Oral Ectoderm Regulatory States from Cleavage to Gastrulation.

As shown in Fig. S1A, cohorts of regulatory genes are expressed in increasingly complex patterns as development proceeds. The process of regulatory state formation is summarized in diagrams (Fig. 1), which are based on single and double in situ hybridization from earlier studies of regulatory gene expression (5, 8⇓⇓–11, 14, 15) and from detailed additional observations that we reproduce in Fig. S1. Among 35 oral ectoderm and apical regulatory genes previously reported, 12, generally the earliest markers of each individual regulatory state domain, were selected for further examination. These genes play critical roles in regulatory state formation and diversification. Their quantitative expression dynamics had been measured earlier (16), and the relevant kinetic data are reproduced in Fig. S2. The detailed sequence of expression patterns shown in Fig. S1 can be abstracted to provide the dynamically changing Boolean expression matrices shown in Fig. 2, where the contributions of the 12 genes to the regulatory state of each domain can be read horizontally (6, 7). These are the specific patterns of expression for which we seek causal explanation in the encoded architecture of the ectodermal GRNs.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Diagram of ectodermal regulatory states with increasing complexity from the late cleavage stage to the onset of gastrulation. (A) Maps showing regulatory state formation. Developmental stages include the late cleavage stage (9 h), blastula stage (12, 15, and 18 h), mesenchyme blastula stage (24 h), and early gastrula stage (30 h). The embryos are shown with oral ectodermal views unless specified otherwise. lv, lateral view. The asterisk (*) following “oral animal ecto” indicates near-apical + central ectoderm. (B) Events of cell-state divergence during early embryogenesis, highlighting ectoderm diversification. Red orthogonal arrows indicate separation of regulatory states along the oral-aboral axis, and green arrows indicate the animal-vegetal axis; black orthogonal arrows indicate cell lineage separation.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Expression matrix of ectodermal regulatory genes. Expression territories of the regulatory genes were mapped using WMISH (Fig. S1) and summarized in the matrix table. All domains and subdomains refer to the ectoderm unless otherwise specified. The matrix table includes only regulatory states on the oral or the lateral side of the embryo. Developmental stages include 12 h (early blastular stage), 18 h (late blastular stage), 24 h (mesenchyme blastular stage), and 30 h (early gastrula stage).

The ectodermal boundary formation events in both time and space that are implied in the patterns of Fig. 1A are abstracted in the regulatory process diagram shown in Fig. 1B, where the orthogonal pairs of arrows mark the institution of regulatory state boundaries—red for boundaries in the oral/aboral axis and green for boundaries that arise along the animal/vegetal axis. These boundaries are the subject of the experimental work in this paper (except for one arising within the aboral ectoderm, the bottom-right image in Fig. 1A). An initial transient regulatory state boundary, already evident at 7 h, forms at the equatorial cleavage planes where it separates eve-expressing blastomeres that will give rise to endomesodermal constituents (macromere descendants) from foxq2-expressing blastomeres ancestral only to ectodermal and neurogenic components of the embryo (mesomere descendants) (Fig. S1, 7 and 9 h). A new regulatory state boundary is then established, which separates the nonapical ectoderm from the apical domain (10–15 h). A boundary between the transient regulatory states of all veg1 lineage cells and that of the overlying oral ectoderm forms at 12–18 h; at 15–18 h, another regulatory state boundary delimits the near-apical from the central-oral ectoderm regulatory states; and finally, at ∼24 h, the definitive regulatory state boundary separating veg1 posterior endoderm from veg1 ectoderm is established. Boundaries within the apical neurogenic domain have not been analyzed, whereas formation of boundaries within the endomesoderm was solved previously (5, 6). We adduce earlier evidence regarding oral ectoderm regulatory state boundaries that form along the oral/aboral axis and integrate it with the animal/vegetal boundary mechanisms below.

GRN Interactions Controlling the Apical Domain/Oral Ectoderm Boundary.

