The role of micromere signaling in Notch activation and mesoderm specification during sea urchin embryogenesis

In the embryos of many sea urchin species, the mesoderm is subdivided into the primary mesenchyme cells (PMCs), which ingress from the vegetal plate before archenteron formation, and the secondary mesenchyme cells (SMCs), which form from the archenteron throughout gastrulation. The PMCs are the sole descendants of the large micromeres and form only the skeletogenic mesenchyme. The SMCs differentiate into non-skeletogenic mesoderm including blastocoelar cells, pigment cells, coelomic pouch cells and circumesophageal muscle cells. The SMCs derive almost exclusively from veg2 descendants (Hörstadius, 1973; Cameron et al., 1987, 1991; Logan and McClay, 1997), though the small micromeres contribute to the coelomic pouches (Pehrson and Cohen, 1986). With the possible exception of the small micromere derivatives, cell-cell interactions are thought to be involved in the specification of the SMCs. Veg2 derivatives give rise to the vegetal plate territory which consists of prospective endoderm and SMCs (Hörstadius, 1973; Ruffins and Ettensohn, 1996; Logan and McClay, 1997). There is evidence that the micromeres induce overlying cells to form the endodermal component of the vegetal plate territory (reviewed by Davidson et al., 1998).

When the micromeres are removed at the 16-cell stage, there is a significant decrease in the number of cells that express the vegetal plate- and endoderm-specific marker Endo16, and gastrulation is delayed (Ransick and Davidson, 1995). Because the Endo16 marker cannot be used to distinguish between prospective endoderm and mesoderm, it was unclear whether micromere signaling is necessary for the normal development of SMCs.

The signaling properties of the micromeres have also been demonstrated using transplantation experiments. Mesomeres at the animal pole of the embryo generally produce only ectoderm during normal development (Hörstadius, 1973; Cameron et al., 1987, 1991; Logan and McClay, 1997). When micromeres are transplanted to the animal pole at the 32-cell stage, they can induce an ectopic archenteron (Hörstadius, 1973). The micromeres have a stronger inductive effect if they are transplanted to the animal pole at the 8-cell stage (Ransick and Davidson, 1993). Animal halves produce only ectoderm in isolation (Hörstadius, 1973; Henry et al., 1989; Wikramanayake et al., 1995; Wikramanayake and Klein, 1997), but pigment cells are induced when animal cells are recombined with micromeres (Khaner and Wilt, 1991; Amemiya, 1996). Thus, the micromeres are sufficient to induce both endoderm and pigment cells.

Recent studies have begun to address the molecular mechanisms involved in the specification of the SMCs and have demonstrated a role for the Notch signaling pathway. Beginning at the blastula stage, LvNotch protein is downregulated in prospective SMCs as expression at the cell surface decreases, while there is an increase in staining within intracellular vesicles (Sherwood and McClay, 1997, 1999). Overexpression of a constitutively active form of LvNotch causes an increase in the domain of LvNotch downregulation and an increase in SMCs, while overexpression of a dominant negative form reduces the domain of LvNotch downregulation and causes a significant decrease in the numbers of SMCs (Sherwood and McClay, 1999). Thus, the loss of LvNotch expression is associated with the activation of the Notch signaling pathway and SMC specification.

The purpose of this study is to examine the roles of micromere signaling in Notch activation and SMC specification. Normal staining patterns are described for two new mesoderm-specific monoclonal antibodies. Micromere removal experiments demonstrate that micromere signaling is necessary for the specification of many non-skeletogenic mesodermal cell types. Transplantation experiments demonstrate that the micromere-derived signal is sufficient to induce all SMC types examined. Finally, micromere signaling is shown to be necessary for the downregulation of LvNotch protein, providing a link between micromere signaling, Notch signaling and mesoderm specification.

Adult Lytechinus variegatus were obtained from Susan Decker in Davie, FL, and Tracy Andacht and Jennifer Jackson at the Duke University Marine Laboratory, Beaufort, NC. Gametes were obtained by injecting 0.5 M KCl into the adult coelomic cavity. Embryos were raised in artificial sea water (ASW; Instant Ocean) or natural sea water (provided by Gail Cannon). For most experiments, embryos were raised at 23°C.

