SpMyb functions as an intramodular repressor to regulate spatial expression of Cyllla in sea urchin embryos

The CyIIIa gene of Strongylocentrotus purpuratus encodes a cytoskeletal actin that is specifically transcribed in the aboral ectoderm of the embryo and larva, under the control of a 2.3 kb ds-regulatory domain. Studies of this cis-regulatory system have served as a paradigm for the analysis of information processing systems that regulate transcription during embryonic development (see Kirchhamer et al., 1996, for review). In vitro analyses have identified over twenty protein target sites within the CyIIIa regulatory domain, at which specific interactions with at least nine different transcription factors occur (Calzone et al., 1988; Thézé et al., 1990). All of these factors have been isolated by affinity chromatography (Calzone et al., 1991; Coffman et al., 1992; Char et al., 1993; Zeller et al., 1995a; Coffman et al., 1996), and eight of them have been cloned (see Fig. 1A). Concomitantly, in vivo studies utilizing gene transfer technology have identified regulatory functions for each target site (Franks et al., 1990; Hough-Evans et al., 1990; Kirchhamer and Davidson, 1996; Coffman et al., 1996).

The CyIIIa cis-regulatory domain consists of two functionally distinct modules, the locations of which are shown in Fig. 1A (Kirchhamer and Davidson, 1996; Kirchhamer et al., 1996). The proximal module includes both positively and negatively acting spatial control elements which together suffice to initiate expression exclusively in the aboral ectoderm early in embryogenesis. The middle module functions dominantly later in embryogenesis. It is required both for the sharp increase in CyIIIa transcription that occurs after blastula stage, and for spatially localizing that transcription to the aboral ectoderm (Kirchhamer and Davidson, 1996). It was recently demonstrated that the positive regulatory function of the middle module is mediated by the SpRunt-1 target site (Coffman et al., 1996). Two negative spatial regulatory functions have also been identified within the middle module. Interactions with the zinc finger repressor SpZ12-1 (see Fig. 1A) are required to prevent transcription in the skeletogenic mesenchyme (Wang et al., 1995). A second middle module repressor was inferred from in vivo competition (Hough-Evans et al., 1990), and from the results of deletion of a region of the middle module known as the P7II target site (Fig. 1A; Kirchhamer and Davidson, 1996). Interactions at this site were found to be required to prevent ectopic transcription in both the oral ectoderm and skeletogenic mesenchyme. The P7II target site had been mapped to a 63 bp sequence within the middle module which is shown in Fig. 1B. The purpose of the present investigation was to determine what protein target sites within the sequence shown in Fig. 1B mediate its negative regulatory function, and to identify the responsible repressor protein.

Previous analysis of the P7II region revealed that it contains binding sites for SpGCF1 (Fig. 1A,B), a transcription factor that binds multiple target sites in the regulatory domains of various genes (Zeller et al., 1995a). The ability of SpGCF1 to homomultimerize and loop DNA in vitro led to the hypothesis that its function may be to facilitate direct communication between proteins bound at distant target sites (Zeller et al., 1995b). The facts that target sites for SpGCF1 are widespread, occurring in both the proximal and the middle modules, and that where analyzed, these target sites function positively in transcriptional regulation (Franks et al., 1990; Kirchhamer et al., 1996), suggested that SpGCF1 is not the factor responsible for the negative spatial function of the P7II region. However, because of the high abundance of SpGCF1 in nuclear extracts, it was initially difficult to detect any other proteins that bind in the same region. To overcome this difficulty we made use of automated sequential affinity chromatography (Coffman et al., 1992) to analyze P7II-specific binding proteins that remain in nuclear extract following passage over columns bearing specific SpGCF1 sites. We thus discovered that a sea urchin myb-domain transcription factor, SpMyb, specifically binds the P7II target site. Gene transfer experiments demonstrate that the ectopic expression mediated by the SpRunt-1 interaction within the middle module of the CyIIIa gene (Coffman et al., 1996; Kirchhamer et al., 1996) is repressed by the addition of a single synthetic Myb site. SpMyb is thus a key negative spatial regulator of the middle module.