The mechanism leading to specification of the apical neurogenic domain begins with the zygotic expression of foxq2, which, as we have just seen, is activated during the fifth cleavage in all cells of the animal hemisphere (Figs. S1 and S2). A sharp upward retraction in the domain of foxq2 expression occurs between 9 and 12 h, however, and this process continues, so that by 15 h the spatial expression of foxq2 marks the apical plate region; thereafter, this gene is stably expressed in the central region of the apical plate domain (Fig. S1, 9–30 h). Our initial objective, in unraveling the gene interactions that set the boundary between the apical domain and the oral ectoderm, is to understand the cause(s) of retraction of the foxq2 expression domain until it arrives at and defines the apical plate boundary because it is the first apical-specific regulatory gene to be expressed. Some evidence indicated that the earliest 9- to 12-h phase of retraction could be mediated indirectly by Wnt8 signaling from the vegetal blastomeres, and indeed effects of Wnt signaling on the size of the apical domain were reported earlier (17). We show here, however, that after 12 h a known transcriptional repressor expressed in the oral ectoderm, not, which, however, is not a direct target of Wnt signaling, is specifically responsible for preventing foxq2 expression all the way down to the equator. This is demonstrated by the dramatic spatial effects of not morpholino antisense oligonucleotide (MASO) on foxq2 expression at 15 h, which, as seen in Fig. 3A, is then restored down to the oral equator. Fig. 3A also shows that this MASO effect cannot be seen 3 h earlier. The not gene, a direct early target of Nodal signaling (10), is expressed significantly in the equatorial oral ectoderm by 12 h and thereafter (Fig. S2).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

MASO perturbation and expression analysis to identify the roles of regulatory genes in establishing boundaries along the primary axis. Perturbation assays were performed to knock down expression of foxq2, not, or emx (A), eve (B), six3 and/or foxq2 (C), hox11/13b (D), or bmp2/4 (E). Expression pattern changes were investigated using WMISH. The apical region of endogenous foxq2 expression is marked with green arrowheads or dotted circles (A, C, and E). Red and green dashed arcs in B indicate the veg1 and veg2 regions, respectively. In D, the red arrowheads mark hox11/13b expression; the blue arrowheads indicate the gap between the veg1 ectoderm and the veg2 endoderm, and the black arrowheads show expansion of lim1/vegf3 expression into the veg1 endoderm in the hox11/13b morphants. A broader survey of more regulatory genes in response to MASO perturbation was shown in Fig. S3. lv, lateral view; vv, vegetal view; av, apical view; ov, oral ectodermal view.

Fig. S1 also shows that transcripts of the homeobox gene emx, the zinc-finger gene egr/z60 (18), and the Tgfβ gene univin abut the lower boundary of foxq2 expression at 15 and 18 h and that all three are transcribed in the regions from which foxq2 has earlier cleared. This is confirmed in the control emx/foxq2 double in situ hybridization shown in Fig. 3A, and in Fig. S3A the same is shown for univin and z60. The emx gene repressively controls the lower boundary of foxq2 expression after the midblastula stage, as shown by the spread of this domain into the near-apical oral ectoderm in emx MASO-treated embryos (Fig. 3A). Reciprocally, as Fig. 3A shows, foxq2 expression is required to prevent expression within the apical domain of emx (and also of univin and egr) (Fig. S3A). Thus, emx and foxq2 are linked in a mutual exclusion circuit (4, 19, 20), in which, within the normal domain of expression of each, expression of the other is prohibited. Thereby, the boundary separating foxq2 expression from the oral ectoderm regulatory state is locked down. The logical elegance of this boundary formation mechanism is underscored by the use of a common driver by both foxq2 and emx genes. This is the positively acting, pan-ectodermal regulator SoxB1 (21): as shown by the quantitative MASO perturbation experiments reproduced in Fig. S4, by 24 h emx and foxq2 expression are, respectively, about 90 and 97% depressed by knockdown of SoxB1 translation. Note that, later in blastulation, expression of emx, univin, and egr is cleared from the oral and/or aboral ectoderm (Fig. S1A) by repressors active in that domain, whereas their boundaries with foxq2 persist on the oral side of the embryo.

