ABSTRACT
The specifications of cell types and germ-layers that arise from the vegetal plate of the sea urchin embryo are thought to be regulated by cell-cell interactions, the molecular basis of which are unknown. The Notch intercellular signaling pathway mediates the specification of numerous cell fates in both invertebrate and vertebrate development. To gain insights into mechanisms underlying the diversification of vegetal plate cell types, we have identified and made antibodies to a sea urchin homolog of Notch (LvNotch). We show that in the early blastula embryo, LvNotch is absent from the vegetal pole and concentrated in basolateral membranes of cells in the animal half of the embryo. However, in the mesenchyme blastula embryo LvNotch shifts strikingly in subcellular localization into a ring of cells which surround the central vegetal plate. This ring of LvNotch delineates a boundary between the presumptive secondary mesoderm and presumptive endoderm, and has an asymmetric bias towards the dorsal side of the vegetal plate. Experimental perturbations and quantitative analysis of LvNotch expression demonstrate that the mesenchyme blastula vegetal plate contains both animal/vegetal and dorsoventral molecular organization even before this territory invaginates to form the archenteron. Furthermore, these experiments suggest roles for the Notch pathway in secondary mesoderm and endoderm lineage segregation, and in the establishment of dorsoventral polarity in the endoderm. Finally, the specific and differential subcellular expression of LvNotch in apical and basolateral membrane domains provides compelling evidence that changes in membrane domain localization of LvNotch are an important aspect of Notch receptor function.
INTRODUCTION
The vegetal plate of the sea urchin mesenchyme blastula embryo is a structurally simple, yet developmentally dynamic, region. This concave single cell-layer epithelium contains all three future germ-layers of the embryo concentrically arranged within a few cell diameters (Ruffins and Ettensohn, 1996). During gastrulation, vegetal plate cells undergo dramatic movements and shape changes to invaginate and form the archenteron (reviewed in Hardin, 1996). Cells at tip of the archenteron segregate and become secondary mesenchyme cells (SMCs). Behind the SMCs, the archenteron gives rise to the endoderm, which subdivides into several regions. Finally, the ectoderm is excluded from the archenteron. These changes that occur in cells derived from the vegetal plate during gastrulation offer a relatively simple cellular model for studying the integration of mechanisms underlying cell fate diversification and morphogenesis.
Cellular studies of vegetal plate regionalization suggest that local cell-cell interactions play an essential role in the diversification of cell-types that arise from the vegetal plate (McClay and Logan, 1996; reviewed in Davidson, 1993). The molecular basis of these cell-cell interactions, however, is unknown. One segregation that must occur in cells derived from the vegetal plate is the differential specification of SMCs and endoderm cells. Lineage analysis has shown that the presumptive SMCs lie in the central region of the mesenchyme blastula vegetal plate and are clonally distinct from the presumptive endoderm cells, which are positioned in a ring around the SMCs (Ruffins and Ettensohn, 1996). The clonal isolation of presumptive SMCs from endoderm cells is consistent with these germ-layers being differentially specified before vegetal plate invagination. To date, however, molecular evidence has suggested otherwise. The vegetal plate marker Endo16 and sea urchin homologs of brachyury and forkhead are expressed throughout the vegetal plate prior to invagination; only after invagination has begun do these and other identified molecular markers become differentially expressed between SMCs and endoderm lineages (Ransick et al., 1993; Harada et al., 1995, 1996; reviewed in Davidson, 1993). Therefore, it remains unclear when these germ-layers are specified. Given the evidence for cell-cell interactions specifying cell-types in the vegetal plate, an analysis of evolutionarily conserved intercellular signaling pathways may shed light into the time and mechanism by which these two germ-layers become distinct.
The Notch signaling pathway mediates cell-cell interactions leading to the specification and patterning of a wide array of cell types in both invertebrate and vertebrate development (Conlon et al., 1995; de Celis et al., 1996; reviewed in Artavanis-Tsakonas et al., 1995). The conserved role and pleiotropic function of this pathway made it a likely candidate for involvement in the diversification of cell-types arising from the vegetal plate. The Drosophila Notch gene and two related genes, lin-12 and glp-1 in C. elegans, have been identified as receptors in the Notch pathway. These genes, and four identified vertebrate homologs of Notch, all encode single-spanning transmembrane proteins (Uyttendaele et al., 1996; reviewed in Artavanis-Tsakonas et al., 1995). Studies on the localization of both Drosophila Notch and C. elegans GLP-1 have led to the proposal that the activity of this signaling pathway may in part be controlled by the precise distribution of the receptor (Fehon et al., 1991; Crittenden et al., 1994). Therefore, knowledge of the localization of Notch proteins is a likely indicator of where this pathway functions.
We have isolated and generated antibodies to the first echinoderm Notch homolog (LvNotch). Analysis of LvNotch expression demonstrates that the presumptive SMCs and endoderm are indeed differentially specified in the mesenchyme blastula embryo, earlier than any previous markers have shown. In addition, there is a dorsal bias in the expression of LvNotch in the mesenchyme blastula vegetal plate that is maintained in the presumptive endoderm during invagination. These results, together with several experimental manipulations and a quantitative analysis of LvNotch expression, suggest that the Notch signaling pathway is involved in the segregation of SMCs from endoderm cells, and in the establishment of a dorsoventral axis in the endoderm. Finally, we present compelling evidence that shifts in membrane domain localization are an important component of Notch receptor function.
MATERIALS AND METHODS
Animals
Sea urchins (Lytechinus variegatus) were obtained from Susan Decker (Hollywood, FL) and Tracy Andacht (Duke University Marine Laboratory). Gametes were harvested and cultured as described by Hardin et al. (1992).
Cloning of LvNotch
Poly(A)+ RNA was prepared and cDNA made from various stages of embryos as described (Bachman and McClay, 1995). Degenerate oligonucleotide primers were designed as in Stifani et al. (1992), and used in a PCR reaction with the above cDNA pools using the following conditions: 95°C, 60 seconds; 40°C, 60 seconds; 72°C, 2 minutes for 45 cycles. An amplified 429 bp band was cloned into Bluescript SK-vector (Stratagene) and sequenced (Sequenase 2.0 kit; Amersham Life Science). Initial clones of LvNotch obtained from a lambda-zap phage mid-gastrula cDNA library (Stratagene), using the cloned PCR product as a probe, led to the identification of the 3′ end of the coding sequence. However, multiple screens and the isolation of 28 clones were necessary to locate the 5′ end of the coding region. Three clones, which together covered the entire open reading frame, were subcloned and subjected to a transposon-based sequencing strategy (Strathmann et al., 1991) to obtain the full sequence for both strands of DNA encoding LvNotch. DNA sequences were analyzed using AssemblyLign (International Biotechnologies, Inc.). Alignment of the 5′ and 3′ ends of all clones identified to the final sequence of LvNotch gave no evidence for differential processing of the LvNotch gene.