Synthetic oligonucleotides used in the constructs shown in Fig. 1C are listed below. The following double-stranded oligonucleotides (corresponding to the boxed sequence in Fig. 1B) were used both as probes in DNA-binding assays and as artificial target sites in reporter constructs (Fig. 1C). Myb target sites are in larger boxes and in bold face, GCF1 sites are in smaller boxes, nucleotide base substitutions that alter the wild-type sequence are underlined, and sequences containing SacII sites for the purpose of cloning are in lower case.

Oligonucleotide pair MG:

Note that four of the five base substitutions in oligonucleotide pair G fall outside of the canonical myb target site, within sequences that in the wild-type are partial myb recognition sequences (4/6: TAAC), and also that these changes introduce an additional GCF1 site. For the purpose of cloning, the two individual oligonucleotides of each pair were annealed, concatamerized using T4 DNA ligase, and digested with SacII.

The 9.4 kb CylIIaCAT construct (line 1 of Fig. 1C) described by Kirchhamer and Davidson (1996) was used as control construct. Constructs MG.CAT, M.CAT and G.CAT (lines 3, 4 and 5 of Fig. 1C) were generated by opening plasmid DP7II.CAT (line 2 of Fig. 1C; Kirchhamer and Davidson, 1996) at the unique SacII site and inserting the oligonucleotide pairs MG, M or G, respectively (see above). The resulting constructs were sequenced to verify that they contained only a single oligonucleotide insert, and that the insert was in the proper orientation with respect to the wild-type sequence.

All plasmids carry the remaining complement of regulatory sites of the CyIIIa 5’ regulatory region, transcription initiation site, 5′ leader sequence of the primary transcript and a CyIIIa-CAT fusion gene. The plasmids were linearized for microinjection at unique SphI or KpnI sites.

Gametes of Strongylocentrotus purpuratus were obtained, microinjections carried out and embryos cultured as described previously (McMahon et al., 1985; Kirchhamer and Davidson, 1996). Embryos were collected at the appropriate stage and processed for whole-mount in situ hybridization (Ransick et al., 1993; Kirchhamer and Davidson, 1996). Embryos were analyzed and images stored with an imaging system using ProgRes software and a Zeiss Axioskop microscope.

Large-scale embryo cultures, preparation of nuclear extracts, and automated sequential affinity chromatography (ASAC) were carried out as previously described (Calzone et al., 1991; Coffman et al., 1992). For protein microsequencing, peak fractions from four separate ASAC runs of an affinity column bearing oligonucleotide pair M (see above) were pooled, dialyzed against Buffer 0.1 (20 mM Hepes-KOH, pH 7.9; 0.1 M KCl, 0.1 mM EDTA, 1 mM dithiothreitol, 0.1% Nonidet P-40, 0.1 mM PMSF, 20% glycerol), brought to 20 ng/q l with poly(dIdC/dIdC), and reloaded onto a tandem series of four columns bearing the M oligonucleotide pair. The columns were loaded and eluted using a Gilson ASPEC programmed with our standard ASAC Software (Coffman et al., 1992), and the fractions analyzed by both electrophoretic mobility shift and SDS-PAGE. This second round of affinity purification in the presence of nonspecific competitor DNA resulted in a substantial enrichment of the myb-specific DNA-binding activity. Peak fractions were again pooled, dialyzed against Buffer 0.1 (without glycerol), and concentrated by ultrafiltration using Centricon 10 mini-concentrators (Amicon). The pooled, concentrated M oligonucleotide column eluate was brought to 1% trifluoroacetic acid, subjected to reverse phase HPLC (SMART, Pharmacia) on a C4 column (ABI), and eluted with a linear acetonitrile gradient. Individual peaks were collected and analyzed by electrophoretic mobility shift and by SDS-PAGE. Myb-specific DNA-binding activity was found in a single peak which contained two species of protein, an abundant species of approx. 30 kDa and a minor species of approx. 80 kDa. The approx. 30 kDa species corresponds in size to the predominant myb-specific DNA binding protein identified by 2-dimensional EMSA (J. A. Coffman, unpublished result), and probably represents a proteolytic fragment of the full-length protein (see Results). This fraction was subjected to digestion with a lysylendopeptidase (Achromobacter Protease 1; Wako). The resultant peptides were resolved by reverse phase HPLC (Pharmacia SMART equipped with a C18 column) and sequenced on a Porton automated sequencer.