Gene Interactions Controlling the Boundary Between Animal and Vegetal Ectoderm.

All cells deriving from the animal half of the sea urchin embryo give rise to ectoderm, but, in addition, a few cells of the vegetal half located immediately beneath the equator also assume ectodermal cell fates. These cells are descendants of the veg1 lineage, which also gives rise to posterior endoderm. In the ancestors of veg1 ectodermal cells, eve expression initiates at 7 h, representing the first regulatory distinction from animal ectodermal cells (5, 6, 22). Until about 12 h, eve remains broadly expressed in the vegetal half, but by 15 h its expression is confined to the veg1 lineage, and transcript levels accumulate (Fig. S2) (5). Expression of eve precedes expression of the genes constituting the regulatory state of the adjacent nonapical ectoderm such as emx and lim1 (Figs. S1A and S2). When emx is activated, and until after 15 h, it is expressed all over the nonapical ectoderm, and in addition its lower boundary of expression overlaps with that of eve. The univin gene is expressed in exactly the same way (Fig. S1A). After 18 h, however, transcripts of both genes have cleared from the veg1 eve domain (emx is then expressed in the lateral and aboral ectoderm). This behavior suggests that eve is acting as a repressor of genes that define the animal ectoderm regulatory states on both the oral and the aboral side of the embryo.

The experiments of Fig. 3B demonstrate this eve function. In the first column, we see double in situ hybridizations showing eve and emx and eve and univin. Expression of univin still partly overlaps that of eve at 18 h, but its expression and that of emx have become exclusive with respect to eve on either side of the veg1/animal ectoderm boundary by 24 h. Two additional genes expressed in the oral ectoderm GRN, gsc and foxG (9, 10), are also expressed right down to the boundary with the veg1 eve expression domain at 24 h. The second column of Fig. 3B shows expression of these same four ectoderm genes together with that of the veg2 regulatory gene foxa: a gap of three to four rows of cells deep, which corresponds to the veg1 domain, is clearly visible for all four genes. The third column demonstrates for each of these genes that, if eve expression is blocked by introduction of MASO, their expression extends right down to the veg2 cells, and the veg1 gap no longer exists. Therefore, eve, which defines veg1 regulatory identity and initiates the veg1 GRN (6), also functions to exclude expression of these ectoderm genes from the veg1 domain. Thereby, the veg1 lineage boundary is maintained as a regulatory state boundary, i.e., until the following boundary transition in this region, which subdivides veg1 into posterior endoderm and ectoderm.

Transcriptional Repressions Further Partitioning the Oral Ectoderm.

The lim1 regulatory gene is activated by ∼10 h, 2–3 h after foxq2 and six3 (Fig. S2). Its expression is also driven by the pan-ectodermal activator SoxB1 (Fig. S4). A priori this gene should therefore be able to express throughout the ectoderm, but, instead, from the outset its expression appears as a band extending around the embryo from the veg1 domain below the equator up to about halfway into the animal oral ectoderm (Fig. S1B, 12–21 h). Its upper boundary of expression implies repression of lim1 in the apical and near-apical animal ectoderm. At 15 and 18 h, the expression pattern of six3 (15, 23) appears perfectly reciprocal to that of lim1. Thus, as seen in lateral view, six3 transcripts occupy the apical domain plus the near-apical animal ectodermal domains (Fig. S1B). An experiment shown in Fig. 3C demonstrates that six3 and foxq2 are indeed responsible for excluding lim1 expression from the apical and near-apical ectoderm because foxq2 + six3 MASOs cause lim1 expression to extend to the whole animal half of the embryo. The central animal ectoderm and the near-apical animal ectoderm regulatory states are thereby separated by the lim1/six3 boundary. This boundary soon pertains only to the oral side, where it persists, as lim1 expression is lost from the aboral ectoderm between 21 and 24 h, as is six3 expression (Fig. S1B). Furthermore, an additional spatial repression mediated by foxq2 precludes six3 expression in the central apical domain (Fig. 3C). Thus, the relations foxq2 repressing six3; foxq2+six3 repressing lim1 produce a central disk of foxq2 expression; a surrounding torus of six3 expression, the lower boundary of which bisects the nonapical animal ectoderm; and an abutting lower torus of lim1 expression. Following the confinement of lim1 and six3 to the oral side of the embryo by aboral ectoderm repression (see below), these boundaries persist in the oral ectoderm and oral apical domains (Fig. S1B).

