Altered expression of spatially regulated embryonic genes in the progeny of separated sea urchin blastomeres

The sea urchin pluteus is a relatively simple larva that develops from six major groups of cells (reviewed by Horstadius, 1973; Okazaki, 1975a, b). This embryo has been the subject of extensive and elegant experimental embryology since Driesch (1891, 1892) first began investigation of the developmental capacities of embryo fragments. This work defined the fate map of the embryo and demonstrated the importance of localized maternal information in early development. In addition, analysis of the developmental capacities of parts of embryos and of abnormal combinations of blastomeres clearly established a crucial role for cell-cell interactions in the differentiation of some tissues (reviewed by Horstadius, 1973). The techniques of molecular biology made possible an equally extensive analysis of gene expression in this embryo, focusing mainly on genome structure and temporal patterns and levels of control of gene expression in the whole embryo (reviewed by Davidson, 1986). To a large extent these two bodies of knowledge remain unconnected: Consequences of experimental manipulations have been interpreted primarily by morphological criteria, while characterization of gene expression in whole embryos has revealed little about which individual gene products are involved in the determination and differentiation of diverse cell types, or when distinctions are first made among cells at the molecular level.

Three approaches have allowed us to begin to relate expression of specific gene products to specific developmental events. These are techniques for enrichment of individual tissues (Okazaki, 1975a; McClay & Marchase, 1979; Harkey & Whiteley, 1980; reviewed by McClay, 1986), production of antibodies directed toward determinants restricted to specific regions or cell types of the embryo (reviewed in McClay & Wessel, 1985), and development of in situ hybridization techniques for elucidating spatial patterns of distribution of individual mRNAs (Cox et al. 1984; reviewed by Angerer et al. 1987). These approaches have identified individual proteins and RNAs that provide sensitive indicators of early commitments of cells to specific pathways of differentiation (reviewed by Angerer & Davidson, 1984). An important conclusion that emerged from use of molecular markers is that the major groups of blastomeres in sea urchin embryos have already established distinctive patterns of gene expression at the transition between cleavage and early blastula stage.

Analysis of specific gene products that signify early steps in the commitment of cells to different developmental fates provides new criteria for analyzing the effects of experimental manipulations of the embryo. These criteria have the advantages of greater independence because individual gene products may be expressed, although an entire morphological structure is not created, and of being much earlier indicators of developmental decisions than are the morphological consequences of those decisions (Angerer & Davidson, 1984; Sargent et al. 1986; Gurdon, 1987; Wilt, 1987). This kind of approach has been used recently to help unravel the interactions of genes involved in Drosophila segmentation, for example, by examining mutations in individual genes not for their morphological effects, but for the alterations they produce in the patterns of expression of mRNAs or proteins encoded by other genes (for recent reviews see Akam, 1987 and Scott & Carroll, 1987). In studies of inductive interactions in Xenopus embryos, several clones representing mRNAs expressed specifically in endoderm, mesenchyme or ectoderm have been used as indicators of commitment of cells to specific pathways of differentiation (Gurdon et al. 1984, 1985; Sargent et al. 1986). In the present study, we analyze the extent to which programs of expression of different genes in the sea urchin embryo can be executed by cells when they are deprived of cell-cell and cell-substratum interactions and positional cues normally provided by the intact embryo. To this end, we have examined the ability of separated cells derived from blastomeres continually dissociated beginning at 16-cell stage to regulate appropriately a variety of individual genes with diverse and well-characterized spatial and temporal patterns of expression.

Gametes from Strongylocentrotus purpuratus adults (Marinus Co., Westchester, CA) were collected after intracoelomic injection of 0·5M-KC1. After washing, eggs were fertilized in filtered Harvey’s artificial sea water, pH8T (SW), containing lOmm-para-aminobenzoic acid (Sigma) to prevent hardening of fertilization membranes (McCarthy & Spiegel, 1983). When the embryos reached 16-cell stage (approximately 6h; Fig. 1A), fertilization membranes were removed by passing the embryo suspension through 54 μm Nitex cloth (Showman & Foerder, 1979). Embryos were dissociated by washing three times in calcium- and magnesium-free artificial sea water containing Imm-EDTA (CMFSW; Hynes & Gross, 1970) at 4°C unless otherwise noted. In later experiments (numbers 8 and 9, Table 2) embryos were dissociated by washing in ‘cell dissociation medium’ [CDM: 20% distilled H₂O, 40% CMFSW, 40% iM-dextrose; Harkey & Whiteley (1985)] rather than CMFSW. In this case, separated cells were also collected from CDM by centrifugation at 300g, but onto a 1m-sucrose cushion. Dissociated blastomeres were cultured at 15°C starting at ∼1× 10⁵ cells ml^-1 in calcium-free sea water (CFSW; Giudice & Mutolo, 1970) containing 1 mm-[ethylene-bis(oxyethylenenitrilo)]-tetraacetic acid (EGTA) to separate daughter cells after each division. During the entire dissociation procedure, intact control embryos were held at 4°C; culture in normal SW at 15 °C was resumed at the same time that dissociated cells were placed in CFSW at 15 °C.