Electrophoretic mobility shift assays (EMSA) were performed essentially as described by Calzone et al. (1988); Coffman et al. (1996). For detection of myb DNA-binding activity in either affinity column eluate or in vitro translation reactions, 20 ng/μl poly (dIdC/dIdC) was used as nonspecific competitor in each 10 μl binding reaction, which included 1 μl of protein extract. For detection of immunoprecipitated, renatured myb DNA-binding activity, 10 ng/μl poly (dIdC/dIdC) was used in each 15 μl binding reaction, which included 10 μl of renatured protein eluted from the antibody complex (see below). EMSAs were visualized using a phosphorimager and ImageQuant software (Molecular Dynamics).

Immunoprecipitation of affinity-enriched myb DNA-binding activity was performed as follows: 100 pl of P7II affinity column eluate was brought to 1% SDS and heated to 90°C to denature the proteins, then brought to 1 ml in immunoprecipitation buffer (60 mM Na2HPO4, 40 mM NaH₂PO₄, 1% sodium deoxycholate, 0.1% SDS, 2 mM EDTA, 150 mM NaCl, 1 mM PMSF). 250 μl aliquots of this preparation were incubated overnight at 4°C with 25 μl protein A sepharose that had been pre-coated with either anti-BP-2 antibodies (Boyle et al., 1986) or antibodies from nonimmune serum from a different rabbit. The resulting immunoprecipitates were collected by centrifugation, washed three times in 1 ml each of PBS, and eluted with 100 μl denaturation buffer (20 mM Hepes-KOH, pH 7.9; 0.1 mM EDTA, 1 mM dithiothreitol, 6 M guanidinium-HCl, 1 mM MgCl2). The eluted proteins were renatured over 0.5 ml Biogel P6 (Biorad) columns that had been pre-equilibrated in renaturation buffer (10 mM Hepes-KOH, pH 7.9; 0.1 M KCl, 1 mM dithiothreitol, 0.1% Nonidet P-40, 10% glycerol) and subjected to EMSA as described above.

A highly conserved region of the myb DNA binding domain was amplified by polymerase chain reaction, using the degenerate primers designed by Stober-Grasser et al. (1992). Plasmid prepared from a blastula-stage λZap cDNA library was used as template. The amplified 168 bp fragment was excised, electroeluted, cloned into pGEM-T (Promega), and sequenced. This fragment was then used to screen a blastula-stage λZap cDNA library by standard procedures. A first round of screening yielded two clones, 7.1.1 and 15.1.1, which contain inserts of approx. 1.2 and approx. 1.5 kb respectively. The latter insert was sequenced and found to contain an uninterrupted open reading frame. An approx. 500 bp XbaI-EcoRI fragment from the 3’ end of the 15.1.1 insert was therefore used to rescreen the blastulastage λZap cDNA library, yielding clones 9.1.1, 10.1.1, and 12.1.1. Clone 10.1.1 contains the longest insert (approx. 2.2 kb) and was sequenced. The SpMyb sequence presented here was compiled from the combined double stranded sequences of clones 15.1.1 and 10.1.1.

DNA sequencing and sequence analysis was carried out as previously described (Coffman et al., 1996), using an ABI 373A automated sequencer and IBI Pustell sequence analysis software. Nested deletions of clones 10.1.1 and 15.1.1 were constructed using a Pharmacia kit.

