, to our knowledge, in the primary mesenchyme cells. (3) Both of these cells form anucleate fragments, which are indistinguishable from fragments containing the nucleus (Fig. 3A: arrow), a fact also not reported for primary mesenchyme cells. So far as we know, the only other cell type morphologically similar to ours is the network of cultured cells of the respiratory system of Holothuria sp. (Rannou, 1971).

Our results show that the network is fundamentally three-dimensional in nature (Fig. 8) and that the number of crossing sites is likely to depend on the amount of environmental ECM (Fig. 9). Although further experimentation is necessary to confirm this point, it suggests a mechanism by which the ECM supports the apparently ‘cellular’ state of the mesenchyme cells in vivo. The local net-like conformation of mesenchyme cells was found in vivo not only by us (Fig. 10) but also by Chia (1977), supporting the possibility that these cells actually form a disperse, three-dimensional, acellular network in vivo.

One of the most important questions to arise from the present study is what physiological function(s) of the normal development does the acellularity of the mesenchyme cells reflect. Assuming that the acellularity is not an artifact due to in vitro conditions, two possibilities are considered. (1) It may reflect the manner in which mesenchyme cells exchange and process positional information in morphogenetic events. For example, the mesenchyme cells participating the formation of the mouth (Crawford and Abed, 1983) may form a local network that extends over the blastocoel between the ectoderm and endoderm of the presumptive mouth region (cf. Fig. 8 of Crawford and Abed, 1983). The situation may also be true for those shown in Fig. 10B and C of this paper. Such a network will not only be efficient for transmitting information between distant cells, but will also stabilize the field of morphogenesis. A physical role, similar to that of connective tissue, is also suggested for the ‘fibrous network’ of mesenchyme cells (Chia, 1977).

Another possible function is that the mesenchyme cells play a defensive role in the larval body. This speculation is based on the reported fact that embryonic mesenchyme cells of the starfish show a phagocytic activity in vivo (Metchnikoff, 1968; Kominami, 1985), which sometimes results in the formation of a large syncytium in the blastocoel of normal embryos (Metchnikoff, 1968). This idea, in turn, allows us to interpret the formation of large syncytia, at the beginning of the culture, as a reaction of phagocytes against contact with a foreign body: a reaction commonly seen in vertebrate monocytes in variety of inflammatory reactions (Papadimitriou, 1978). However, further biochemical and comparative studies are necessary to verify this possibility.

We thank Drs Jun-ichi Taniguchi and Chizuko Koseki of the National Cardiovascular Center Research Institute for kindly offering us the use of their differential interference microscope. Staff members of Asamushi Marine Biological Station of Tohoku University, of Ushimado Marine Laboratory of Okayama University and of Noto Marine Laboratory of Kanazawa University were generous enough to supply the starfish and to allow us to use their facilities during the course of this investigation. Thanks are also due to Mr Noriya Miyata for preparing photographs.