RNA was extracted essentially according to Nemer et al. (1984), and the concentration and purity were determined spectrophotometrically.

RNA samples were denatured and 5 μ glane^-1 fractionated by electrophoresis on 1 % agarose gels containing formaldehyde as described by Maniatis et al. (1982). After transfer overnight in 20 × SSC, membranes (Gene Screen Plus; DuPont/NEN) were baked at 80°C in vacuo for 2h. Individual comparisons of mRNA levels were made among samples analyzed on the same blot. Single-stranded RNA probes were transcribed from restriction endonuclease truncated DNA templates with SP6 or T7 RNA polymerase according to Lynn et al. (1983) to a final specific activity of 2×10⁸disintsmin^-1μg^-1 using [a-PJUTP (>3000 Ci mmol^-1; Amersham). Template DNA was removed by digestion with DNase I (RNase-free; FPLC-pure grade; Pharmacia) and, after addition of ammonium acetate to 2 M, RNA was precipitated with isopropanol (Maniatis et al. 1982). Blots were prehybridized in a solution containing 50% deionized formamide, 2 × SET (l× SET is 0·15m-NaCl, 0·03 m-Tris-HCl, 2mm-EDTA, pH8·0), 1% SDS, 1 × Denhardt’s, I mg ml^-1 phenol: chloroformextracted yeast RNA (type VI, Sigma), and 10% polyethylene glycol 8000 (Sigma) at 55°C for 1·2h. The prehybridization solution was replaced with fresh solution containing approximately 1 ×10⁻¹ cts min^-1 of RNA probe ml^-1, and the filters were incubated overnight at 55 °C in bottles on a tissue culture roller (Rollacell™, New Brunswick Scientific). Filters were washed at 68 °C (approximately T_m- 15 °C) according to the method of Church & Gilbert (1984), with two washes in wash buffer 1 (40mm-sodium phosphate, pH6·8, 0·5% bovine serum albumin [Sigma], 1 mm-EDTA, 5 % SDS) for a total of 10 min, and three washes in wash buffer 2 (40mm-sodium phosphate, pH 6·8, 1 mm-EDTA, 1% SDS) for a total of 30 min, then autoradiographed with Kodak XAR-5 film. Filters were rehybridized with different probes after removal of the previous probe with two 15 min washes in 0·1 × SET, 0·1% SDS at 95°C. Autoradiography was carried out to several extents of exposure for each assay and signals were quantified by laser densitometry to identify exposures in the linear response range of the film. Signals were corrected for minor differences in quantity of RNA on the blot by normalizing to signals obtained with a probe complementary to a 512 nt region of 5. purpuratus mitochondrial 16S rRNA (Wells et al. 1982; Eldon et al. 1987).

Cells from cultures or control embryos were adjusted to 10⁶ cells ml^-1 and 5 μl drops were pipetted onto microscope slides which had been treated with triethoxyaminopropyl silane and activated with paraformaldehyde according to the procedure of Gottlieb & Glaser (1976). For controls, intact embryos were collected by centrifugation, resuspended in OTvol. of CMFSW at 0°C and gently agitated until the cells separated. The time needed to separate control cells was 30 min and all manipulations were performed on ice to minimize RNA degradation. Slides were dried by brief heating at 50°C and then placed on an ice-cold metal tray, and the cells were fixed with 0·3 % glutaraldehyde in 2·5 % (w/v) NaCl, 50mm-sodium phosphate, pH7·5, for lh. Slides were washed briefly in fixation buffer several times and passed through graded ethanols and xylene. Samples were then processed for in situ hybridization as described by Cox et al. (1984), beginning after the normal deparaffinization step in xylene and using ³H-labeled RNA probes (lAxICdisintsmin^-1-1) at saturating concentrations. Slides were autoradiographed and developed as described by Angerer & Angerer (1981).

The DNA templates used to transcribe RNA probes are described in the following references: cytoplasmic actins Cyl, Cylla, Cyllb, and Cyllla (Cox et al. 1986); Sped and Spec2a (Hardin et al. 1988); Spec3, the 1·74 kb EcoRI fragment used in Eldon et al. (1987); collagen (Venkatesan et al. 1986); SpHboxl (Angerer et al. 1989); SpMW5 and SpMW9 (M. Winkler, Univ, of Texas, Austin, personal communication). All probes represent S. purpuratus sequences, and the spatial distributions during embryogenesis of the corresponding mRNAs are described in ‘Results’.