Northern analysis
A 1.0% agarose/formaldehyde gel was loaded with 3 μg/lane poly(A)+ RNA (isolated with QuickPrep; Pharmacia), fractioned by electrophoresis, transferred onto a nylon membrane (Gene Screen, NEN Research Products), and hybridized with a 948 bp fragment of LvNotch corresponding to a portion of the extracellular domain (amino acids 798-1114). The blot was then probed and washed as described (Bachman and McClay, 1995). Identical results to those shown in Fig. 2 were also obtained with a probe corresponding to a portion of the intracellular domain (data not shown).
Antibody production
Eight fusion proteins were expressed using subcloned or PCR-amplified fragments of LvNotch in pGex1, 2 and 3 vectors (Glutathione S-transferase (GST) expression system; Smith and Johnson, 1988). The fusion proteins and corresponding LvNotch amino acids they encompass are as follows: Pet1, 29-287; Pet2, 198-464; 2s, 488-799; Bam3, 715-878; Bam4, 1002-1130; Bam1, 1128-1472; Ank, 1751-2095; CloneB, 1886-2531. All fusion proteins were expressed and purified as described (Smith and Johnson, 1988), with the exception of 2s and CloneB. The fusion protein 2s was insoluble and purified as outlined for Dmoesin (McCartney and Fehon, 1996). Isolated fusion proteins were injected into animals to generate polyclonal antibodies (pAb; Harlow and Lane, 1988); clone B, which could not be eluted from the glutathione-agarose beads, was injected coupled to these beads. Animals injected with fusion proteins were as follows: mice, all fusion proteins; guinea pigs, Bam1 and Ank; Rabbits, Ank. Whole serum from mice was prepared by ammonium sulfate precipitation (Harlow and Lane, 1988). Serum from guinea pigs and rabbits was affinity purified as described (McCartney and Fehon, 1996). Specificity of antibodies to LvNotch was determined by staining gastrula-stage embryos and comparing immunofluores-cence patterns. Controls for specificity included staining with preimmune serum, staining of embryos with antibodies to GST, and staining after pre-incubation of pAb with the appropriate fusion protein.
Western analysis
Protein extracts were prepared by pelleting embryos, adding 10 μl of pelleted embryos to ice-cold SDS Lysis buffer (200 μl; 100 mM Tris, pH 6.8; 4% SDS; 20% glycerol; 1 mM PMSF; 1 μg/mL Leupeptin), douncing 4× with a pestle, and then immediately boiling for 5 minutes. Samples were then spun at 16,000 relative centrifugal force (RCF) for 15 minutes, and the supernatant frozen at −70°C until use. The number of embryos homogenized in the 10 μl sample was determined by serial dilutions taken from the original pellet of embryos. Samples were run on 3%-15% gradient gels, transferred to nitrocellulose and probed with affinity-purified guinea pig α-Bam1 (1:1000; 100 μg/ml stock) or affinity-purified rabbit α-Ank (1:2500; 1 mg/ml stock) pAbs. The blots were processed and developed as described (McCartney and Fehon, 1996). Pre-immune sera used as controls for these antibodies gave light, nonspecific background banding patterns.
Immunolocalization and image analysis
Embryos were fixed in ice-cold methanol for 20 minutes and rehydrated and washed 2× with ice-cold artificial sea water (ASW). Embryos were then washed with PBS, blocked for 10 minutes in PBS/2% normal goat serum (GibcoBRL), incubated overnight at 4°C in primary antibody, washed, blocked as above, then incubated at room temperature in secondary antibody (Cy5- or Cy3-conjugated; Jackson Immunoresearch Laboratories) for 1.5 hours. Embryos were washed and mounted in 9:1 (v/v) glycerol:1 M Tris, pH 8.0. For all images shown, pAbs generated to the fusion protein Bam1 were used; either from mice whole sera at 1:1000 or affinity-purifiedfrom guinea pig at 1:100 (25 μg/ml stock). However, pAbs generated from a minimum of two non-overlapping fusion proteins were analyzed for all stages shown to confirm the specificity of the staining pattern. The PMC-specific monoclonal antibody (mAb) 1G8 (McClay et al., 1983), and guinea pig α-LvCadherin and α-Lvβ-catenin pAbs (Miller, 1995), were used with the above fixation and incubation conditions. A Zeiss confocal laser scanning microscope was used to aquire all images. Mesenchyme blastula-stage embryos were fixed for cell-count analysis of LvNotch localization in vegetal plate cells 30-50 minutes after the start of PMC migration (10-45 minutes before vegetal plate invagination was observed). Lvβ-catenin/LvNotch and 1G8/LvNotchstained vegetal plates were obtained by sequential confocal sectioning of double-labeled vegetal plates, rendering 2-D projections of the sections, and then overlaying these projections using Adobe Photoshop 3.0.5.
LiCl and NiCl2 treatment and embryo manipulation
Batches of fertilized eggs were divided into control and treated cultures. Embryos were treated with 30 mM LiCl from the 2-4 cell stage until the mid-mesenchyme blastula stage; embryos treated with NiCl2 were placed in 1 mM NiCl2 immediately after fertilization until the early gastrula stage (Hardin et al., 1992). After treatment, embryos were concentrated, rinsed 3× with ASW and cultured in ASW. Treated and untreated embryos were immunostained in parallel. Single embryo manipulations and antibody staining was performed as described (McClay and Logan, 1996).
Phylogenetic analysis
Full-length Notch family member protein products were obtained in a BLAST search (Version 8, 1994, Genetics Computer Group (GCG), Madison, WI) and sequence alignment was performed using the PileUp program of GCG (gap weight: 3.0; length weight: 0.100). Ambiguously aligned positions were excluded from phylogenetic analysis, which produced the same tree using both weighted parsimony (MACCLADE 3.04, Maddison and Maddison, 1992) and unordered parsimony (PAUP 3.1; Swofford, 1993). Bootstrap values were obtained for the unordered parsimony tree using PAUP 3.1 (Swofford, 1993).