In vitro transcription and translation were performed using Ambion T3 mMessage Machine and reticulocyte lysate kits, respectively. For in vitro transcription, clone 15.1.1 was linearized with BsmI, which cuts at nucleotide position 711 of the sequence depicted in Fig. 5 (126 bp downstream from the myb domain). In vitro translation reactions were carried out using the manufacturers protocol, except that the reactions were supplemented with 0.1 mM L-methionine. The use of nonradioactive methionine allowed for unambiguous detection of the translation products by EMSA (described above).

Genomic DNA and RNA blot hybridizations were carried out as described previously (Coffman et al., 1996), using as probe either an XbaI-EcoRI fragment of SpMyb corresponding to a C-terminal portion of the coding sequence, or the cloned PCR fragment corresponding to repeat 3 of the DNA-binding domain. Probes were labeled with [³²P]dCTP (NEN), using the Oligolabeling kit from Pharmacia. Washes were carried out in either 0.1x SSC, 1% SDS (high stringency) or in 1× SSC, 0.2% SDS (lower stringency) at 65°C.

Fig. 1B shows the 63 bp sequence, referred to here as the P7II region, which was previously shown to be required for negative spatial regulation of CyIIIa (Hough-Evans et al., 1990; Kirchhamer and Davidson, 1996). Earlier studies revealed that this segment of DNA interacts with one or more proteins (Calzone et al., 1988; Thézé et al., 1990). Our initial experiments were therefore designed to determine exactly what target sites lie within the P7II region. When the entire P7II region is used as a probe, the protein most commonly detected in crude nuclear extract by electrophoretic mobility shift assays (EMSA) is SpGCF1, a relatively abundant protein that functions positively in regulating CyIIIa (Franks et al., 1990; Zeller et al., 1995a; Kirchhamer and Davidson, 1996). Because the P7II region clearly functions negatively in regulating CyIIIa, we suspected that it might contain a target site for another protein, which would be obscured by the high abundance of SpGCF1 in nuclear extracts. We found previously that a tandem series of columns bearing oligonucleotides containing a highly specific SpGCF1 target site (P8II from Thézé et al., 1990) can be used to remove this protein efficiently from nuclear extracts, allowing downstream purification of rarer proteins in the same sequential affinity chromatography run (Coffman et al., 1992). Nuclear extract was thus passed over two columns bearing a SpGCF1 target site, and then over a column bearing an oligonucleotide fragment of the P7II region (boxed sequence in Fig. 1B). In addition to the GCF1 site (shown in blue in Fig. 1B), the P7II oligonucleotide contains a sequence (shown in red Fig. 1B) that matches perfectly the consensus binding site for the myb family of DNA-binding proteins, viz PyAAC^G/TG (in reverse orientation in Fig. 1B; see Lipsick, 1996; Graf, 1992; and Luscher and Eisenman, 1990 for reviews).

In order to distinguish between SpGCF1 and putative myb DNA-binding activities eluted from the P7II affinity column, we constructed synthetic 22bp oligonucleotides that dissociated these sites. Three different double-stranded oligonucleotides were synthesized: (1) a fragment of the wild-type sequence with both the myb and GCF1 sites (oligo MG) that served as a control; (2) a mutated sequence that abolishes the myb site but adds an additional GCF1 site (oligo G); and (3) a mutated sequence that abolishes the GCF1 site but retains the myb site (oligo M; see Materials and Methods for oligonucleotide sequences). Each of these oligonucleotides was used as a probe in EMSA experiments in order to analyze the DNA-binding specificity of proteins eluted from the P7II affinity column. As is shown in Fig. 2, both oligonucleotides containing the myb target site (i.e., oligonucleotides MG and M) specifically bound a protein eluted from the wild-type P7II affinity column (Fig. 2A). In contrast, when the same oligonucleotides were used in EMSA to probe total nuclear extract, only SpGCF1 was detected, by probes G and MG (Fig. 2B). Probe M detected nothing under these conditions in total nuclear extract from blastula-stage embryos, indicating that the active myb-site-specific factor is relatively rare in these extracts. Nonetheless, an estimated 100-fold enrichment of this factor achieved by a single round of affinity chromatography allowed its detection. These results demonstrate that while the wild-type oligonucleotide MG interacts with both SpGCF1 and a myb site-specific factor, oligonucleotide M interacts exclusively with the myb-site-specific factor, and oligonucleotide G interacts exclusively with SpGCF1.