In these experiments, we dissociated embryos of S. purpuratus at 16-cell stage (approximately 6h), cultured the blastomeres under conditions in which daughter cells continue to separate, and compared levels of specific mRNAs in these cell cultures to those in control embryos when the latter reach mesenchyme blastula stage (24h; see Fig. 1). We chose 16-cell stage for dissociation because it presents the first morphological distinction among blastomere types: 8 mesomeres, 4 macromeres and 4 micromeres arrayed along the animal-vegetal axis (Fig. 1A). Micromeres can execute a normal pattern of gene expression (Harkey & Whiteley, 1983) and, in the presence of serum, differentiate autonomously to form primary mesenchyme cells in culture (Okazaki, 1975a). The developmental capacities of macromeres and mesomeres have not been analyzed by molecular assays. However, interpretations based on morphology indicate that these regions of the embryo, especially the animal hemisphere, cannot differentiate normally in isolation (Horstadius, 1973). These observations suggest that regulation of individual genes expressed with different tissue specificities might be differently affected in blastomeres separated at this stage. We chose 24h as the time to measure accumulation of individual mRNAs for several reasons. (1) By this time most, if not all, different cells in the normal embryo have initiated distinctive patterns of gene expression (reviewed by Angerer & Davidson, 1984 and Davidson, 1986). (2) In most cases, it avoids ambiguity of interpretation that might result for those mRNAs whose tissue specificity changes significantly in intact embryos at later stages (see below). (3) By this time mRNAs accumulate to levels that can be measured accurately.

We used probes for 11 different mRNAs whose spatial patterns of accumulation throughout development have been well characterized by in situ hybridization. Distributions of these mRNAs in late cleavage (∼12h) and mesenchyme blastula stage embryos (24 h) are diagrammed in Fig. 2. These mRNAs encompass a variety of patterns of developmental prevalence and cell type specificity, of which we describe only those features most important for the work presented here.

(1) SpMW5 and SpMW9 are abundant maternal mRNAs that probably represent the set of genes encoding proteins required for cell division (Grainger et al. 1986). Both messages are uniformly distributed in eggs and embryos (L. Angerer, unpublished observations) and decrease in abundance gradually during cleavage so that their concentrations are much reduced by 24 h. Variations in abundance of these messages during cleavage suggest that these genes are transcribed for a short period after fertilization (L. Kelso & M. Winkler, personal communication).

(2) The temporal and spatial patterns of expression of the linked cytoplasmic actin genes Cyl and Cyllb are indistinguishable, although the abundance of Cyllb mRNA is several-fold lower in all cases examined (Shott et al. 1984; Lee et al. 1986; Cox et al. 1986). Cyl mRNA, which has been examined in greater detail, is present at very low concentration in maternal RNA and accumulates rapidly and uniformly in the embryo towards the end of cleavage (Lee et al. 1986; Cox et al. 1986) as the result of transcriptional activation between 9 and 12 h (Hickey et al. 1987). At mesenchyme blastula stage (20–23 h), this message begins to disappear rapidly from aboral ectoderm and, to a lesser extent, from presumptive oral ectoderm and primary mesenchyme cells, while continuing to accumulate in presumptive secondary mesenchyme and endoderm (Cox et al. 1986).

(3) The Spec3 gene product is associated with cilio-genesis and synthesis of this message is coordinated with that of a β-tubulin mRNA (Eldon et al. 1987; Harlow & Nemer, 1987). This mRNA initially accumulates uniformly in the embryo at late cleavage stage and then, shortly before hatching blastula stage, disappears from all cells at the vegetal pole while continuing to accumulate uniformly in ectoderm during the next phase of development (Eldon et al. 1987).

(4) Cytoplasmic actin Cyllla and Sped mRNAs are present at relatively low concentration in maternal RNA, only about 3-4 % of the levels found at gastrula stage (Lee et al. 1986; Bruskin et al. 1981; Hardin et al. 1988). The maternal copies of Cyllla mRNA are not detectably localized in the egg (Cox et al. 1986). Both messages begin to accumulate at about 9 h of development as the result of transcriptional activation (Sped, Cabrera et al. 1984; Cyllla, Hickey et al. 1987) and through the remainder of development they are confined to aboral ectoderm and its precursors (Lynn et al. 1983; Cox et al. 1986; Hardin et al. 1988). The Sped (and Spec2a, see below) message encodes a protein belonging to the calmoduiin/troponin C/myosin light chain group of calcium-binding proteins (reviewed in Hardin et al. 1988).

Two other probes detect mRNAs that are also restricted to aboral ectoderm, but which accumulate later during the differentiation of this tissue. Spec2a mRNA, encoded by a small gene family whose members are structurally related to Sped (Hardin et al. 1988), does not accumulate until after hatching blastula stage. SpHboxl, a homeobox-containing mRNA, accumulates in aboral ectoderm beginning at mesenchyme blastula stage (Dolecki et al. 1986) and becomes progressively restricted to the aboral end of the pluteus larva (Angerer et al. 1989).