RESULTS
Isolation of sea urchin LvNotch cDNA clones
Using degenerate oligonucleotide primers (Stifani et al., 1992), a single band at the predicted size of 400 bp was amplifed by RT-PCR using staged pools of cDNA from embryos of the sea urchin Lytechinus variegatus. This amplification produced a single PCR fragment homologous to previously isolated Notch genes, and was used to screen a mid-gastrula stage cDNA library. This led to the identification and the sequencing of three cDNA clones, which together make up a 7,646 bp cDNA sequence with a predicted open reading frame of 2,531 amino acids (Fig. 1A). Based on homology to other Notch genes we have named this gene LvNotch.
(A) An alignment of the deduced amino acid sequence of LvNotch (GenBank # AF00634) and Drosophila Notch (for example, see Wharton et al., 1985) proteins using the BestFit program (GCG; gap weight, 3.0, length weight, 0.001). Identical amino acids are shaded. Positions of the first amino acid residues in EGF-repeats, Notch/Lin-12 repeats and Ankyrin repeats of Drosophila Notch are indicated by number (Coffman et al., 1990). Conserved structural domains are outlined. (B) Parsimony analysis of LvNotch and the four vertebrate Notch family members indicates that the vertebrate Notch proteins are more closely related to each other than to LvNotch. Drosophila Notch was used to root the tree. Bootstrap values from 550 pseudoreplicates are indicated above the nodes (PAUP 3.1; Swofford, 1993). Scale bar indicates the number of amino acid substitutions along a given branch length (calculated by MACCLADE; Maddison and Maddison, 1992).
(A) An alignment of the deduced amino acid sequence of LvNotch (GenBank # AF00634) and Drosophila Notch (for example, see Wharton et al., 1985) proteins using the BestFit program (GCG; gap weight, 3.0, length weight, 0.001). Identical amino acids are shaded. Positions of the first amino acid residues in EGF-repeats, Notch/Lin-12 repeats and Ankyrin repeats of Drosophila Notch are indicated by number (Coffman et al., 1990). Conserved structural domains are outlined. (B) Parsimony analysis of LvNotch and the four vertebrate Notch family members indicates that the vertebrate Notch proteins are more closely related to each other than to LvNotch. Drosophila Notch was used to root the tree. Bootstrap values from 550 pseudoreplicates are indicated above the nodes (PAUP 3.1; Swofford, 1993). Scale bar indicates the number of amino acid substitutions along a given branch length (calculated by MACCLADE; Maddison and Maddison, 1992).
Developmental northern blot of LvNotch expression. The amount of poly(A)+ RNA was 3 μg/lane, calculated by OD260 (left) or estimated by isolation from the same number of embryos (right). E, egg; 7th, 7th cleavage; TVB, thickened vegetal plate blastula; EG, early gastrula; LG, late gastrula; Pr, prism; Pl, pluteus larva.
Developmental northern blot of LvNotch expression. The amount of poly(A)+ RNA was 3 μg/lane, calculated by OD260 (left) or estimated by isolation from the same number of embryos (right). E, egg; 7th, 7th cleavage; TVB, thickened vegetal plate blastula; EG, early gastrula; LG, late gastrula; Pr, prism; Pl, pluteus larva.
Deduced amino acid sequence and phylogenetic analysis of LvNotch
Alignment of the predicted LvNotch protein sequence with Drosophila Notch demonstrated that LvNotch contains all conserved domains shared by Notch proteins (Fig. 1A). LvNotch has multiple EGF-like repeats, three Notch/Lin-12 repeats, a transmembrane domain, six Ankyrin repeats and a putative PEST domain. However, LvNotch contains only 35 EGF-like repeats, in contrast to the 36 repeats found in Drosophila Notch and vertebrate Notch1 and 2. An alignment of Notch family members, including at least one of the four vertebrate Notch proteins and Drosophila Notch (GCG PileUp program, gap weights: 3, 10 and 15), revealed a deletion of EGF-repeat 14 in LvNotch (alignment not shown; Fig. 1A). Overall amino acid sequence comparisons between vertebrate Notch1, 2, 3, 4 and Drosophila Notch to LvNotch showed identities of 43%, 41%, 41%, 38% and 44%, respectively.
To address whether a duplication event or events in the Notch gene may have occurred before the divergence of sea urchins and vertebrates from a common ancestor, a parsimony analysis of Notch protein sequences was undertaken (Swofford, 1993; MACCLADE 3.04, Maddison and Maddison, 1992). The resulting phylogenetic tree (Fig. 1B) suggests that the duplication events that produced multiple Notch genes in vertebrates occurred in the vertebrate lineage after the divergence of vertebrates and sea urchins from a common ancestor.
LvNotch expression and localization
LvNotch mRNA was present during all stages of sea urchin embryogenesis (Fig. 2). Low levels of maternal mRNA were found in the egg. After fertilization the amount of LvNotch mRNA appeared to increase, peaking in abundance at the gastrula stage, and then decreasing to lower levels in the pluteus larva.
To determine the temporal and spatial pattern of LvNotch protein distribution, eight fusion proteins were made, encompassing different regions of LvNotch (Fig. 3A). Specific antibodies were generated to five LvNotch fusion proteins (Bam1, 3, 4, Pet1, Ank), and were used to examine the expression of LvNotch (see Materials and Methods).
(A) Schematic diagram of LvNotch. Regions to which eight fusion proteins were made are indicated. * marks fusion proteins that produced specific pAbs. (B,C) Western analysis of protein extracts using intracellular-directed rabbit α-Ank and extracellular-directed guinea pig α-Bam1 pAbs. (B) Long exposure of gastrula protein extracts (200 embryos/lane); arrow indicates what is presumably full length LvNotch at 350×103Mr, and arrowheads the predominant intracellular and extracellular fragments of LvNotch at 116×103Mr and 260×103Mr, respectively. (C) Developmental western analysis of protein extracts (200 embryos/lane) shows that fragmentation of LvNotch occurs at all stages where LvNotch is expressed (arrowheads). Affinity-purified antibodies from two separate animals for both fusion proteins yielded identical banding patterns, while pre-immune serum gave nonspecific, weaker banding. * denotes yolk protein, which cross-reacts with α-Bam1 pAb. 16, 16-cell stage; other abbreviations as in Fig. 2.