Gene transfer experiments were performed to determine the function of each of the target sites contained within the P7II region. Toward this end, each of the synthetic oligonucleotides used for the DNA/protein binding experiments described above was separately cloned into the deleted region of the ΔP7II.CAT construct (Fig. A 1C, lines 3–5). The resulting constructs were microinjected into eggs, which were then allowed to develop to gastrula stage and analyzed by whole-mount in situ hybridization (WMISH). The results of this experiment are summarized quantitatively in Table 1, and representative examples are shown in Fig. 3. Remarkably, each of the constructs that contained a ‘wildtype’ myb site, i.e., MG.CAT and M.CAT gave rise to spatially normal patterns of expression (Fig. 3A,B). This result is quantitated in lines 3 and 5 of Table 1. Note that the small fractions of embryos displaying ectopic expression of these constructs are comparable to the fractions obtained with the wild-type CyIIIa.CAT construct (Table 1, line 1; see also Kirchhamer and Davidson, 1996). In contrast, G.CAT, the construct containing the mutated myb target site, gave rise to significant ectopic expression in both the oral ectoderm and the skeletogenic mesenchyme, as illustrated in Fig. 3C1, C2. This result is shown in line 4 of Table 1. It is important to note that the amount of ectopic expression obtained with this construct is similar to that observed when Δ P7II.CAT is injected (Table 1, line 2; Kirchhamer and Davidson, 1996). Since G.CAT includes a perfectly good GCF1 site, we conclude that the latter site plays no role in the negative regulatory function of the P7II region. In contrast, M. CAT, the construct lacking GCF1 sites in the oligonucleotide insert, exhibited somewhat lower levels of expression (Table 1, line 3), a result that is consistent with the earlier evidence cited above that SpGCF1 facilitates transcriptional activation. These results demonstrate that, within the context of the rest of the CyIIIa regulatory system, the myb target site contained within the P7II region is necessary and sufficient to prevent ectopic transcription in the oral ectoderm and skeletogenic mesenchyme. Furthermore, the five point mutations that eliminate interactions with the myb site binding protein in vitro produce quantitatively the same effect on spatial expression in vivo as does the entire 63 bp P7II deletion studied earlier by Kirchhamer and Davidson (1996).

To determine whether the affinity purified myb-binding protein in fact contains a myb related DNA-binding domain, we first made use of a polyclonal antibody (anti-BP-2) generated against the vertebrate myb DNA-binding domain (Boyle et al., 1986). Anti-BP-2 has been shown to immunoprecipitate myb-related proteins from a wide variety of metazoan species. The EMSA depicted in Fig. 4 shows that the protein eluted from the P7II affinity column is indeed immunoprecipitated with anti-BP-2, but not with non-immune antibodies.

Affinity purified myb-binding proteins were further purified to near homogeneity by reverse-phase HPLC (data not shown), and subjected to amino acid sequence analysis. Two amino acid sequences were recovered that fall within the highly conserved first and third repeats of the myb DNA-binding domain (see Fig. 5, highlighted sequence). This result directly demonstrates the existence in the nuclei of sea urchin embryos of a myb domain protein that can be purified by virtue of interaction with its CyIIIa target site.