(5) The collagen probe detects an mRNA that accumulates mainly in primary and, to a lesser extent, in presumptive secondary mesenchyme cells beginning just before hatching blastula stage. This message is one of the first detectable zygotic gene products in primary mesenchyme cells (Angerer et al. 1988).

(6) Actin Cylla mRNA is first detectable just after hatching (around 20 h; Lee et al. 1986). At mesenchyme blastula stage, in situ hybridization shows that it is found primarily in presumptive secondary mesenchyme cells and, briefly, in some primary mesenchyme cells. This mRNA is restricted to secondary mesenchyme cells through late gastrula stage, but subsequently disappears from these cells and accumulates to similar high levels in part of the endoderm (Cox et al. 1986).

Dissociated blastomeres were cultured in stirrer flasks at 15 °C in calcium-free sea water containing EGTA (CFSW). Under these conditions there was little, if any, aggregation of separated cells. Cell viability at 24 h, monitored by trypan blue dye exclusion, was always greater than 95 %. Viable cells were counted at 24, 48 and 72 h and, as shown in Fig. 3, the kinetics of increase in cell number were similar to those of intact embryos: There is a rapid increase between 6 and 24 h followed by a slower increase between 24 and 72 h. Although viable cell counts are probably less accurate at later stages, due to the small size of the cells, the average number of separated cells at 72 h was at least within a factor of 2 of that in normal embryos. At all times we observed some variation in cell size. At 24 h, the largest cells (3% of the population), had volumes about one-half that of micromeres, but the volumes of the large majority of cells were two- to fivefold less than those of these large cells. Thus virtually all cells make at least some divisions after initial separation, and the population on average makes almost the normal number of divisions.

Dissociated cells showed no morphological differentiation characteristic of cells in the mesenchyme blastula. They remained spherical, with centrally located nuclei, and were not visibly ciliated. At longer culture times (48 or 72 h), there was no evidence of pigment granules characteristic of some secondary mesenchyme cells or of spicule material normally synthesized by primary mesenchyme cells.

To compare concentrations of individual mRNAs in separated cells and control embryos, total RNA was purified at 24 h as described in Materials and methods and analyzed on blots hybridized with ³²P-labeled RNA probes. All blots were rehybridized with a probe for mitochondrial 16S rRNA that is present at constant concentration throughout development (Wells et al. 1982; Eldon et al. 1987), and these signals were used to normalize for slight differences in loading. This correction was usually 5−10% and never more than 20%. Examples of these data, shown in Fig. 4, demonstrate that the levels of different mRNAs are affected by cell dissociation in dramatically different ways. SpMW5 and SpMW9, abundant maternal mRNAs, are the only messages examined that decrease in abundance in normal embryos between 6 and 24 h and these continue to decay, although to reduced extents in dissociated cells. (The blot shown in Fig. 4 has been overexposed to show the signal for RNA extracted from intact cells. Quantification of the time course of this decay is illustrated in Fig. 5.) In the particular experiment shown in Fig. 4, effects of dissociation on other mRNAs range from almost complete failure to accumulate (e.g. Sped), through levels similar to those in control embryos (e.g. actin Cyl) to significant overaccumulation (e.g. actin Cylla).

In initial experiments, we tested two variations in culture conditions that might affect accumulation of these mRNAs. To determine the effect of absence of calcium in the medium, we cultured dissociated cells at 10-fold lower concentration in normal sea water with faster stirring in order to maintain separation (Arceci & Gross, 1980). This control is especially important in the case of Sped (and probably Spec2a) mRNA, which encodes a calcium-binding protein (Muesing et al. 1984). We also cultured cells in the presence of 2% horse serum, which has been reported to improve viability of dissociated cells from embryos of Lytechinus (Arceci, 1980) and Strongylocentrotus (McCarthy & Spiegel, 1983) and to be required for differentiation of primary mesenchyme cells (Okazaki, 1975a). No effect on mRNA accumulation was observed when Ca²⁺ or serum was added separately (data not shown) or when they were added simultaneously (Fig. 4 and Table 1). Under these different conditions we detected no differences in division rate or morphology of the separated cells compared to those of cells cultured by our standard method (data not shown).