To remove the fertilization envelope, eggs were fertilized in sea water with 0.5 mM 3-amino-1,2,4-triazole, washed, and later passed through a pulled pipette with a diameter slightly smaller than the fertilization envelope. All micromanipulations were performed within a 10 minute time interval. Micromere removal was performed according to the technique of Ransick and Davidson (1995). Immediately after the third cleavage, embryos were placed in hyaline extraction medium (HEM; McClay and Fink, 1982) in a 35 mm Petri dish coated with 2% agar. Micromeres were removed with a hand-pulled glass needle and the resulting micromere(−) embryos were transferred to sea water in an agar-coated dish. Experimental embryos were examined at a later time to ensure that subsequent cleavages were normal.

The technique for micromere transplantations was based on methods used by Ransick and Davidson (1993) and Amemiya (1996). Donor embryos were labeled with rhodamine isothiocyanate (RITC; Ettensohn and McClay, 1988). For transplantations to the animal pole, labeled 16-cell stage embryos and unlabeled host 8-cell stage embryos were placed together in HEM in an agar-coated dish and quartets of donor micromeres were isolated. The micromeres and host embryos were transferred to an agar-coated dish with sea water and the micromeres were transplanted to the animal pole of the host embryo.

The animal pole can be identified in 8-cell stage embryos of this species because the animal cells are slightly larger than the vegetal cells. For transplantations to animal caps, the animal pole of 8-cell stage embryos was marked with carbon particles treated with 1 mg/ml poly-L-lysine or 1% protamine sulfate, the animal four cells were isolated, and RITC-labeled micromeres were isolated and transplanted to the vegetal pole of the animal cap. Experimental embryos were scored only if they contained all of the descendants of the micromeres.

PMC transplantation was performed according to the method of Ettensohn and McClay (1986, 1988). RITC-labeled mesenchyme blastula-stage embryos were loaded into microinjection chambers, a microneedle was inserted into the blastocoel and PMCs were pulled into the needle. After PMCs were collected from several embryos, they were expelled into the blastocoel of host embryos in a different microinjection chamber.

For whole-mount immunostaining, embryos were fixed in 3.7% formaldehyde in ASW for 10 minutes at room temperature, washed in ASW and postfixed in ice-cold 100% methanol for 5 minutes. The embryos were washed several times in phosphate-buffered saline (PBS) and incubated in primary antibody for 2 hours at room temperature or overnight at 4°C. Primary antibodies included SMC1 and SMC2 which are IgM monoclonal antibodies produced using an in vitro immunization method (Hodor, 1998; P. H. and C. A. E., unpublished data); 6a9 (Ettensohn and McClay, 1988) and 6e10 (Ingersoll, 1993), which are monoclonal antibodies that recognize skeletogenic mesenchyme; a polyclonal antibody against myosin (a gift from G. M. Wessel; Wessel et al., 1990); a polyclonal antibody against LvG-cadherin (CAD-1; a gift from D. R. McClay; Miller and McClay, 1997); and a polyclonal antibody against LvNotch (a gift from D. R. McClay; Sherwood and McClay, 1997). Monoclonal antibodies were used at full strength. Anti-myosin was used at a dilution of 1:100 in PBS, CAD-1 was used at 1:200 in all experiments except SMC1, anti-LvNotch, CAD-1 triple-staining where it was used at 1:1000 and anti-LvNotch was used at 1:100. After primary antibody incubation, the embryos were washed several times in PBST (PBS with 0.05% Tween), once in PBS and incubated in secondary antibody (1:50) for 2 hours at room temperature. Secondary antibodies included fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse IgM (for SMC1, SMC2 and 6a9), Texas Red-conjugated goat anti-mouse IgG (for 6e10), Texas Red-conjugated goat anti-rabbit IgG (for anti-myosin) and Cy3-conjugated goat anti-guinea pig IgG (for CAD-1 and anti-LvNotch; Jackson ImmunoResearch Laboratories, Inc. and ICN Biomedicals). After incubation in secondary antibody, the embryos were washed as described above and mounted in 1:1 PBS:glycerol with 2.5% 1,4-diazabicyclo[2.2.2]octane to prevent photobleaching. The embryos were examined with epifluorescence or laser scanning confocal microscopy. For double- and triple-labeling experiments, primary antibodies were pooled together and secondary antibodies were pooled together.