(A) Schematic diagram of LvNotch. Regions to which eight fusion proteins were made are indicated. * marks fusion proteins that produced specific pAbs. (B,C) Western analysis of protein extracts using intracellular-directed rabbit α-Ank and extracellular-directed guinea pig α-Bam1 pAbs. (B) Long exposure of gastrula protein extracts (200 embryos/lane); arrow indicates what is presumably full length LvNotch at 350×103Mr, and arrowheads the predominant intracellular and extracellular fragments of LvNotch at 116×103Mr and 260×103Mr, respectively. (C) Developmental western analysis of protein extracts (200 embryos/lane) shows that fragmentation of LvNotch occurs at all stages where LvNotch is expressed (arrowheads). Affinity-purified antibodies from two separate animals for both fusion proteins yielded identical banding patterns, while pre-immune serum gave nonspecific, weaker banding. * denotes yolk protein, which cross-reacts with α-Bam1 pAb. 16, 16-cell stage; other abbreviations as in Fig. 2.
Examination of western blots of gastrula protein extracts using intracellular (α-Ank), and extracellular (α-Bam1) antibodies revealed prominent immunoreactive bands at 116×103Mr and 260×103Mr, respectively, as well as additional lower abundance fragments (Fig. 3B). This banding pattern suggested that the majority of endogenous LvNotch product is present as two fragments, corresponding roughly in size to the intracellular and extracellular domains. Rapid lysis with the addition of an extensive protease cocktail during protein extraction did not reduce the amount of fragmented LvNotch product (data not shown). Similar complex fragmentation of the Notch receptor has been reported in Drosophila, for C. elegans GLP-1, and in cell lines expressing vertebrate Notch homologs (Fehon et al., 1990; Crittenden et al., 1994; Zagouras et al., 1995). The functional significance of this processing or degradation of Notch is unknown. A developmental western blot showed that LvNotch was fragmented at all stages where it was expressed, and that abundance peaked in the gastrula embryo, consistent with the mRNA expression profile (Fig. 3C). While western analysis suggested that LvNotch may exist predominantly as an extracellular and intracellular fragment, whole-mount immunofluorescence using these extracellular- and intracellular-specific antibodies revealed no differences in localization pattern (data not shown).
Whole-mount immunofluorescent analysis showed that LvNotch protein was dynamically expressed at the cellular and subcellular level after cleavage stages. During cleavage stages, expression was uniform on the surfaces of all blastomeres (Fig. 4A). In the early blastula embryo (6 hours in development), loss of LvNotch expression was seen in a sector of the embryo that later becomes the center of the vegetal plate (Fig. 4B). An increase in basolateral membrane staining was observed at this time in cells that continued to express LvNotch. This pattern of expression persisted through the thickened vegetal plate stage (10 hours), when the area lacking detectable LvNotch was identified as the central thickened vegetal plate (Fig. 4C). Only 1 to 2 hours later (mid to late-mesenchyme blastula stage), LvNotch was strongly upregulated on the apical surface of cells surrounding the central region of the vegetal plate (referred to as apical LvNotch; Fig. 4G). Low levels were also observed inconsistently in the cytoplasm of the eight small micromere descendants at the center of the vegetal plate (data not shown). At this same time, LvNotch, which previously was localized predominantly basolaterally in cells of the animal half of the embryo, was now distributed uniformly in cell membranes. Examination of cell surfaces in the early mesenchyme blastula vegetal plate revealed a slight upregulation of LvNotch along the apical surface of cells bordering the central vegetal plate (Fig. 4D). Midway through the mesenchyme blastula stage (approx. 11.5 hours) a subset of cells began to express high levels of apical LvNotch in a striking cell-by-cell manner (Fig. 4D,E). By the end of this stage (12-12.5 hours), cells with high levels of apical LvNotch formed an asymmetric ring, with one side of the ring having both increased levels of apical expression and more cells expressing apical LvNotch (Fig. 4F). In the mid-gastrula (14 hours), high levels of apical LvNotch expression were maintained in cells now identifiable as presumptive endoderm cells, while the SMCs at the tip of the archenteron lacked detectable LvNotch expression (Fig. 4H). By the late gastrula (16 hours), when most of the SMCs have migrated away from the tip of the archenteron, apical LvNotch extended to the tip of this structure (Fig. 4I). Notably, throughout most of gastrulation the highest levels of apical LvNotch were distributed along one side of the archenteron (Fig. 4H, inset; Fig. 4I), except in the last presumptive endoderm cells to invaginate, where LvNotch was expressed at high levels in a symmetric manner (data not shown).
LvNotch displays dynamic cellular and subcellular distribution during sea urchin development. (A) LvNotch is expressed uniformly along all cell surfaces in the 60-cell-stage embryo. (B) In the early blastula, LvNotch is downregulated in a sector of the embryo (between arrowheads) and becomes concentrated along basolateral membranes of cells. (C) In the thickened vegetal plate stage, LvNotch maintains a concentration along basolateral membranes of cells, and the region of LvNotch downregulation is identifiable as the thickened vegetal plate (vp). (D-F) Surface views of mesenchyme blastula-stage embryos. (D) The vegetal plate of a mid-mesenchyme blastula embryo shows a narrow ring of cells expressing slightly increased levels of apically localized LvNotch and a single cell expressing higher levels. (E) In a slightly older embryo (side view), more cells express apical LvNotch. (F) In the late mesenchyme blastula embryo, vegetal plate cells express apical LvNotch in an asymmetric ring. (G) A cross-section of a late mesenchyme blastula embryo shows that apical LvNotch (arrow) extends into cells in the center of the vegetal plate, beyond the limits of cells in which LvNotch is expressed basolaterally (arrowhead). In contrast to the vegetal plate, the cells of the animal half of the embyro express LvNotch at lower levels with a nonpolar distribution in membranes. (H) In the mid-gastrula stage, high levels of apical LvNotch expression are found in the presumptive endoderm while expression is absent in SMCs at the tip of the archenteron (arrow). A cross-section of the archenteron at the level of the presumptive endoderm (inset) shows that the asymmetic distribution of apical LvNotch continues into the archenteron. (I) LvNotch expression extends to the end of the archenteron in the late gastrula, and continues to be asymmetrically distributed along the sides. Bars: 25 μm figures; 10 μm inset.