Results from 9 separate experiments using the standard culturing protocol are summarized in Table 2. The values given are mRNA abundances in dissociated cells at 24 ± 1 h expressed as percentages of the values for corresponding control mesenchyme blastulae. Included are data from one experiment (5) in which cells were dissociated at 15°C, the normal culture temperature, rather than at 4°C as in all other experiments. This variable also did not have a significant effect. The levels of some mRNAs in dissociated cells relative to those in normal control embryos varied considerably among different cultures. In order to determine what portion of this variation could be attributed to differences in levels of individual mRNAs among control cultures, we hybridized six of the probes to blots containing RNAs from seven batches (experiments 1-7) of control 24 h mesenchyme blastulae. Table 3 shows the standard errors for these measurements, with the average of values among different batches for each mRNA taken as 100 %. The small variation in mRNA levels among batches of control embryos derived from different females at different times spanning two breeding seasons indicates that absolute levels of these mRNAs are rather tightly regulated in normal embryos. We conclude that differences observed for the ratios presented in Table 2 reflect primarily differences in mRNA abundance among batches of dissociated cells. The data for dissociated cells do not reveal any consistent correlation of values that could be attributed to differences in cell viability among experiments. Levels of different messages expressed primarily or exclusively in ectoderm (Cyllla, Sped, Spec3 and Spec2a) show some correlation which we consider further in the Discussion.

To determine whether dissociation of cells leads to alterations in timing of accumulation of individual messages, we examined intermediate time points covering the period during which, in normal embryos, levels of the abundant maternal mRNAs (SpMW5 and SpMW9) decrease while those of the other mRNAs increase: 8h, allowing only time for any transient effects of the dissociation to pass; 13h, when intact embryos (∼200-cell stage) first detectably accumulate some of these mRNAs; 18h, about 300-cell blastula stage, the usual time of hatching and before primary mesenchyme cells have ingressed; and 24 h, about 400-cell mesenchyme blastula stage. To minimize variation, all samples were derived from the same single-pair fertilization. Results for individual mRNAs are shown in Fig. 5, in which values for separated cells are shown as percentages of control levels at 24 h; insets in each panel show the RNA blot signals.

Based on differences in the kinetics of accumulation of individual mRNAs in dissociated cells, we consider them in four classes. (1) Both maternal RNAs, SpMW5 and SpMW9, decay in control embryos and in dissociated cells, although the latter maintain consistently higher levels. (2) Three mRNAs, those encoding actins Cylla, Cyllb, and Cyl, surprisingly show initial accumulation to higher levels in separated cells and then level off or decline while they continue to accumulate in control embryos. (3) Spec3 mRNA accumulates and decays with kinetics similar to those in control embryos, although at all stages levels are reduced compared to controls. (4) All other messages accumulate at lower (collagen) or much lower (Sped, actin Cyllla) rates in dissociated cells than in controls, but beginning at about the same time. Analyses at 48 and 72 h show that the levels of all of these mRNAs decline after 24 h (data not shown).

These data provide additional insight to the sources of variation in the standard 24 h assays shown in Table 2. First, the kinetics of accumulation of actin Cylla and Cyllb and, to a lesser extent, Cyl mRNAs, differ significantly in dissociated cells and embryos. Clearly small differences of timing in different cultures would have large effects on the ratios presented in Table 2 and, as expected, these transcripts show the greatest variability of the newly accumulating mRNAs at 24h. Second, although the abundances of SpMW5 and SpMW9 mRNAs are about 5-fold greater, on average, in dissociated cells at 24h, this difference appears to result from a slight pause in decay after dissociation, followed by resumption at a rate within 2-fold of that in control cells. For Sped and actin Cyllla, these data show that at no time are dissociated cells capable of accumulating levels of mRNA similar to those in control embryos.

RNA blot assays do not distinguish whether changes in abundance of individual mRNAs in dissociated cells are due to corresponding changes in the number of cells expressing those mRNAs, the abundance of the mRNA per cell, or both. We used in situ hybridization to address this question for two of the messages. Because separated cells cannot be identified by morphology, this approach is useful only when changes in mRNA expression are relatively large. Furthermore, levels of hybridization that appear high on sections of intact embryos correspond to relatively few autoradiographic grains per single cell. We analyzed Sped mRNA because it is the most abundant representative of the set of messages that is restricted to presumptive aboral ectoderm of the normal embryo (Lynn et al. 1983) and because its expression is greatly reduced in dissociated cells. We examined actin Cylla message because, in the normal 24 h blastula, it is confined to a small subset of cells, the secondary mesenchyme, but it is markedly overexpressed in dissociated cells.