Developmental western blots were carried out as described by Towbin et al. (1979). Embryos were suspended in 7 volumes 2× SDS-PAGE sample buffer containing 100 mM dithiothreitol as a reducing agent and boiled for 3 minutes. Equal amounts of protein were determined by the method of Lowry (1951) and separated on 5-15% acrylamide linear gradient gels and electrotransferred to nitrocellulose sheets. The presence of proteins on the nitrocellulose was checked by staining with 0.1% Ponceau S in 5% acetic acid. Blots were incubated overnight at 4°C with hybridoma culture supernatants diluted 1:25 to 1:50 in 5% powdered milk in Tris-buffered saline (TBS). Blots were washed in TBS, incubated in horseradish peroxidase-conjugated goat anti-mouse immunoglobulins (Cappel), and visualized with the ECL chemiluminescence kit and Hyperfilm-ECL film (Amersham Corp). Controls consisted of blots incubated with secondary antibody alone.

Whole-mount immunofluorescence demonstrated that the SMC1 antigen was present in punctate structures in the apical end of cells in the vegetal plate beginning at the hatched blastula stage, 8 hours after fertilization (Fig. 1A; Table 1). At the mesenchyme blastula stage, SMC1-positive cells circled a region of unstained cells, probably the small micromere derivatives (Figs 1C, 2A,B). During early gastrulation, SMC1-positive cells remained at the tip of the archenteron (Fig. 1D). As gastrulation proceeded, however, the number of SMC1-positive cells decreased progressively and there was an increase in extracellular staining lining the blastocoel (Fig. 1E,F). The staining pattern suggests that SMC1 specifically recognizes an antigen in presumptive SMCs in the vegetal plate and archenteron. When SMCs ingress, the antigen may be released into the blastocoel. The SMC1 antibody did not yield a detectable signal when used for immunoblotting.

To determine more precisely which cells express the SMC1 antigen, the staining pattern was compared to the fate map of the vegetal plate. Late mesenchyme blastula stage embryos were double-stained with SMC1 and CAD-1, an antibody against LvG-cadherin that outlines cell borders (Fig. 2A). The vegetal plate contained about eight SMC1-negative cells surrounded by an average of 46 SMC1-positive cells (Table 1). At this stage, the vegetal plate contains approximately 66 macromere-derived SMC precursors and eight small micromere derivatives (Ruffins and Ettensohn, 1996). Thus, there are fewer SMC1-positive cells in the vegetal plate than macromere-derived presumptive SMCs. The number of SMC1-positive cells does not closely match the estimated number of precursors of any single SMC subtype (Ruffins and Ettensohn, 1996). The domain of SMC1 staining was also compared to the boundary of apical LvNotch expression that marks the boundary between presumptive endoderm and mesoderm (Sherwood and McClay, 1997). Late mesenchyme blastula stage embryos were triple-stained with SMC1, CAD-1 and anti-LvNotch (Fig. 2B). There was a sharp boundary between cells expressing apical LvNotch (prospective endoderm) and cells expressing the SMC1 antigen, further supporting the idea that SMC1-positive cells constitute most of the prospective mesoderm within the vegetal plate. However, there were a few SMC1-negative/LvNotch negative cells, suggesting that not all prospective mesodermal cells express the SMC1 antigen.

At the mesenchyme blastula stage, the SMC1-positive cells were positioned symmetrically around the vegetal pole in most embryos (Fig. 2A,B). Just prior to the onset of archenteron invagination, however, staining was lost on one side of the vegetal plate. To determine the polarity of SMC1 staining, early to mid-gastrula stage embryos were double-stained with SMC1 and 6e10, an antibody specific for skeletogenic cells which are distinctively patterned along the oral/aboral axis at this stage (Fig. 2C). SMC1-positive cells were absent from the side opposite the ventrolateral clusters of PMCs, indicating that SMC1 staining was lost from the aboral side of the vegetal plate.