LvNotch displays dynamic cellular and subcellular distribution during sea urchin development. (A) LvNotch is expressed uniformly along all cell surfaces in the 60-cell-stage embryo. (B) In the early blastula, LvNotch is downregulated in a sector of the embryo (between arrowheads) and becomes concentrated along basolateral membranes of cells. (C) In the thickened vegetal plate stage, LvNotch maintains a concentration along basolateral membranes of cells, and the region of LvNotch downregulation is identifiable as the thickened vegetal plate (vp). (D-F) Surface views of mesenchyme blastula-stage embryos. (D) The vegetal plate of a mid-mesenchyme blastula embryo shows a narrow ring of cells expressing slightly increased levels of apically localized LvNotch and a single cell expressing higher levels. (E) In a slightly older embryo (side view), more cells express apical LvNotch. (F) In the late mesenchyme blastula embryo, vegetal plate cells express apical LvNotch in an asymmetric ring. (G) A cross-section of a late mesenchyme blastula embryo shows that apical LvNotch (arrow) extends into cells in the center of the vegetal plate, beyond the limits of cells in which LvNotch is expressed basolaterally (arrowhead). In contrast to the vegetal plate, the cells of the animal half of the embyro express LvNotch at lower levels with a nonpolar distribution in membranes. (H) In the mid-gastrula stage, high levels of apical LvNotch expression are found in the presumptive endoderm while expression is absent in SMCs at the tip of the archenteron (arrow). A cross-section of the archenteron at the level of the presumptive endoderm (inset) shows that the asymmetic distribution of apical LvNotch continues into the archenteron. (I) LvNotch expression extends to the end of the archenteron in the late gastrula, and continues to be asymmetrically distributed along the sides. Bars: 25 μm figures; 10 μm inset.
Relationship of apical LvNotch expression to the zonula adherens junction and to developmental compartments
The striking appearance and specific localization of apical LvNotch in vegetal plate cells of the mesenchyme blastula embryo suggested that this specific distribution may have a functional significance. As a first step towards understanding possible functions for this localization pattern, we undertook a detailed analysis of the expression of apical LvNotch in relation to cellular and developmental compartments.
The appearance of apical LvNotch does not correlate with the formation of the zonula adherens junction
In Drosophila, Notch also undergoes changes in subcellular distribution during development, from a lateral membrane distribution in early embryonic epidermal cells to a close association with the zonula adherens (ZA) junctions in embryonic hindgut cells and imaginal epithelia (Fehon et al., 1991). We thus speculated that the expression of apical LvNotch might correlate with the formation of the ZA. Previous electron micrograph studies have indicated that the ZA forms during the early blastula stage in the sea urchin embryo (Spiegel and Howard, 1983). To directly examine the localization of LvNotch in relation to the ZA, embryos were double stained for LvNotch and sea urchin Lv-cadherin and Lvβ-catenin proteins (conserved components of the ZA; Kemler, 1993). The ZA was present between all cells in the mid-blastula stage embryo (9 hours in development), a time when LvNotch was localized primarily in basolateral membranes of cells (Fig. 5A,B). In the early gastrula (13 hours), cells that expressed apical LvNotch had LvNotch colocalized with the ZA, as well as over the apical surface (Fig. 5C,D). These results demonstrate that the apical shift in LvNotch localization that occurs in the mesenchyme blastula embryo is neither correlated with the formation of the ZA, nor restricted to this junction.
Apical shifts in LvNotch subcellular localization do not correlate with the formation of the ZA. (A-D) Embryos were doublelabeled with guinea pig α-Lv-cadherin pAb and mouse α-LvNotch pAb. (A, B) In the post-hatch blastula embryo, LvNotch (A) predominates in the basolateral membranes of cells. LvG-cadherin (B) is concentrated in the ZA, which in cross-section appears as bright, apical points between cells. (C,D) In the early gastrula embryo, cells that express high levels of apical LvNotch (C) have LvNotch distributed over the apical surface (inset), as well as colocalized to the ZA (arrowheads). Double staining for Lvβ-catenin and LvNotch produced identical results (data not shown). Bars, 25 μm (10 μm in insets).
Apical shifts in LvNotch subcellular localization do not correlate with the formation of the ZA. (A-D) Embryos were doublelabeled with guinea pig α-Lv-cadherin pAb and mouse α-LvNotch pAb. (A, B) In the post-hatch blastula embryo, LvNotch (A) predominates in the basolateral membranes of cells. LvG-cadherin (B) is concentrated in the ZA, which in cross-section appears as bright, apical points between cells. (C,D) In the early gastrula embryo, cells that express high levels of apical LvNotch (C) have LvNotch distributed over the apical surface (inset), as well as colocalized to the ZA (arrowheads). Double staining for Lvβ-catenin and LvNotch produced identical results (data not shown). Bars, 25 μm (10 μm in insets).
Apical LvNotch is coincident with the presumptive SMC-endoderm boundary
Apical LvNotch in the mid-gastrula embryo appeared to be restricted to the presumptive endoderm and excluded from most, if not all, SMCs (Fig. 4H). In addition, the absence of expression in the center of the vegetal plate of the mesenchyme blastula embryo suggested that apical LvNotch may be specifically excluded from the presumptive SMCs even before invagination begins (Fig. 4G). We therefore asked if the apical expression was indeed coincident with the SMC-endoderm boundary in the mesenchyme blastula vegetal plate by comparing the number of cells either expressing or lacking apical LvNotch to a fate-map of cells of this stage (Fig. 6A,C; Ruffins and Ettensohn, 1996). The number of cells lacking apical LvNotch expression in the center of the vegetal plate correlated almost exactly with the number of presumptive mesoderm cells fate-mapped to this position, 75.5 versus 74 (66 SMC precursors and 8 small micromere descendants), respectively (Table 1). The cells (roughly 130) that express apical LvNotch were therefore coincident with the fate map’s 155 endodermal precursor cells that surround the presumptive mesoderm cells.
Numbers and asymmetry of cells expressing apical LvNotch in the mesenchyme blastula stage vegetal plate

Apical LvNotch expression in the mesenchyme blastula vegetal plate delineates the presumptive SMC-endoderm boundary and has a dorsal bias. (A) Cells expressing or lacking apical LvNotch in the vegetal plate were quantified (Table 1) by double-labeling embryos with guinea pig α-Lvβ-catenin pAb (red), which outlines the apical sides of epithelial cells, and mouse α-LvNotch pAb (blue). (B,C) Embryos double-stained with 1G8 mAb (green) specific for PMCs and guinea pig α-LvNotch pAb (blue). The bias of apical LvNotch expression is towards the dorsal side of the embryo, both in the vegetal plate of the early gastrula (B) and in the presumptive endoderm of the archenteron in the late gastrula (C). (D) A schematic map of apical LvNotch expression (blue) on the mesenchyme blastula vegetal plate was inferred by combining the above results, (A-C; Table1) and the mesenchyme blastula fate map (Ruffins and Ettensohn, 1996). The number of cells lacking LvNotch expression in the center of the vegetal plate is shown in parenthesis next to the estimated number of presumptive mesoderm cells, and the number of cells expressing apical LvNotch is similarly shown next to the estimated number of presumptive endoderm cells. Note: the fate map of Ruffins and Ettensohn (1996) did not determine the outer boundary of the presumptive endoderm in relation to the dorsoventral axis. Therefore cell counts cannot place it at this boundary. Based on the restriction of high levels of apical LvNotch expression to the presumptive endoderm in the late gastrula, we draw it at the edge of the presumptive endoderm boundary in (D).
