Cloning. Fully degenerate primers were designed to anneal to the conserved motifs YISLITMAIQ and FNDCF(V/L)KV in the forkhead domain of Fox sequences and KLTNEMIVTK and DTQLKIKY in the Tbox domain of brachyury genes. RT-PCR was performed by using AmpliTaq, following manufacturers recommendations (Perkin–Elmer, Boston) with 3 mM MgCl₂, and with the following cycling parameters: 94°C for 1 min, then six cycles of 94°C for 20 s, 40°C for 1 min, 72°C for 30 s, then 35 cycles of 94°C for 20 s, 45°C for 1 min, then 72°C for 30 s. Template was 50 ng of cDNA, random-primed from hatching and gastrula stage total RNAs. The resultant 171-bp Fox and 468-bp Bra PCR products were ligated into pGEM-T Easy Vector and transformed by using the supplier’s recommendations (Promega).

Phylogenetic Analyses. Each conceptually translated Asterina miniata sequence was subjected to phylogenetic analysis to confirm that the true Strongylocentrotus purpuratus ortholog had been cloned; for Krox, Gatae, and Otx sequences see refs. 1-3; for the Tbr and Bra orthologs from the closely related starfish Asterina pectinifera, see ref. 4. For the analysis of A. miniata FoxA shown in Fig. 8, full-length amino acid sequences corresponding to FoxA/Hnf3 orthologs from selected taxa were aligned in CLUSTALX and adjusted and trimmed to include only the forkhead domain, plus 18 conserved C-terminal flanking amino acids in MACCLADE 3.06. PAUP* 4.0 was used to estimate tree topology by using a neighbor-joining analysis with mean character difference as the distance measure. The confidence of each node was assessed by 100 bootstrap pseudoreplications, by using a full heuristic search based on distance measures.

Perturbation Assays. Morpholino-substituted antisense oligonucleotide (MASO) technology (Gene Tools, Philomath, OR) was used to perturb the function of gatae, krox, foxa, and tbr by inhibiting translation of the endogenous protein. The MASOs were designed to anneal to bases –22 to +3 of foxa, –12 to +13 of krox, –45 to –21 of gatae or –19 to +6 of tbr, where +1 is the adenine residue of the predicted ATG translation start. The corresponding antisense sequences are CTCCTAGAATCAAGTGGGTTGCAG (gatae); CAGGTCCTTTCATTCTGGTACTCAG (krox); CATGGGTTTCCCTCGA CTCGTTGGG (foxa), or AAGCATACTCGATACAGATCCAAAC (tbr).

Injection solutions were prepared as 600 or 800 m M a foxa MASO, 1 mM a gatae MASO, 300–800 m M a krox MASO, or 600 m M a tbr MASO in 200 mM KCl. Based on the estimated injection volume, the in vivo concentrations were » 1/125th of injection solution concentrations. MASOs of three different sequence compositions that should not have targets to A. miniata mRNAs, i.e., a random, control MASO provided by the manufacturer and two MASOs targeted to mRNAs from the sea urchin S. purpuratus were also injected at 1-mM concentrations. Injection of the control MASO did not lead to any abnormal development through to the end of the assay period, which was the end of gastrulation.

Another type of perturbation was used to disrupt Otx function, because the starfish otx, like the sea urchin otx, has multiple transcripts resulting from alternative transcription start sites. This procedure entailed injecting mRNA encoding a fusion of the Drosophila Engrailed (Eng) repressor domain with the DNA-binding domain of S. purpuratus Otx, as described in ref. 3. This mRNA is translated in vivo, binds to the endogenous cis-regulatory target site(s) and represses the expression of its downstream targets. Note that in S. purpuratus, the effects of the Otx-Eng fusion are the same as observed on combining MASOs targeted to the different otx transcripts (C.-H. Yuh and E.H.D., unpublished data).

The sequence of the DNA-binding domains of the A. miniata otx and S. purpuratus otx genes are highly similar (56 of 60 are identical amino acids) so the sea urchin otx-Eng fusion mRNA [SpEng-OtxHD(K); ref. 5] was used to suppress starfish Otx downstream targets. As a control, we used an Eng-Otx fusion mRNA which has a mutation of the 50th residue of the homeodomain from a K to Q [SpEng-OtxHD(Q); ref. 7]. This protein fails to bind to the Otx target sequence.

SpEng-OtxHD(K) and SpEng-OtxHD(Q) mRNAs were injected at 0.4 pg/pl in 200 mM KCl.

Efficiency of MASO. The efficiency of the MASO to bind to its target site and block translation in vivo was tested by following the strategy described in ref.6. Partial 5' UTR and ORF sequences flanking the predicted translation start of krox, gatae, or foxa were PCR amplified by using the following primer pairs; ACTTGTTGTAGACAGCAACTTTTCGAG, CAAGTGGACGTCGAAGACCTCAACGTCCTCGCTCTG for gatae; CCCAAGCTTCGCTCCACAGCCTGGCG, TAGCGACGTCGGTAGCTGGATGGTAGGCCGAC for foxa; and CCCAAGCTTGCACCGCATCTACTTGGCGAG, TGGGACGTCCTGGCACGGCTTGTCGTTCAC for krox. These products were digested with AatII and HindIII and ligated into these sites in the PBS-GFP vector (7) to produce an in-frame fusion of these leader sequences with sequence encoding the GFP. In vitro-synthesized mRNA corresponding to these fusions, i.e., 5'AmKrox-gfp; 5'AmGataE-gfp, or 5'AmFoxA-gfp was synthesized, purified, and quantified as described (7).