For this analysis, dissociated cells were allowed to settle onto slides and processed for in situ hybridization. For a control population, 24 h embryos from the same experiment were dissociated into individual cells in CMFSW, and these separated cells were immediately processed in the same manner. Histograms of the percentage of cells with different numbers of autoradiographic grains are presented in Fig. 6. For Sped mRNA, these data show a class of heavily labeled cells in control embryos (Fig. 6A) that is absent from the dissociated cell population (Fig. 6B). Fig. 6C shows a difference histogram constructed by subtracting the values for each grain density class of Fig. 6A from the corresponding value of Fig. 6B. This difference shows that dissociated cells lack the class of labeled cells present in control embryos, denoted by the shaded bars in Fig. 6C, corresponding to about 40% of the total cells. In situ hybridization to sections of intact embryos in the same experiment shows a single labeled region including ∼35 % of cells that constitute aboral ectoderm (Lynn et al. 1983; Hardin et al. 1988; data not shown). It is clear that few, if any, dissociated cells accumulate quantities of Sped mRNA as high as those found in presumptive aboral ectoderm of intact embryos. This result excludes an interpretation in which a subset of presumptive aboral ectoderm cells expresses Sped mRNA at normal levels while other cells do not.

Hybridization with the actin Cylla probe revealed that many more cells accumulate this mRNA in dissociated populations than do in control embryos. Comparison of the histogram for control cells (Fig. 6D) to that for dissociated cells (Fig. 6E) clearly demonstrates a large increase in the percentage of cells with higher grain densities (mode of 6-7/arbitrary unit area). The difference histogram shown in Fig. 6F shows that ∼35 % of the dissociated cell population (shaded bars) accumulate Cylla mRNA whereas only about 8% of cells, the presumptive secondary mesenchyme, express Cylla message in normal blastulae (Cox et al. 1986). We conclude that the large increase in abundance of Cylla mRNA in dissociated cell populations reflects primarily a similar increase in number of cells expressing this message.

Analyses of spatial and temporal patterns of gene expression in the early sea urchin embryo have shown that the major regions that give rise to different tissues establish distinctive patterns of mRNA accumulation by blastula stage (reviewed by Angerer & Davidson, 1984; Davidson, 1986). Many of the messages that have been examined in this study are essentially absent from the maternal RNA population and accumulate in only single or restricted sets of tissues as a result of new embryonic transcription. Several others exhibit initial uniform accumulation in the embryo followed by tissuespecific decay. Synthesis of high complexity nuclear RNA is thought to be initiated around 16-cell stage (Ernst et al. 1980) and accumulation of new zygotic tissue-specific mRNAs is first detectable around the 100- to 150-cell stage (Lee et al. 1986; Eldon et al. 1987; Hickey et al. 1987; Hardin et al. 1988). In the intact embryo, initiation of tissue-specific accumulation and decay of different messages presumably signal early steps in commitment of blastomeres to restricted fates. The experiments presented here examine the extent to which separated cells can execute these early events of regulation of gene expression.

A concern in this experimental approach is that failure of cells to accumulate some mRNAs results from damage to the cells. We do not believe that the cell separation procedure, per se, is damaging because an extensive literature shows that separated blastomeres, including those separated in calcium-free sea water, can contribute differentiated cells in chimeras (reviewed by Horstadius, 1973). Furthermore, experiments in which 16-cell blastomeres are immediately reaggregated show that these cells can reform embryo-like structures with differentiated tissues (reviewed by Spiegel & Spiegel, 1985) and can carry out patterns of RNA accumulation much more similar to those of control embryos (Hurley et al. in preparation). To avoid questions of long-term cell viability, we used a minimal separation time that is sufficient for the appearance and accumulation of a number of specific mRNAs. Several observations attest to the healthiness of separated cells during this period. (1) Between 6 and 24 h separated blastomeres continue to divide at rates close to normal and are still capable of dividing several more times, on average, during the next 48 h. Cells continue to exclude trypan blue dye during this period. (2) Separated cells accumulate significant quantities of several mRNAs, including actin Cyl and Cyllb messages, that accumulate uniformly in all cells of normal embryos during this period. They also turn over several abundant maternal mRNAs (SpMW5 and SpMW9) at rates similar to those in normal embryos. Arceci & Gross (1980) have previously shown that micromeres, mesomeres and macromeres cultured separately are also able to execute the switch in synthesis from early (a) to late (J3, y, etc.) chromosomal histone variants. (3) Addition of 2% horse serum, reported to improve viability of separated cells (Arceci, 1980; McCarthy & Spiegel, 1983), and/or culturing in the presence of calcium, also.had no detectable effect on message accumulation (Table 1, Fig. 3). (4) We found that cells dissociated at 16-cell stage and cultured to 24 h do not carry out unsolicited synthesis of the major 70×10³Af_r heat shock protein (data not shown), although Roccheri et al. (1981) have shown that normal blastulae as well as cells dissociated and cultured as in our experiments develop the capacity to mount a heatshock response. (5) Cells cultured up to 24 h in CFSW and then reaggregated incorporate [³⁵S]methionine at rates equivalent to those of intact embryos (our unpublished observations).