In a developmental Western blot, the SMC2 antibody recognized a prominent band of about 300 kDa, which was first present at the hatched blastula stage and persisted through the pluteus stage (Fig. 3A). Whole-mount immunostaining demonstrated that all cells at the hatched blastula stage stained weakly with SMC2 (Fig. 3B) but, just prior to PMC ingression, staining was strongest in cells at the vegetal pole (Fig. 3C). At the mesenchyme blastula stage, SMC2 stained PMCs (Fig. 3D). Throughout gastrulation, PMC staining was lost and SMCs strongly expressed the SMC2 antigen as they formed at the tip of the archenteron (Fig. 3E-G). Pigment cells within the ectoderm transiently stain with SMC2 (Fig. 3F) but, at the prism and pluteus stages, SMC2 stains only blastocoelar cells (Fig. 3H-J).

To determine whether a micromere-derived signal is necessary for SMC specification, the micromeres were removed immediately after they formed at the 16-cell stage. The vegetal plate formed on schedule but did not invaginate until 16 hours after fertilization, at a time when controls were at the mid-to late gastrula stage. Mesenchyme cells began to form from the archenteron by 20 hours in micromere(−) embryos.

Antibody staining with SMC1 demonstrated that the onset of expression of the SMC1 antigen in micromere(−) embryos did not occur at the same chronological time as controls, nor at the equivalent developmental stage (Fig. 4; Table 1). In six batches, almost no SMC1-positive cells developed. Overall, 82% of all embryos had no staining (94/115; Fig. 4A-D). In one batch, the SMC1 antigen was expressed in low numbers but only at the equivalent of the late gastrula stage (32-36 hours after fertilization; Fig. 4E,F). Thus, micromere signaling is necessary for the development of SMC1-positive cells. Even though the development of SMC1-positive cells was rare, the micromere(−) embryos did go on to develop several SMC subtypes, suggesting that the expression of the SMC1 antigen is not necessary for SMC development under these conditions.

Normal pluteus larvae have about 100 pigment cells, which begin to accumulate at the early to mid-gastrula stage, between 12 and 16 hours after fertilization (Table 1). In seven out of ten batches of embryos, no pigment cells developed following micromere removal. Of all late stage embryos (40 to 96 hours after fertilization in Table 1), the average number of pigment cells was only 7 per embryo and 58% never developed pigment cells (Fig. 5; Table 1), demonstrating that micromere signaling plays a critical role in pigment cell specification.

Muscle cells first begin to differentiate at the prism stage, about 24 hours after fertilization (Wessel et al., 1990). Micromere(−) embryos began to express myosin by 36 hours after fertilization. All late stage embryos examined developed this cell type and they formed more muscle cells than controls. This increase was small, but statistically significant (Fig. 6; Table 1).

Blastocoelar cells stain with SMC2 and normally begin to accumulate by 16 hours after fertilization (Table 1). As young PMCs also stain with SMC2 in normal embryos (Figs 3D, 7A), it was necessary to double-stain micromere(−) embryos with SMC2 and 6e10 to distinguish skeletogenic cells from blastocoelar cells. Distinct SMC2-positive/6e10-negative cells began to accumulate by 24 hours after fertilization (Fig. 7C; Table 1), and they formed in all five batches examined. By 44 hours, they resembled blastocoelar cells (Fig. 7D). Of all late stage embryos (40 to 96 hours), 27% never developed this cell type (8/30). Overall, late stage micromere(−) embryos formed an average of only 11 blastocoelar cells per embryo, which is significantly less than the number found in control larvae (Table 1). We conclude that micromere signaling is necessary for the normal development of blastocoelar cells.