Apical LvNotch expression in the mesenchyme blastula vegetal plate delineates the presumptive SMC-endoderm boundary and has a dorsal bias. (A) Cells expressing or lacking apical LvNotch in the vegetal plate were quantified (Table 1) by double-labeling embryos with guinea pig α-Lvβ-catenin pAb (red), which outlines the apical sides of epithelial cells, and mouse α-LvNotch pAb (blue). (B,C) Embryos double-stained with 1G8 mAb (green) specific for PMCs and guinea pig α-LvNotch pAb (blue). The bias of apical LvNotch expression is towards the dorsal side of the embryo, both in the vegetal plate of the early gastrula (B) and in the presumptive endoderm of the archenteron in the late gastrula (C). (D) A schematic map of apical LvNotch expression (blue) on the mesenchyme blastula vegetal plate was inferred by combining the above results, (A-C; Table1) and the mesenchyme blastula fate map (Ruffins and Ettensohn, 1996). The number of cells lacking LvNotch expression in the center of the vegetal plate is shown in parenthesis next to the estimated number of presumptive mesoderm cells, and the number of cells expressing apical LvNotch is similarly shown next to the estimated number of presumptive endoderm cells. Note: the fate map of Ruffins and Ettensohn (1996) did not determine the outer boundary of the presumptive endoderm in relation to the dorsoventral axis. Therefore cell counts cannot place it at this boundary. Based on the restriction of high levels of apical LvNotch expression to the presumptive endoderm in the late gastrula, we draw it at the edge of the presumptive endoderm boundary in (D).
Levels of apical LvNotch expression are differentially localized along the dorsoventral axis of the presumptive endoderm
The expression of high levels of apical LvNotch along one side of the presumptive endoderm led us to examine whether this polarity correlated with the dorsoventral axis of the embryo. This was achieved by comparing the position of the asymmetric ring of apical LvNotch in the early gastrula embryo to the orientation of the ventrolateral clusters of primary mesenchyme cells (PMCs), one of the first morphological features distinguishing the ventral from the dorsal side of the embryo. Double staining with a PMC-specific monoclonal antibody, 1G8, and LvNotch showed that 86% (12/14 embryos) of the LvNotch apical rings examined had a clear dorsal polarity while the remaining embryos appeared to have either symmetric (1/14) or lateral concentrations of apical LvNotch (1/14; Fig. 6B). To determine if this asymmetry was maintained in the presumptive endoderm during vegetal plate invagination, late-gastrula embryos were similarly evaluated. All of these embyros (19/19) had higher levels of apical LvNotch along the dorsal side of the presumptive endoderm (Fig. 6C). Apical LvNotch thus both delineates a boundary between the presumptive endoderm and mesoderm of the vegetal plate, and has a consistent dorsal polarity (Fig. 6D).
A rapid downregulation of apical LvNotch accompanies conversion of endoderm cells to SMCs
Previous studies have shown that, after microsurgical removal of the presumptive SMCs early in gastrulation, a new SMC population is established via conversion of presumptive endoderm cells to SMCs at the tip of the recovering archenteron (McClay and Logan, 1996). To confirm the association of apical LvNotch with the presumptive SMC-endoderm boundary, we repeated this experiment and asked whether apical LvNotch expression was altered during this cell lineage conversion (Fig. 7A). Immediately after removing the presumptive SMCs, apical LvNotch extended to the tip of the archenteron (12/14 embryos; Fig. 7B,C). In embryos allowed to recover for only 1.5 hours, however, apical LvNotch was no longer present at the tip of the archenteron in 88% of embryos examined (Fig. 7D; 14/16). Apical LvNotch thus appears to be rapidly downregulated in response to the lineage conversion of presumptive endoderm to SMCs.
Sequence of SMC removal, recovery and LvNotch apical expression in early gastrula embryos. (A) Diagram illustrating SMC removal via micropipet. (B) Prior to surgery, apical LvNotch is absent in the tip of the archenteron. (C) Immediately after surgery, apical LvNotch extends to the tip of the archenteron (arrowhead), but is rapidly downregulated by 1.5 hours later (D) and a new boundary of apically expressed LvNotch is established (arrow). Bar, 25 μM.
Sequence of SMC removal, recovery and LvNotch apical expression in early gastrula embryos. (A) Diagram illustrating SMC removal via micropipet. (B) Prior to surgery, apical LvNotch is absent in the tip of the archenteron. (C) Immediately after surgery, apical LvNotch extends to the tip of the archenteron (arrowhead), but is rapidly downregulated by 1.5 hours later (D) and a new boundary of apically expressed LvNotch is established (arrow). Bar, 25 μM.
Perturbations of the animal/vegetal and dorsoventral axes cause specific alterations in apical LvNotch expression
The expression of apical LvNotch along the presumptive SMC-endoderm boundary and increased expression on the dorsal side of the vegetal plate suggested that this specific pattern might be tied to both the animal/vegetal and dorsoventral axes of the embryo. To test this possibility, embryos were first treated with lithium chloride (LiCl), which increases the amount of vegetally derived tissues (Horstadius, 1973). LiCl treatment caused an expansion in the number of presumptive SMCs that lacked LvNotch and in the number of presumptive endoderm cells that express apical LvNotch in mesenchyme blastula embryos (Fig. 8A,B): the number of cells that lacked detectable LvNotch expression in the center of the vegetal plate increased 36% (P<0.05), and the number of surrounding cells that expressed apical LvNotch increased 60% (P<0.05; Table 1). LiCl also reduced the asymmetry of apical LvNotch, consistent with its less well-characterized, but known radializing effect (Horstadius, 1973; Table 1). The expansion of cells lacking and expressing apical LvNotch in LiCl-treated embryos provides further evidence that apical LvNotch is specifically expressed in the presumptive endoderm and is excluded from the presumptive SMCs.
The apical LvNotch ring and cells within this ring are expanded in mesenchyme blastula embryos treated with LiCl. Untreated (A) or LiCl-treated (B) were double-labeled for guinea pig α-Lvβ-catenin pAb (red) and mouse α-LvNotch pAb (blue). Bar, 25 μm.