Results from studies of classical experimental embryology, which examined the developmental capacities of parts of embryos and of abnormal combinations of early blastomeres, have been interpreted as providing evidence that the fates of most regions of the embryo are determined by reciprocal gradients of animalizing and vegetalizing tendency emanating from the corresponding poles and mediated by localized maternal factors (Rünnstrom, 1928; reviewed by Horstadius, 1973, and Rünnstrom, 1975). Micromeres have clearly demonstrated inductive powers and the ability of their progeny to differentiate cytologically and biochemically as primary mesenchyme cells when cultured in isolation from other blastomeres (Okazaki, 1975a) strongly suggests that localized maternal factors are sufficient for their determination. In contrast, it is not clear whether the large mass of presumptive ectoderm contains a region of animal pole cytoplasm with analogous but reciprocal function to that of the vegetal pole (reviewed by Angerer & Davidson, 1984; Davidson, 1989). Isolated animal hemispheres cannot carry out much of the cytological differentiation observed in the ectoderm of intact embryos. Regardless of the exact mechanism envisioned, the classic experiments indicate that fates of macromeres and mesomeres are not irreversibly determined by 16-cell stage. The daughter cells of these 16-cell blastomeres retain considerable plasticity of fate at least through blastula stage, which can be revealed by altering the relative numbers and positions of different blastomeres (reviewed by Horstadius, 1973). A straightforward, although probably simplified framework for discussing our results is that cell separation distinguishes between events that can be initiated within lineages of cells entirely by maternal molecules and those that are affected by the organization of the embryo. This ‘organization’ may encompass interactions among cells mediated by cell-cell contacts, by contacts with extracellular materials deposited by individual cells or their neighbors, and by short-range diffusion of molecules. Our results are consistent with greater maternally derived independence of events of gene expression for cells derived from the vegetal pole, than for those derived from the presumptive ectoderm.

As illustrated in Fig. 2, five of the mRNAs examined initially are expressed uniformly in the intact embryo. In general, early uniform expression and decay of messages appear to be maternally programmed. (1) The abundant maternal messages SpMW5 and 9 decay rapidly and uniformly in normal embryos. They also disappear from dissociated cells, although to a slightly lesser extent, perhaps related to the fact that a small percentage of these cells divides more slowly. (2) The mRNA encoding the cytoplasmic actin Cyl accumulates early and uniformly in the embryo as does, presumably, the coordinately regulated actin Cyllb message. Subsequently, the abundance of these messages is modulated differently in different regions of the embryo. This includes down-regulation between 12 and 20 h in a portion of the ectoderm, which is one of the earliest manifestations of the differentiation of oral and aboral regions. The simplest explanation of the overaccumulation of both of these mRNAs is that it results from the inability of some separated cells to execute this tissuespecific down-regulation. The later decline in abundance of each message in separated cells presumably results from the inability of these cells to carry out the subsequent accumulation that normally occurs in other tissues. (3) Spec3 mRNA, encoding a cilia-associated protein, also accumulates early and uniformly in the normal embryo. It continues to be expressed in all presumptive ectoderm (oral and aboral) but by blastula stage is undetectable in the smaller fraction of cells that are the precursors of mesenchyme and endoderm. Without the knowledge that the Spec3 protein is associated with cilia, we would predict that Spec3 mRNA would accumulate to nearly normal levels. In fact, while its kinetics of appearance and decay mimic those in normal embryos, its levels are somewhat lower at all stages, which is undoubtedly related to the fact that separated cells do not generate cilia. Suppression of Spec3 mRNA accumulation in dissociated cells may result from some feedback control, perhaps dependent on products of other genes. For example, it has been shown that Spec3 mRNA abundance is coordinately regulated with that of a message encoding β-tubulin in normal embryos and throughout cycles of deciliation and reciliation (Eldon et al. 1987; Harlow & Nemer, 1987).

Zygotic transcripts of four of the genes examined, Sped, actin Cyllla, Spec2a and SpHboxl, accumulate specifically in precursors of aboral ectoderm during normal development. Expression of these messages in separated cells is reduced, on average, at least 7-fold (Table 2). This is a large reduction, especially in view of the tight control of mRNA abundance in different cultures of control embryos (Table 3). Two observations indicate that this reduction does not represent a simple retardation of development of ectoderm. First, the number of cell divisions by 24 h is not significantly reduced in separated cells. Second, assays at 48 and 72 h showed no further increase in the levels of these mRNAs (our unpublished observations).