Exclusion of the Oral Ectoderm GRN from the Apical Domain.

Repression by foxq2 plus six3 has a further spatially specific effect, the ultimate significance of which expands as early development proceeds. This is the repression of nodal expression. Although the repressive role of foxq2 was previously proposed (14), the regulatory circuit governing the dynamic nodal expression is more complicated and involves synergetic repression. As summarized in Fig. 2, the nodal gene is expressed strongly in the central oral ectoderm and more weakly in the near apical oral ectoderm (see foxq2/nodal double in situ hybridizations in Fig. S1C). In the absence of [foxq2+six3] expression, nodal transcription spreads upward over the whole oral apical domain (Fig. 3C), although treatment with either foxq2 or six3 MASO alone has only minor effects (Fig. S3). This observation, at 21 h, suggests that persisting Six3 protein is responsible together with Foxq2 protein for apical nodal repression (by 21 h, six3 is no longer being transcribed in the apical domain; Fig. 3C). Transcriptional target genes of Nodal signaling such as lefty are, as expected, affected by [foxq2+six3] MASOs in exactly the same way as is nodal expression (Fig. 3C). Partial repression by Six3 probably accounts for the relatively weak nodal expression in the near-apical animal oral ectoderm. Expression of the nodal gene is the primary transcriptional response to the redox polarization that in causal terms initially generates the future oral/aboral axis (24⇓⇓⇓–28). Therefore, because much of the oral ectoderm-specific GRN is wired downstream of Nodal response genes (9⇓–11, 29, 30), Foxq2 repression of nodal hierarchically confines the whole oral ectoderm GRN to the region below the foxq2 expression boundary, that is, the “apical/near apical” boundary of Fig. 1 A and B.

Transcriptional Repression Defining the Boundary Between Ectoderm and Endoderm.

The last of the pregastrular boundaries formulated on the animal/vegetal axis to be considered here is that separating all ectoderm fates from endodermal fates. This boundary forms within the veg1 cell lineage, which gives rise to posterior endoderm and to ectodermal cells located just below the equator (31, 32). Directly or indirectly, Wnt5 signaling is involved in initial veg1 specification (33). Further separation of the veg1 regulatory state, and ultimately of embryonic fate, occurs after 24 h in the late mesenchyme blastula stage (Fig. 1B). By the end of gastrulation, veg1-derived endoderm has constituted the hindgut and part of the midgut, whereas veg1-derived ectoderm has formed the wall of the embryo surrounding the anus. The first spatial regulatory state changes denoting formation of this boundary are separation of the expression domains of lim1 and vegf3 from that of hox11/13b (Fig. 2). Up to 24 h, lim1 and vegf3 are transcribed in all veg1 cells, including the future endoderm precursors; thus the expression domains of these genes abut that of the veg2 gene foxa as can be seen in the double in situ hybridizations of Fig. S1B (foxa/lim1) and Fig. S1C (foxa/vegf3). By 24 h, hox11/13b expression is initiated in the lower rings of veg1 cells, defining those destined for endodermal fate (6). Thereafter, lim1 and vegf3 cease to be expressed in these presumptive endoderm cells, and, by 30 h, the veg1 endoderm below the newly formed endoderm/ectoderm boundary expresses hox11/13b and not lim1 or vegf3, whereas the veg1 oral ectoderm cells do not express hox11/13b but do express lim1, and those oral ectoderm cells immediately lateral to the lim1-positive cells express vegf3. Put more generally, veg1 cells expressing hox11/13b become endoderm and veg1 cells not expressing hox11/13b become ectoderm. Because the lim1 and vegf3 genes are expressed only a few hours earlier coincidentally with hox11/13b, a reasonable prediction is that, when hox11/13b is transcribed in the lower veg1 cells, this gene establishes the endoderm/ectoderm boundary by repressing ectoderm-specific genes within its domain of expression.