AmFoxA-gfp mRNA, 5'AmKrox-gfp mRNA, or 5'AmGataE-gfp mRNA (» 5 × 10⁵ copies) was injected into eggs, which were allowed to cleave once before either 300 m M a krox MASO, 600 m M a foxa, or 1 mM a gatae MASO was injected into one of the first two blastomeres. These embryos were allowed to develop until hatching/early gastrula, when they were assessed for GFP expression by using fluorescent microscopy. In A. miniata, the plane of first cleavage is the plane of bilateral symmetry of the embryo. This experiment provides the opportunity of using one half of the embryo as a control for the other half. We were thus able to confirm that GFP expression was absent from the half-embryo that resulted from the single blastomere receiving both the appropriate gfp-fusion and the MASO, but was still expressed in the half-embryo not injected with MASO.

Quantitative Perturbation Analysis. Epistatic interactions were determined by assessing the relative change in prevalence of the mRNA transcripts of otxb -a, otxb -b, krox, gatae, bra, foxa, tbr, and D in embryos that were perturbed as described above when compared with control embryos. For each experimental replicate, » 200 eggs were injected with 600 m M a foxa, 300 m M a krox, 1 mM a gatae MASOs or 0.4 pg/pl SpEng-OtxHD(K) mRNA. To provide controls for each of these four analyses, the same number of eggs were injected with an equivalent amount of random sequence MASO or 0.4 pg/pl SpEng-OtxHD(Q) mRNA as appropriate. The embryos were allowed to develop to 19-24 h or 28-32 h and total RNA was extracted by using the GeneElute mRNA miniprep kit (Sigma-Aldrich). One-twentieth of the RNA was put aside to later test for genomic DNA contamination, which, for all samples used in the analyses, was found to be insignificant. The remainder of the sample was reverse transcribed with superscript II reverse transcriptase primed with random hexamers (Invitrogen).

The quantitative effects of each perturbation were determined by using real-time quantitative PCR (QPCR), which was performed as described (7, 8). All values were normalized to Am18S rRNA transcript levels, which were assumed to be constant in experimental and control A. miniata. Cycle threshold differences (D C_T) were calculated as the normalized cycle threshold (C_T) of the control sample minus the C_T of the experimental treatment. Each PCR cycle was assumed to increase the product by 1.94 times, and, therefore, fold differences were calculated as 1.94D ^CT.

Embryos were assayed 19–24 h and 28–32 h after fertilization. Each data point indicated in Fig. 2 in the main text represents the average D C_T ± SE of 3, 8, 3, and 1 independent injections of Otx-Eng, a krox MASO, a gataE MASO, and a foxa MASO, respectively, and is based only on embryos assayed from the 19- to 24-h window, except where indicated. The D C_T of A. miniata orthologs of Ubiquitin transcripts in each perturbation assay was also tested and found to not vary significantly between any sets of experimental and control cDNAs. QPCR primer sequences are available on request.

Gene Expression. Whole-mount in situ hybridization (WMISH) was performed as described for AmKrox, AmGatae, and AmOtx (1–3). Dig-labeled antisense RNA probes were synthesized by using either SP6 or T7 RNA polymerases on templates corresponding to foxa (540 nucleotides of 5' UTR and ORF), bra (980 nucleotides of 5' UTR and ORF) or tbr (594 nucleotides of 5' UTR and ORF) cDNAs subcloned into pGEM-T-Easy (Promega).

The developmental time course of foxa and gatae transcript abundance was performed by using QPCR on cDNA templates prepared as described (3).

WMISH was also used to examine the spatial expression of selected transcripts in blastulae/early gastrulae that had developed from zygotes injected with certain perturbation reagents. In many cases, this method was used to confirm QPCR data. This procedure was particularly important when QPCR indicated that different regulatory connections may exist between the starfish and sea urchin GRNs. Hence, all of the connections that were present only in A. miniata were confirmed by WMISH. These connections were the requirement of gatae and otx for tbr expression, the requirement for gatae for its own expression, and foxa for its repression. We injected one blastomere of a two-cell embryo with either SpOtx-Eng(K) mRNA or AmGataE MASO so that only half of the resultant blastula/gastrula embryo would have a diminished expression of this factor. The uninjected half-embryo thus provides an internal control. Unlike Otx and Gatae, Foxa normally acts as a repressor. We therefore examined expression of foxa and gatae in embryos in which the egg had been injected with a foxa MASO, because a change in spatial expression may reveal the nature of foxa repression. We also confirmed that the spatial domains of bra, krox, and tbr were unaffected in blastulae resulting from AmFoxA MASO-injected zygotes.

1. Hinman, V. F. & Davidson, E. H. (2003) Gene Expr. Patterns 3, 419–422.

2. Hinman, V. F. & Davidson, E. H. (2003) Gene Expr. Patterns 3, 423–426.

3. Hinman, V. F., Nguyen, A. & Davidson, E. H. (2003) Mech. Dev. 120, 1165–1176.

4. Shoguchi, E., Satoh, N. & Maruyama, Y. K. (2000) Dev. Growth Differ. 42, 61–68.

5. Li, X. T., Wikramanayake, A. H. & Klein, W. H. (1999) Dev. Biol. 212, 425–439.

6. Rast, J. P., Cameron, R. A., Poustka, A. J. & Davidson, E. H. (2002) Dev. Biol. 246, 191–208.

7. Oliveri, P., Carrick, D. M. & Davidson, E. H. (2002) Dev. Biol. 246, 209–228.

8. Rast, J. P., Amore, G., Calestani, C., Livi, C. B., Ransick, A. & Davidson, E. H. (2000) Dev. Biol. 228, 270–286.