To obtain cDNA clones encoding this protein, degenerate primers (Stober-Grasser et al., 1992) were used in a polymerase chain reaction to amplify a blastula-stage cDNA specifying a highly conserved fragment of the myb DNA-binding domain. The amplified fragment was cloned, sequenced, and used as a probe to screen a blastula-stage cDNA library. With this probe we isolated clones for SpMyb (see Materials and Methods), the combined sequences of which are shown in Fig. 5. Note that the open reading frame begins at the 5 end of the sequence. The first methionine residue, which aligns approximately with the N-termini of vertebrate myb proteins (see Fig. 6), occurs at amino acid position 4 in the open reading frame. However, we cannot be certain that this methionine represents the actual N terminus of SpMyb. The open reading frame depicted in Fig. 5 encodes an approximately 77 kDa protein, which is in the size range of known myb proteins (Boyle et al., 1986). The size also corresponds to the minor approx. 80 kDa species found in the purified preparation of SpMyb from which the amino acid sequence was obtained (see Materials and Methods).

Fig. 6 depicts an alignment of SpMyb with various other members of the myb family from vertebrates and Drosophila. Consistent with their common deuterostome relationship, the highest degree of similarity occurs between SpMyb and vertebrate A-Myb (39% identity), while slightly less correspondence occurs between SpMyb and vertebrate B-Myb (36% identity), and c-Myb (33% identity). The simi-larity between SpMyb and the Drosophila Myb is lowest (31% identity); moreover, DmMyb contains a long stretch of amino acids at the N terminus that are not found in any of the other aligned sequences, and is lacking certain regions that are found in both SpMyb and in all the vertebrate homologues. As can be seen in Fig. 6, the myb DNA-binding domain is the highest region of conservation, containing the three tandem repeats that form the multiple helix-turn-helix motifs of this domain. More C-terminal regions are also conserved. Some of these contain potential sites of phosphorylation by proline-directed serine/threonine kinases, including S528 of murine c-Myb (S533 in chicken c-Myb). This site (boxed in Fig. 5) is S549 of SpMyb. Phosphorylation of this serine has been shown to contribute to negative functions of vertebrate c-Myb, possibly by modulating specific protein-protein interactions (Aziz et al., 1995; Miglarese et al., 1996).

In order to demonstrate that the cloned SpMyb is able to bind specifically to the myb target site from the CyIIIa regulatory domain, a fragment of SpMyb corresponding to the DNA-binding domain was transcribed and translated in vitro, and subjected to EMSA with the M oligonucleotide as probe. As can be seen in Fig. 7, the SpMyb protein fragment specifically interacts with the functional myb target site (oligonucleotide M), but not with the mutated target site (oligonucleotide G).

The results illustrated in Fig. 8 indicate that SpMyb is the only myb-family member in sea urchin embryos. Fig. 8A shows a genome blot of EcoRI-digested DNA from three different individuals that was hybridized with a fragment encoding the C-terminal region of SpMyb. The single bands in lanes 1 and 2 indicate that these individuals are each homozygous for two different restriction fragment variants. The two bands in lane 3 indicate that this individual is either homozygous for a third variant containing an internal EcoRI site, or is heterozygous, carrying two different alleles. The degree of polymorphism apparent on this blot is consistent with that found in natural populations (Britten et al., 1978). When the genome blot was stripped and rehybridized at lower stringency with a probe representing the myb DNA-binding domain, essentially the same result was obtained (data not shown), suggesting that there is only one canonical myb-domain protein encoded in the genome of 5. purpuratus.

To determine the size of the SpMyb transcript, poly(A)⁺ RNA isolated from blastula-stage embryos was transferred to nylon and hybridized with the same fragment as used to probe the genome blot. The results are illustrated in Fig. 8B, which shows that SpMyb is encoded by a single transcript at this stage, of approximately 6 kb. No additional transcripts were detected when the conserved region of the myb domain was used to probe this same blot (data not shown). We therefore conclude that, within the limits of detection employed here, SpMyb encodes the only myb-domain protein expressed in blastula-stage embryos of 5. purpuratus.