To determine if the dorsal bias in LvNotch expression changes with experimental perturbation of the dorsoventral axis, this axis was disrupted by treatment with NiCl2, which increases ventral ectoderm at the expense of dorsal ectoderm, and alters the patterning of mesodermal structures associated with the endoderm and ectoderm (Hardin et al., 1992). NiCl2 treatment did not affect the number of cells lacking LvNotch expression in the center of the mesenchyme blastula vegetal plate (P>0.05; Table 1). However, disruption of the dorsoventral axis did cause a marked reduction in the overall intensity, asymmetry and number of cells expressing apical LvNotch in the vegetal plate (28% fewer cells; P<0.05; Table 1; Fig. 9A,B). This decrease in apical LvNotch continued through most of the gastrula stage (Fig. 9C,D); protein extractions revealed that mid-gastrula embryos treated with NiCl2 had an approximately 30% decrease in LvNotch protein (Fig. 9E). Thus, the presence of a dorsoventral axis appears to be required for the asymmetric expression of apical LvNotch.
NiCl2 decreases the asymmetry and the amount of apical LvNotch expression in the vegetal plate and archenteron. Compare surface views of untreated (A) and NiCl2-treated (B) vegetal plates of mesenchyme blastula embryos stained for LvNotch, as well as sections of untreated (C) and NiCl2-treated (D) late gastula archenterons. (E) Quantitative western analysis of protein extracts (200 embryos/lane) reveals a 28±5.3% s.e.m. decrease in LvNotch protein abundance in NiCl2-treated mid-gastrula embryos. The abundance of the intracellular fragment (116×103Mr) recognized by the rabbit α-Ank pAb was measured, and three different extracts from two separate trials within the linear range of exposure were analyzed (Image QuaNT; Molecular Dynamics). Bars, 25 μm (A,B); 10 μm (C,D).
NiCl2 decreases the asymmetry and the amount of apical LvNotch expression in the vegetal plate and archenteron. Compare surface views of untreated (A) and NiCl2-treated (B) vegetal plates of mesenchyme blastula embryos stained for LvNotch, as well as sections of untreated (C) and NiCl2-treated (D) late gastula archenterons. (E) Quantitative western analysis of protein extracts (200 embryos/lane) reveals a 28±5.3% s.e.m. decrease in LvNotch protein abundance in NiCl2-treated mid-gastrula embryos. The abundance of the intracellular fragment (116×103Mr) recognized by the rabbit α-Ank pAb was measured, and three different extracts from two separate trials within the linear range of exposure were analyzed (Image QuaNT; Molecular Dynamics). Bars, 25 μm (A,B); 10 μm (C,D).
DISCUSSION
Identification of a sea urchin Notch
We report the identification of the first echinoderm Notch gene LvNotch, isolated from the sea urchin Lytechinus variegatus. The deduced amino acid sequence of LvNotch shares significant homology to Drosophila and vertebrate Notch proteins (see Artavanis-Tsakonas, 1995; Uyttendaele et al., 1996). A phylogenetic analysis of the relationship of LvNotch to vertebrate Notch genes suggests that the duplication of Notch genes in vertebrates occurred after the divergence of sea urchins and vertebrates from a common ancestor, which is consistent with the proposed scenario of two main phases of gene duplication early in the vertebrate lineage (Holland et al., 1994). This analysis also implies that a single Notch gene may be present in the sea urchin genome, as appears to be the case in Drosophila. To date we have neither isolated additional sea urchin Notch genes with the degenerate primers and PCR search nor detected other cross-reacting Notch proteins with the extensive polyclonal antibodies we have generated. These data are consistent with the presence of a single Notch gene in sea urchins, but do not rule out the possiblity that additional Notch genes have been created by independent duplication events within the sea urchin lineage.
The most distinctive difference of LvNotch to other family members is the absence of EGF-repeat 14. Since two of the four known vertebrate Notch proteins contain the full complement of 36 EGF-repeats found in Drosophila Notch, the loss of EGF-repeat 14 most likely occurred in the sea urchin lineage after the divergence of sea urchins and vertebrates from a common ancestor. Interestingly, EGF-repeat 14 is the same repeat in which the split mutation in Drosophila Notch is found; a mutation that results in an eye-specific phenotype (Hartley et al., 1987; Kelley et al., 1987). While it is tempting to speculate that a specific function(s) of EGF-repeat 14 necessary for eye development in Drosophila has been lost and is not required in the sea urchin, it has been shown that two Notch-like proteins in C. elegans with different numbers of EGF-like repeats are redundant in embryonic development and functionally interchangeable (Lambie and Kimble, 1991; Fitzgerald et al., 1993). Therefore, loss of EGF-repeat 14 in sea urchin may not necessarily affect the range of interactions that LvNotch participates in during development.
LvNotch expression has an early polarity along the animal-vegetal axis and reveals regionalization of the mesenchyme blastula vegetal plate
The earliest differential pattern of LvNotch protein expression that we detected was the absence of LvNotch protein in the early blastula vegetal pole, and localization to primarily basolateral membranes in cells throughout the animal half of the embryo. This cellular pattern of expression is similar to the mRNA localization of three genes (VEB genes; Reynolds et al., 1992). Perturbation experiments with one of the VEB gene products, BP10/SpAN, which is a secreted metalloprotease and homolog of the human BMP-1 and Drosophila tolloid protein, has demonstrated a role for this protein in the patterning of the sea urchin embryo along the animal-vegetal axis (Lepage et al., 1992). It is therefore possible that LvNotch may similarly mediate cell-cell interactions important in the specification of cell-types along the early animal-vegetal axis.
Lineage analysis has demonstrated that the differential specification of SMCs and endoderm could occur at the mesenchyme blastula stage (Ruffins and Ettensohn, 1996). However, previous studies have not revealed the expression of molecular markers distinguishing these two germ-layers until invagination of the vegetal plate has begun during the gastrula stage (Ransick et al., 1993; Harada et al., 1995, 1996; reviewed by Davidson, 1993). Our analysis of the expression of apically localized LvNotch demonstrates that the SMCs and endoderm are in fact differentially specified at least by the mid-to late mesenchyme blastula stage, earlier than previous molecular studies have indicated. In addition, the dorsal bias in apical LvNotch localization and specific change in expression of apical LvNotch in vegetalized and ventralized embryos establishes that the late mesenchyme blastula vegetal plate is organized with reference to both the animal/vegetal and dorsoventral axes of the embryo before invagination of the vegetal plate begins. The continued polarity of apical LvNotch along the dorsal side of the presumptive endoderm during invagination, and reduction of this expression by NiCl2 treatment, further shows that: (1) the presumptive endoderm has a dorsoventral polarity in register with the ectodermal dorsoventral axis, (2) cells of the presumptive endoderm maintain a dorsal bias in apical LvNotch expression during the convergent-extension movements that extend the archenteron, and (3) the endoderm may be ventralized by NiCl2 treatment, much like the ectoderm is.