Fig. 3D shows that, if hox11/13b expression is blocked by MASO treatment, clearance of neither lim1 nor vegf3 expression from prospective posterior endoderm cells fails to occur. This is demonstrated by the extension of the domains of expression of these genes right to the veg2 boundary of foxa expression. Thus, hox11/13b acts as a critical domain-specific repressor on the endoderm side of the boundary separating endodermal from ectodermal cell fate in the sea urchin embryo.

Lateral Boundaries.

Fig. 1A illustrates two bilateral boundaries on the oral face of the 30-h embryo that separate regulatory states along the oral/aboral axis. These are the boundaries on each side between the animal lateral ectoderm and the medial ectoderm territories, i.e., the near-apical and central-oral ectoderm and, within the veg1 ectoderm, the boundaries on each side that separate the lateral veg1 ectoderm from the central veg1 oral ectoderm (in addition, there are the boundaries of the future stomodaeum, which are not treated here). It is to be noted that appearances can be deceiving, so to speak, in that the regulatory state map of the embryo is significantly more complex than the morphological map of the oral side of the embryo would suggest. Thus, as shown by comparing the 30-h lateral and oral views of Fig. 1A, the ciliated band actually consists of the two bilateral ectoderm regulatory states—an oral apical regulatory state and the central and lateral veg1 regulatory states—even though the regulatory gene one-cut (hnf6) is expressed early all around the future ciliated band.

The regulatory states of the aboral ectoderm domains (11, 12) differ from those of the oral ectoderm domains, although some key genes contributing to each are at first expressed in both oral and aboral ectoderm. We have already encountered two examples, six3 and lim1 (Fig. S1B). Resolution of their expression patterns to the oral side again depends on repression. Many genes of the aboral ectoderm require a positive boost from Bmp signals emitted from the oral ectoderm, as confirmed at the cis-regulatory level (12). Because of the extensive feedbacks within the aboral ectoderm GRN, Bmp MASO essentially down-regulates the whole of this GRN. As shown in Fig. 3E, use of Bmp MASO demonstrates that the aboral ectoderm GRN includes repressor(s) that function to abolish transcription of six3 and lim1 in the aboral ectoderm.

Returning to the two pairs of oral/aboral boundaries within the oral face, much has been learned about the specific repressions responsible for the boundaries between the medial and lateral ectodermal territories. Two known genes, not and gsc, encode spatial repressors that are expressed in the oral ectoderm, both activated by Nodal signaling. The not gene is expressed in the near apical, central, and veg1 oral ectoderm by 15 h, and gsc is expressed in the same domains except for the veg1 oral ectoderm. Repression by not silences multiple genes of the lateral oral ectoderm regulatory state in the medial oral ectoderm domains, leaving them to be expressed across the boundary with the lateral oral ectoderm (9, 10). For genes initially expressed across the veg1 ectoderm such as vegf3, not repression confines expression to the regions across the boundary with the veg1 lateral ectoderm (10). Repression by gsc silences the onecut gene in the near apical and central oral ectoderm, confining its expression to the ciliated band domain across the boundaries with the lateral ectoderm on the sides, the apical domain above, and the veg1 ectoderm that constitutes the oral/vegetal arm of the ciliated band (34). Aboral repressors such as irxa restrict the expression of all of the genes of the lateral oral and veg1 oral ectoderm to confined bands of cells, also obliterating their expression in the aboral ectoderm (34).

Thus, an essentially orthogonal, bilateral pattern of bounded oral ectodermal regulatory states is established by the time of gastrulation (Fig. 1A, 30 h). The observations that we summarize in this paper, taken together with those obtained earlier, show that the mechanism by which this complex pattern is established is mainly sequential, spatially confined, transcriptional repression, occurring along both axes of the embryo.