Whole mount in situ hybridization was employed to visualize the distribution of SpMyh transcripts in prism-stage embryos. Examples of the staining pattern obtained with an antisense, digoxigenin-labeled SpMyb probe are shown in Fig. 9B,D. As a control, embryos were stained with Endo16, which is specifically localized to the midgut and hindgut at this stage (Fig. 9A,C; Ransick et al., 1993). In contrast, SpMyb transcripts appear to be largely localized to the oral ectoderm, mesenchyme and endoderm at this stage (Fig. 9B,D). SpMyb is therefore present in regions where it is required to repress CyIIIa transcription (oral ectoderm and mesenchyme), and in other regions as well (endoderm). The temporal expression pattern, as well as a more detailed analysis of the spatial pattern of both SpMyh transcripts and protein will be the subject of a separate report.

We report here the cloning and initial characterization of SpMyb, a sea urchin member of the myb family of transcription factors. SpMyb was identified as a nuclear protein in sea urchin embryos that specifically interacts with a functional target site in the ds-regulatory domain of CyIIIa. This target site, which lies within the region referred to as P7II in earlier work (Hough-Evans et al., 1990; Kirchhamer and Davidson, 1996), is required to prevent ectopic expression of CyIIIa in the oral ectoderm and skeletogenic mesenchyme. Six lines of evidence indicate that SpMyb is the protein that mediates the negative regulatory function of this target site in vivo: (i) when alternative is now improbable, since no other proteins were found to interact specifically with the P7II region, either by EMSA or by the highly sensitive technique of affinity chromatography combined with EMSA.

Several regions of the sequence in SpMyb are also conserved in various other myb proteins. As expected, the highest conservation occurs within the myb DNA-binding domain, the primary structure of which is highly conserved in organisms as distant as Dictyostelium discoideum (Stober-Grâsser et al., 1992; Lipsick, 1996). A C-terminal region previously shown to be conserved in both Drosophila and vertebrate myb proteins (reviewed by Lipsick, 1996) is also found in SpMyb. In addition, SpMyb contains blocks of residues that are also found in the ‘negative regulatory domain’ of vertebrate c-Myb (Dubendorff et al., 1992). In vertebrates, c-Myb serves as an cloned into the region of CyIIIa.CAT constructs from which the P7II region had been deleted, short oligonucleotides bearing the CyIIIa myb target site reconstitute the spatial regulatory function of that region in vivo, whereas oligonucleotides containing point mutations that alter bases known to be critical for myb target-site recognition do not (Table 1 and Fig. 3); (ii) a protein isolated from blastula-stage nuclear extracts interacts specifically with oligonucleotides containing the myb target site, as is shown by in vitro competition experiments (Fig. 2); (iii) this protein is immunoprecipitated by an antibody specific to the myb domain (Fig. 4); (iv) amino acid sequences obtained from the affinity-purified protein are identical to sequences in the conserved DNA binding domain of SpMyb; (v) DNA and RNA blot hybridization experiments employing different parts of cloned SpMyb as probe indicate that there is only one canonical myb domain protein encoded in the sea urchin genome, and that only a single species of SpMyb transcript is expressed in the blastula-stage embryo (Fig. 8); and (vi) SpMyb is expressed in the territories where the P7II target site is required to prevent ectopic expression of CyIIIa (Fig. 9). While it remains formally possible that a protein other than SpMyb is actually responsible for the negative spatial regulatory function of the P7II target site, this activator of some genes and as a repressor for others (see, e.g. Reddy et al., 1994; Vandenbark et al., 1996; reviewed by Lipsick, 1996). Vertebrate A-Myb has been shown to be an activator where analyzed (see, e.g. Golay et al., 1994; Takahashi et al., 1995). B-Myb has been shown to function both as a repressor (Foos et al., 1992), and as a cell-type-specific activator (Tashiro et al., 1995). We have shown that in the sea urchin embryo, SpMyb functions as a repressor of CyIIIa transcription. In the following we discuss evidence that the specific target of that repression is the SpRunt-1 activator.