Potential roles for the Notch pathway in vegetal plate regionalization
It has been proposed that Notch signaling can maintain cells in an uncommitted developmental state (reviewed in Artavanis-Tsakonas et al., 1995). If this is true, the specific apical expression of LvNotch could retain much of the presumptive endoderm in an uncommitted state. Consistent with this notion is the substantial regulative capacity of the developing endoderm throughout gastrulation, indicated by experiments in which removal of the presumptive SMCs in the early gastrula results in a recovery of this mesodermal population by a group of presumptive endoderm cells switching to an SMC fate (McClay and Logan, 1996). The rapid loss of LvNotch expression at the tip of the recovering archenteron from which the SMCs have been removed suggests that presumptive endoderm cells that switch to an SMC fate rapidly downregulate apical LvNotch expression. It is thus possible that Notch signaling, by maintaining much of the presumptive endoderm in an uncommitted state, normally prevents these cells from adopting a mesodermal fate.
While the ventral bend of the pluteus larval endoderm and bilateral symmetry of associated mesodermal structures is consistent with the sea urchin endoderm containing dorsoventral polarity, the bias in LvNotch expression is the first reported evidence that a specific dorsoventral molecular asymmetry exists in the endoderm of the sea urchin. The Notch pathway provides a potential mechanism for establishing dorsoventral polarity in this structure. Consistent with this notion is that NiCl2, which causes the loss of the apparent morphological dorsoventral polarity in the endoderm (Hardin et al., 1992), also results in a reduction or loss of the LvNotch molecular asymmetry. Involvement of the Notch pathway in axial organization would not be unprecedented, as it is essential for dorsoventral polarity of the Drosophila wing and in the early C. elegans embryo (for examples see de Celis et al., 1995; Mello et al., 1994).
LvNotch expression and implications for Notch signaling
Several experiments using overexpressed fragments of Notch in cell lines and embryos have suggested that Notch may be activated by translocation of the intracellular domain into the nucleus (Jarriault et al., 1995; Kopan et al., 1996; reviewed in Artavanis-Tsakonas et al., 1995). Similar to observations in Drosophila and C. elegans (Fehon et al., 1991; Crittenden et al., 1994), however, we have not yet detected this domain of LvNotch in the nucleus. While these results do not exclude the possibility that the intracellular domain acts in the nucleus during development, they indicate that this domain is either difficult to detect or rapidly degraded upon entering the nucleus.
The specific and dynamic distributions of Drosophila Notch and C. elegans GLP-1 during development have led to the proposal that the activity of the Notch signaling pathway may in part depend upon the precise localization of the receptor (Fehon et al., 1991; Crittenden et al., 1994). Our results with LvNotch are consistent with and extend these findings. We show that dynamic cellular distributions of LvNotch to specific developmental compartments of the embryo are often accompanied by distinct subcellular localizations of LvNotch in apical or basolateral membrane domains. For example, in the early blastula embryo, LvNotch is absent from the vegetal pole but concentrated in basolateral cell membranes throughout the animal half of the embryo (Fig. 4C). Later in development, high levels of LvNotch are specifically localized on the apical surface and colocalized with the ZA of cells that are restricted to the presumptive endoderm, and the highest levels of apical expression are on the dorsal side of this structure (Fig. 6). We also demonstrate, using known components of the ZA (i.e. cadherin and β-catenin), that the ZA are present in these epithelial cells prior to these membrane domain changes in LvNotch localization. Therefore, changes in LvNotch membrane domain localization do not correlate with the formation of the ZA; rather, they appear to be specific to distinct cellular distributions of LvNotch in developmental compartments of the embryo.
While the C. elegans Notch-like GLP-1 protein has not been shown to undergo similar subcellular shifts in membrane localization (Crittenden et al., 1994), Drosophila Notch shifts from lateral membranes in embryonic epithelia to colocalize with the ZA in embryonic hindgut and imaginal epithelia (Fehon et al., 1991). In addition, in the Drosophila ovary, high levels of Notch are expressed over the apical surface of follicle cells within specific developmental stage egg-chambers (Xu et al., 1992). A direct comparison of Drosophila Notch with ZA components has not been performed; however, Notch is not concentrated on the apical surface or colocalized to the ZA in the late gastrula embryonic epidermis (Fehon et al., 1991), a time when the ZA is present in these cells (Tepass and Hartenstein, 1994). Therefore, apical shifts of Drosophila Notch also appear to be independent of ZA formation and are likely to be specific to developmental compartments of the embryo as well. Together, these results offer compelling evidence that the specific subcellular localization of Notch proteins in apical and basolateral membrane domains is a conserved and important element in Notch receptor signaling.
Interestingly, a recent study in Drosophila oogensis has suggested that Notch may mediate distinct functions on apical and lateral membrane domains (Goode et al., 1996). Molecular analysis of the Notch receptor has raised the possibility that Notch could be acting as a multifunctional receptor capable of interacting with numerous ligands (Rebay et al., 1991). Shifts in localization of Notch may therefore represent interactions with ligands differentially localized at the subcellular level. Alternatively, laterally and apically localized LvNotch could be interacting with the same ligands, but generating distinct responses in different membrane domains, as occurs in some cellular contexts with the vertebrate EGF-receptor (Lund et al., 1996).
Knowledge of the cellular and subcellular expression of proteins involved in Notch signaling, combined with experimental manipulation of these components and Notch, will be needed to understand the the full significance and functions that Notch may mediate on apical versus basolateral membrane domains. Recent advances in techniques to overexpress specific proteins in the sea urchin (Cameron et al., 1994; Mao et al., 1996) should allow us to pursue these studies, and enable a more direct understanding of the function of this pathway in the development of the sea urchin embryo.
ACKNOWLEDGEMENTS
We are indebted to Rick Fehon for advice and encouragement throughout this work, and thank Cliff Cunningham for his expertise with the phylogenetic analysis. We thank John Matese for help with figures and Nina Tang Sherwood, Gabrielle Kardon, Phil Hertzler and Rebecca Lamb for comments on the manuscript. This work was supported by NIH grant HD14483 to D. R. M., NSF equipment grant BIR-9318118, and NIH training grant 5T32GM07184 to D. R. S.