Micromeres give rise to about 60 PMCs, which are normally the first mesenchymal cells to differentiate at about 10 hours after fertilization. Skeletogenic cells were also the first mesodermal cell type to form in micromere(−) embryos, although they did not arise until about 20 hours after fertilization (Fig. 7B; Table 1). Skeletogenic cells developed in embryos from all 16 batches examined. Of all late stage embryos, 10% never developed skeletogenic cells (4/41), but the average number of skeletogenic cells in this older group was 53 per embryo, which is not significantly different from controls (Table 1). By 32 hours in most micromere(−) embryos, skeletogenic cells formed a ring and ventrolateral clusters (Fig. 7D), indicating that the ectodermal signals that pattern these structures were present by this time.

When the PMCs are removed from mesenchyme blastula stage embryos, skeletogenic cells form at the apparent expense of pigment and blastocoelar cells (Fukushi, 1962; Ettensohn and McClay, 1988; Ettensohn and Ruffins, 1993). One interpretation of these results is that the PMCs normally suppress some prospective pigment and blastocoelar cells from becoming skeletogenic cells. In some ways, the phenotype of micromere(−) embryos is similar to the phenotype of PMC(−) embryos. Following both manipulations, skeletogenic cells develop from SMCs and there is a decrease in the numbers of pigment and blastocoelar cells. As the PMCs derive from the micromeres, when the micromeres are removed, so essentially are the PMCs. Thus, it was necessary to determine whether the results of micromere removal were secondary effects due to the absence of the PMCs. Specifically, we tested whether PMCs added to micromere(−) embryos can suppress skeletogenic cell development and promote pigment and blastocoelar cell development. The micromeres were removed at the 16-cell stage and RITC-labeled PMCs were transplanted into the blastocoel when control embryos were at the mesenchyme blastula stage. Micromere(−)/PMC(+) embryos were fixed at 24 to 32 hours after fertilization and stained with the skeletogenic cell specific antibody 6a9 and an FITC-conjugated secondary antibody. The number of host-derived skeletogenic cells (FITC-labeled only) was compared to the number of donor-derived PMCs (FITC- and RITC-labeled). With increasing numbers of transplanted PMCs, fewer skeletogenic cells developed from the host, indicating that the donor PMCs suppressed host cells from becoming skeletogenic (Fig. 8A-C). The micromere(−)/PMC(+) embryos were next examined at the pluteus stage to see whether the numbers of pigment cells and blastocoelar cells were rescued. Micromere(−)/PMC(+) embryos developed few pigment or blastocoelar cells (Fig. 8C-F). Thus, in micromere(−) embryos, unlike PMC(−) embryos, skeletogenic cells do not develop at the expense of pigment and blastocoelar cells. The fate of the cells which were blocked from becoming skeletogenic in micromere(−)/PMC(+) embryos is unknown. Because the added PMCs did not rescue the mesodermal deficiencies of micromere(−) embryos, this strongly supports the view that a distinct, micromere-derived signal is indeed necessary for the development of pigment and blastocoelar cells.

To test whether the micromeres are sufficient to induce animal cells to develop non-skeletogenic mesoderm, a quartet of micromeres was transplanted to the animal pole of 8-cell stage embryos and these experimental embryos were stained with the SMC1 antibody at the early to mid-gastrula stage. Three out of ten experimental embryos formed a small, ectopic archenteron at the animal pole and five experimental embryos developed ectopic SMC1-positive cells with an average of two cells per embryo. One embryo with ectopic SMC1 staining also had an ectopic invagination (Fig. 9). Thus, the micromeres are sufficient to induce the expression of the SMC1 antigen, but the induction is not as robust as in the normal induction of the vegetal plate.