If expression of Sped and actin Cyllla mRNAs reflects differentiation of aboral ectoderm, a process that we suggest depends on cell-cell interactions, then why should these mRNAs accumulate at all in separated cells? Our in situ hybridization results show that Sped mRNA does not accumulate preferentially in a small subset of separated cells that might be derived from, for example, the extreme animal pole, or the blastomeres closest to the inductive influence of the cells at the vegetal pole. We cannot determine whether the population of cells expressing at low levels is restricted to presumptive aboral ectoderm. We suspect that continuous contact throughout cleavage is required for accumulation of normal levels of aboral ectodermspecific messages, since experiments in which cells were separated at each cleavage stage from 4-cell to 128-cell have not identified any discrete stage at which expression of these messages becomes autonomous; nor does separation at stages earlier than 16-cell result in further reductions of the levels of these messages (Hurley et al. in preparation). Thus, it seems more likely that initial accumulation of Sped and Cyllla mRNAs might be sponsored by maternal molecules. The fact that low levels of these mRNAs are found in eggs (Bruskin et al. 1981; Lee et al. 1986; Angerer et al. 1989) supports the idea that early zygotic accumulation of these mRNAs is a continuation of an oogenetic program.

Although accumulation of collagen and actin Cylla messages is not a very early event in the normal embryo, the fact that their expression is confined to vegetal pole derivatives, primary and secondary mesenchyme cells (Cox et al. 1986; Angerer et al. 1988), predicts that they should be expressed at levels similar to those in normal embryos. Collagen mRNA, which in normal embryos accumulates mainly in primary mesenchyme cells, reaches a level that is, on average, about one-third that in control embryos. Autonomous differentiation of primary mesenchyme cells has been demonstrated only when they are cultured on a surface in serum and allowed to form a syncytium. Maintenance of normal levels of collagen gene activity may depend on homotypic cell contacts and/or deposition on an appropriate substratum of extracellular matrix material, of which collagen itself is an integral part (Wessel & McClay, 1985). In experiments similar to ours, Stephens, Kitajima and Witt (personal communication) have found that the primary mesenchyme cellspecific message Sm50, which encodes the major spicule matrix protein, is expressed in dissociated cells at levels about 50% of those in control embryos.

Accumulation of Cylla message to abnormally high levels in separated cells is a surprising result. Our in situ hybridization measurements demonstrate that the number of dissociated cells expressing this mRNA is 4- to 5-fold higher than in normal embryos. An interesting possibility is that expression of this message is normally repressed in cells adjacent to micromeres and that separation of cells prevents this repression. Both older experimental embryology (reviewed in Hörstadius, 1973) and more recent experiments indicate that the fates of cells derived from the vegetal portion of macromeres are not determined autonomously. For example, the ability of secondary mesenchyme cells to ‘count’ and replace primary mesenchyme cells removed from the blastocoel (Fukushi, 1962; Ettensohn & McClay, 1988), indicates that some form of communication among vegetal pole cells is involved in specifying different cell fates. It is interesting that the only other set of cells that normally expresses the Cylla message is the differentiating endoderm of the prism and pluteus larva, which also derives from the vegetal portion of the macromeres.

Our molecular data relate to morphological descriptions of the fate of animal half embryos when separated at progressively later stages up to mesenchyme blastula. These show increasing ability to differentiate oral and aboral regions (Hörstadius, 1935, 1936a, b). Animal halves from 16- to 32-cell-stage embryos produce mostly permanent blastulae with elongated cilia, and little or no definition of oral and aboral regions. Animal half embryos also progressively lose their competence for induction by implanted micromeres (Hörstadius, 1936a). These observations are part of the classical evidence that normal differentiation of the ectoderm requires continued communication with more vegetal cells during cleavage and early blastula stages. During this period, we now know that different aboral ectoderm-specific messages accumulate in sequence (for example, see Hardin et al. 1988). Separation of blastomeres from 16-cell stage onwards does not suppress early regulation of mRNA abundance characteristic of all ectoderm, but does interrupt the animal-vegetal interaction required for normal accumulation of aboral ectoderm-specific messages and, probably also, the tissue-specific down-regulation of mRNAs initially expressed uniformly in the ectoderm. The variation in ability of embryos in different cultures to express aboral ectoderm-specific mRNAs is molecular data that parallels morphological descriptions in the literature showing that animal regions from different batches of eggs vary considerably in the extent to which they can differentiate. Such differences may ultimately relate to variations in the size of maternal stockpiles of regulatory molecules.

Further investigation of the exact cellular interactions necessary for specification of oral and aboral ectoderm will require additional markers for messages expressed at different times, especially ones expressed specifically in oral ectoderm, and investigation of the corresponding gene activities in separated blastomere types alone and in different combinations.

This work was supported by a grant from the National Institutes of Health to R.C.A. and L.M.A. (GM25553). R.C.A. is the recipient of a Research Career Development Award from the Public Health Service (HD602). Collaboration with the laboratories of Eric Davidson, Tom Humphreys and Greg Dolecki, Bill Klein, Robert Simpson, and Matt Winkler has made possible development of the set of characterized cloned probes used in this study. We thank Laurie Stephens and Fred Wilt for exchange of ideas and communication of results before they were submitted for publication. We thank Ed Stephenson for helpful suggestions on the manuscript.