Previous analyses demonstrated that the middle module of the CyIIIa regulatory domain becomes the dominant regulatory module during differentiation of the aboral ectoderm after the blastula stage of embryonic development, and that the activation function of the middle module is mediated by the SpRunt-1 target site (Kirchhamer and Davidson, 1996; Coffman et al., 1996). In addition, it was shown that constructs lacking the entire middle module and retaining only the proximal module produce normal patterns of expression even though they lack the P7II regions, whereas constructs lacking the P7II region of the middle module but in which the remainder of the middle module is present are expressed ectopically (Kirchhamer and Davidson, 1996). These results suggested that the P7II region mediates repression by interfering with the transactivation function of SpRunt-1 in the oral ectoderm and skeletogenic mesenchyme. Such regulation could either be direct, by means of protein-protein interactions between SpMyb and SpRunt-1, or indirect, occurring by means of interactions of SpRunt-1 and SpMyb with a common intermediate protein or proteins. Intramodular or local repression, referred to as ‘quenching’ by Gray and Levine (1996), endows the developmental cis-regulatory module with the independent capability of processing spatial regulatory information in given regions or cell types of the embryo (for discussion, see Arnone and Davidson, 1997; Kirchhamer et al., 1996; Gray and Levine, 1996).

It is likely that SpRunt-1 is active not only in the aboral ectoderm, wherein it functions to activate CyIIIa, but also in the oral ectoderm and skeletogenic mesenchyme, where SpMyb appears to be required to quench SpRunt-1-mediated activation. Late in development SpMyb is itself expressed strongly in the oral ectoderm, mesenchyme and endoderm. The repressive functions of the middle module are evident only after gastrulation at least in mesenchyme cells (Wang et al., 1995; Kirchhamer et al., 1996), and the localized expression of SpMyb occurs at this time. Thus its function as a repressor appears to be transcriptionally regulated in the postgastrular embryo, the terminal period of differentiation for many embryonic cell types.

The putative functional interaction between SpMyb and SpRunt-1 is interesting in light of the known interactions between vertebrate homologues of these two transcription factors. Vertebrate myb-domain and runt-domain factors have been shown to interact synergistically in the activation of the T-cell receptor § and y enhancers (Hernandez-Munain and Krangel, 1994; Hsiang et al., 1995) and in the activation of a murine leukemia retroviral enhancer (Zaiman and Lenz, 1996). In the case of the former, the interaction does not involve cooperative DNA binding, suggesting that it may require intermediary proteins (Hernandez-Munain and Krangel, 1995). In vertebrate systems, both runt-domain and myb-domain transcription factors are also known to interact with Ets-domain transcription factors in the regulation of various genes (e.g., Reddy et al., 1994; Wotton et al., 1994; Zaiman and Lenz, 1996; Dudek et al., 1992). We note that a perfect target site for an Ets-domain factor occurs 20 bp upstream of the myb target site in the CyIIIa middle regulatory module (-1252 to -1257 in the sequence published by Thézé et al., 1990), and a close match (5/6) for the Ets target site occurs immediately adjacent to the SpRunt-1 target site. There is no evidence at present that either of these sites bind Ets-domain proteins or function in regulating CyIIIa transcription. However, an Ets factor is expressed in sea urchin embryos, and its mRNA is prevalent in early embryos (Chen et al., 1988).

The interaction between SpMyb and SpRunt-1, and perhaps an Ets factor as well, are specified by the target sites encoded in the cis-regulatory DNA of the CyIIIa middle module, which can thus be thought of as a ‘hard-wired’ genomic regulatory subsystem (Arnone and Davidson, 1997). The comparative evidence indicates the remarkable conservation of the main working component of this subsystem, a sequence element that includes target sites for runt, myb, and Ets transcription factors. Here this element functions as a spatial on-or-off switch, a role that is likely to be useful in many different developmental regulatory contexts, and in many different genes.

The authors are indebted to Dr Joseph Lipsick (Stanford University Medical Center) for the kind gift of BP-2 antiserum. We thank Nicolette Halloran and Jina Yun for technical assistance, and Drs Ellen Rothenberg, Maria Arnone, and Leonard Bogarad for critical reviews of the manuscript. This work was supported by the Stowers Institute for Medical Research, by a grant from the NICHHD (HD-05753), and from NSF subcontract (STC DIR 8809719).