The inductive abilities of the micromeres were tested further by transplanting a quartet of micromeres to the vegetal pole of an animal cap. Animal caps gave rise to ciliated epithelial embryoids (Fig. 10A), which did not express endodermal or mesodermal fates at times when control and experimental embryos were examined. Following micromere transplantation to the animal cap, an archenteron was induced, but its development was delayed by as much as 10 hours compared to controls. The induced archenteron was often very thin, but SMCs formed at the tip (Fig. 10B,C). Few experimental embryos produced SMC1-positive cells (7/24; Fig. 10D) and, among these, there was an average of only three cells per embryo. 4 out of 6 resulting larvae developed blastocoelar cells, but the number appeared to be lower compared to controls (Fig. 10E compared to Fig. 3I). 7 out of 14 resulting larvae examined for pigment cells developed this cell type, though there was an average of only nine pigment cells per larva (Fig. 10F). 4 out of 6 resulting embryos developed muscle cells with an average of 15 cells per embryo (Fig. 10G). Thus, the micromeres can induce the development of all mesodermal cell types examined; however, fewer mesodermal cells developed in these experimental embryos than in control embryos, suggesting that the animal cells may be less responsive to micromere signaling than vegetal cells.

To test this possibility, the micromeres were removed from 16-cell stage embryos and replaced with four RITC-labeled micromeres. These embryos produced an average of 18 SMC1-positive cells per embryo (n=10; s.d.=16), suggesting that the procedure can partially account for the absence of SMC1 staining following micromere transplantation. These embryos, however, went on to develop approximately normal numbers of pigment cells (average=81; n=10; s.d.=19; Fig. 10H,I) supporting the idea that the mesomeres and their derivatives are less responsive to the micromere-derived signal than macromeres.

There is much evidence that the Notch signaling pathway is involved in normal non-skeletogenic mesoderm specification and that activation of the pathway is correlated with the loss of LvNotch protein in prospective SMCs (Sherwood and McClay, 1997, 1999). At the late mesenchyme blastula stage (12 hours), LvNotch protein is expressed throughout most of the embryo at low levels and is strongly expressed at apical surfaces of prospective endodermal cells; however, it is absent from prospective SMCs in the vegetal plate (Fig. 11A; Sherwood and McClay, 1997). To determine whether micromere signaling is involved in the downregulation of LvNotch, the micromeres were removed and the resulting embryos were stained with anti-LvNotch antibody. At 12 hours after fertilization, LvNotch was expressed apically by all cells in the vegetal plate (Fig. 11C; 6/7; LvNotch staining was very faint throughout one embryo). Thus, micromere signaling is not necessary for the apical expression of LvNotch, but is necessary for the downregulation of the LvNotch protein, and presumably for the activation of the signaling pathway and the specification of SMCs in the vegetal plate.

During normal gastrulation, the LvNotch protein continues to be expressed at the apical surface of endodermal cells and absent from prospective SMCs at the archenteron tip (Fig. 11B; Sherwood and McClay, 1997). During early gastrulation in micromere(−) embryos (16 to 20 hours after fertilization), LvNotch was not lost from the archenteron tip, suggesting that it is not activated by this time (Fig. 11D; 15/18). In three embryos, some cells contained only punctate intracellular staining, suggesting that the apical LvNotch protein was internalized. When mesenchymal cells began to form by 24 to 28 hours, the pattern of LvNotch staining was difficult to interpret in most embryos because the abundance of apical LvNotch protein had decreased throughout the archenteron (13/19). In two embryos, LvNotch remained apically localized within the archenteron and, in four embryos, several cells had punctate staining suggesting that the protein was internalized (Fig. 11E). The intracellular expression pattern is consistent with the possibility that the Notch signaling pathway is activated during mesoderm specification at this later stage, independent of micromere signaling.

Our understanding of the specification of non-skeletogenic mesoderm has been fettered by a lack of early molecular markers for these cell types. The generation of the SMC1 and SMC2 monoclonal antibodies has made it possible to examine the mechanisms involved in the development of the SMCs.

The intracellular, punctate pattern of SMC1 staining suggests that the antigen is initially localized to a compartment of the secretory pathway. When these cells ingress the SMC1 antigen appears to be released into the blastocoel. We favor the view that the cells that express the SMC1 antigen are most, but not all, of the macromere-derived presumptive non-skeletogenic mesoderm. Triple-labeling of embryos with SMC1, anti-LvNotch and CAD-1 revealed a sharp boundary between SMC1- and LvNotch-expressing territories, supporting the view that SMC1-positive cells are restricted to prospective mesoderm. The number of SMC1-positive cells (about 46 per embryo), however, is lower than the estimated number of SMC precursors (about 66 per embryo) at the late mesenchyme blastula stage (Ruffins and Ettensohn, 1996). This finding, along with the observation that there are some SMC1-negative cells within the domain of Notch downregulation, leads us to conclude that a minority of prospective SMCs are not recognized by this antibody.

The time of onset of SMC1 staining suggests that SMC specification events begin as early as the hatched blastula stage. This is consistent with the finding that Notch activation and downregulation begins at this time (Sherwood and McClay, 1997). SMC1-positive cells become symmetrically arranged around the vegetal pole, but are lost from the aboral side of the vegetal plate at the early gastrula stage. This is consistent with the findings that pigment cell ingression begins at the early gastrula stage in this species (Ettensohn and McClay, 1988) and that most pigment cells originate from the aboral side of the vegetal plate (Ruffins and Ettensohn, 1996). The remaining SMC1-positive cells in the oral half of the vegetal plate are likely prospective blastocoelar cells as this cell type is concentrated in this region at this time (Ruffins and Ettensohn, 1996). It is unclear whether prospective muscle and coelomic pouch cells express the SMC1 antigen.

The SMC2 antibody initially stains cells within the vegetal plate at the blastula stage and PMCs at the mesenchyme blastula stage. SMC2 also stains pigment cells and blastocoelar cells as they form throughout gastrulation. Thus, this surface antigen appears to be transiently expressed by all populations of migratory mesenchyme cells. At later stages, however, SMC2 is a specific marker for blastocoelar cells.

Previous studies have demonstrated that micromere signaling is necessary and sufficient for the specification of endoderm (Hörstadius, 1973; Ransick and Davidson, 1995, 1993), and sufficient for the development of pigment cells (Khaner and Wilt, 1991; Amemiya, 1996). In the present study, we have examined how micromere signaling is involved in the specification of several SMC types, including SMC1-positive cells, blastocoelar cells, pigment cells and muscle cells. Following micromere removal, there were several abnormalities in mesoderm development. In all batches, mesodermal cells formed after a chronological delay and only after the onset of archenteron invagination. Few SMC1-positive cells, pigment cells and blastocoelar cells developed, suggesting that the micromeres normally induce these mesodermal cell types. Another possibility, but perhaps more unlikely, is that these cells form but subsequently die. Micromere induction is not absolutely necessary for mesoderm development because some SMCs eventually formed. The regulative development of these cells, however, may occur by a somewhat different pathway compared to normal embryos, as these embryos rarely stained with SMC1. This may be the case for blastocoelar cells, but because few pigment cells developed in micromere(−) embryos, it is not clear whether or not they express the SMC1 antigen. The lack of micromere signaling had little effect on the ultimate development of muscle cells.

Though not normally a derivative of SMCs, skeletogenic cells also developed in micromere(−) embryos and the total number was similar to that found in control embryos. Several other studies have demonstrated previously that skeletogenic mesenchyme forms in the absence of micromeres (Hörstadius, 1973; Langelan and Whiteley, 1985; Ransick and Davidson, 1993) or PMCs (Fukushi, 1962; Ettensohn and McClay, 1988). Without the PMCs, some SMCs take on the skeletogenic fate at the apparent expense of pigment and blastocoelar cell fates (Ettensohn and Ruffins, 1993). These studies suggest that the skeletogenic fate is the default fate for some SMCs and that the PMCs normally suppress the SMCs from becoming skeletogenic and promote them to develop as other mesodermal cell types. In the micromere(−) embryos of the current study, the finding that the skeletogenic cells are the first mesodermal cell type to form is consistent with this hypothesis. In addition, the finding that few pigment and blastocoelar cells develop is consistent with the lack of the PMC-derived signal. We show, however, that even though added PMCs suppress the development of skeletogenic cells, they do not promote the development of pigment and blastocoelar cells. Thus, under these conditions, skeletogenic cells do not develop at the expense of pigment or blastocoelar cells. This strongly supports the idea that, during normal development, the specification of pigment and blastocoelar cells requires a separate signal from the micromeres.

By transplanting micromeres to animal cells, we have demonstrated that micromere signaling is sufficient to induce SMC1-positive cells, blastocoelar cells, pigment cells and muscle cells. Fewer mesodermal cells, however, developed in experimental embryos compared to normal embryos and embryos in which micromeres were transplanted back to their normal position. This suggests that mesomere derivatives are less able to respond to micromere signaling than are macromere derivatives. The limited ability of animal cells to develop mesoderm may be due to the signals that occur between mesomere derivatives that are involved in suppressing the development of vegetal cell types (Henry et al., 1989). In addition, it is possible that the limited response of the animal cells is due to their low level of nuclear β-catenin (see below; Logan et al., 1999).

The specification of non-skeletogenic mesoderm involves the activation of the Notch signaling pathway and the redistribution of the LvNotch protein from the cell membrane to intracellular vesicles (Sherwood and McClay, 1997, 1999). In the present study, we have shown that micromere signaling is required for this downregulation of LvNotch protein and presumably for the activation of the Notch signaling pathway. It is not clear whether the micromeres provide a ligand for LvNotch, or whether LvNotch is activated several steps downstream of micromere signaling. Micromere signaling is also involved in the specification of endoderm (Hörstadius, 1973; Ransick and Davidson, 1993, 1995); however, endoderm development does not appear to involve Notch signaling (Sherwood and McClay, 1997, 1999). The following model attempts to link micromere signaling and LvNotch activation in the specification of non-skeletogenic mesoderm. During normal development (Fig. 12A), micromere signaling directly or indirectly causes the activation and downregulation of LvNotch protein. Events downstream of activated LvNotch result in the expression of SMC-specific genes. In the absence of micromere signaling (Fig. 12B), LvNotch is not activated or downregulated, and SMC specification does not occur at the normal time. During gastrulation in micromere(−) embryos, mesodermal cells do form and LvNotch is internalized, suggesting that LvNotch signaling may also be involved in mesoderm development in the absence of the micromeres. It remains possible that mesoderm specification under these conditions occurs by an independent pathway as Notch downregulation is not always associated with Notch activation (Sherwood and McClay, 1997, 1999). LvNotch downregulation, however, has also been shown to be associated with mesoderm specification following archenteron removal at the early gastrula stage (Sherwood and McClay, 1997), providing another example of LvNotch downregulation occurring in a micromere-independent fashion.

Components of the Wnt signaling pathway are involved in the specification of both endoderm and mesoderm (reviewed by Ettensohn and Sweet, 1999). During cleavage, β-catenin is localized to the nuclei of micromeres, macromeres and their derivatives (Logan et al., 1999). Upregulation of nuclear β-catenin causes excess endoderm and mesoderm to develop at the expense of ectoderm, whereas reduction of endogenous β-catenin function results in deficiencies in endoderm and mesoderm (Wikramanayake et al., 1998; Emily-Fenouil et al., 1998; Logan et al., 1999). β-catenin may be necessary for the specification of these cell types by endowing blastomeres with the potential to respond to the micromere signal and/or by giving blastomeres the potential to develop endoderm and mesoderm autonomously. Micromere signaling has no effect on the localization of β-catenin in neighboring cells, suggesting that nuclear localization is controlled autonomously (Logan et al., 1999). The embryos of most echinoderms, including some sea urchins (Schroeder, 1981; Wray and McClay, 1988), do not form micromeres during normal development, and thus endoderm and mesoderm specification must occur by mechanisms other than micromere signaling. If these mechanisms also exist in sea urchin embryos that form micromeres, they may in part explain endoderm and mesoderm specification following micromere removal. It will be important to examine the involvement of β-catenin and Notch in the specification of mesoderm in the embryos of these other echinoderm groups, as well as in the specification of the mesoderm that occurs in the absence of micromere signaling in L. variegatus.

The authors are grateful to G. Wessel, J. Miller, D. Sherwood and D. McClay for their generous gift of antibodies against myosin, LvG-cadherin and LvNotch. This research was supported by NSF Grant IBN-9817988, NIH Grant HD-24690 and an NIH Research Career Development Award to C. A. E.