Ectoderm nuclei from sea urchin embryos contain a Spec–DNA binding protein similar to the vertebrate transcription factor USF

In sea urchins, cell lineages arising from the animal-most two-thirds of the unfertilized egg contribute solely to the embryonic ectoderm. By the larval pluteus stage (3 days after fertilization in Strongylocentrotus purpuratus), several ectodermal cell types are distinguishable. These include aboral ectoderm, a continuous sheet of approximately 470 cells covering the larval surface, and a variety of oral ectoderm cell types such as those constituting the neural ectoderm, ciliary band and stomadeum (Davidson, 1986). Because of its simplicity, the aboral ectoderm has been the focus of several investigations in recent years (Lynn et al. 1983; Hurley et al. 1989; Cameron et al. 1989; Nisson et al. 1989). It derives from 11 founder cells, one of which arises at the third, four at the fifth, and six at the sixth cleavage division (Cameron et al. 1987; 1989). Each of these founder cells contributes a distinct patch of cells on the larval surface; hence the aboral ectoderm is a composite of several invariant lineages (Cameron et al. 1987; 1989). However, in spite of this invariance, blastomeres destined to become aboral ectoderm in the normal embryo can be experimentally manipulated to form other cell types, including those associated with oral ectoderm and more vegetal cell types such as gut, muscle, pigment and skeletal tissue. (Horstadius, 1973; Davidson, 1989; Livingston and Wilt, 1989) The remarkable plasticity of aboral ectoderm precursors suggests that the mechanisms giving rise to this cell type in the early embryo cannot simply be cytoplasmic segregation of maternal factors. A recent review by Davidson (1989) addresses these issues in more detail.

Approaching the problem of aboral ectoderm specification requires the appropriate molecular markers, and many genes that are activated exclusively in this cell type have now been isolated and characterized (Cox et al. 1986; Hardin et al. 1988; Yang et al. 1989). We have been using the Spec genes as markers for aboral ectoderm differentiation. This small gene family, encoding intracellular calcium-binding proteins, is activated at the late cleavage-early blastula stage (Tomlinson and Klein, 1990). Spec mRNA accumulation is highly specific to aboral ectoderm cell lineages (Lynn et al. 1983; Carpenter et al. 1984; Hardin et al. 1988; Tomlinson and Klein, 1990). Our studies have been directed towards identifying the DNA sequence elements responsible for transcriptionally activating the Spec genes in aboral ectoderm cells and towards isolating the protein factors that recognize these elements.

Elucidating the ontogeny of such proteins and the mechanisms that activate them in the egg and early embryo would greatly increase our knowledge of both how the aboral ectoderm cell type originates and the basis of its plasticity.

Recently, we have focused our attention on the structure and expression of three Spec genes: Sped, Spec2a, and Spec2c (Hardin et al. 1988; Gan et al. 1990b). By aligning these genes with a common upstream 600 bp repetitive DNA sequence element, termed RSR, we showed that a conserved DNA block of approximately 800 bp extends from the 3′ end of the first exon to the 5′ end of the RSR element. In Spec2a, the conserved sequence block is a continuous stretch of DNA, but in Sped and Spec2c, 2.4 to 2.9 kb of inserted DNA interrupts the conserved sequence block, thus changing the relative placement of the RSR element and other 5′ flanking DNA. Deletion of the 5′ half but not the 3′ half of the RSR element causes a significant decrease in the chloramphenicol acetyl transferase (CAT) activity induced by Spec – CAT reporter gene fusions injected into Lytechinus pictus eggs. Furthermore, this element has several properties suggesting that it is an enhancer-like element important for Spec gene expression (Gan et al. 1990b).

Here we continue our analysis of the Spec gene conserved sequence block. We demonstraté the presence of a DNA-binding protein in sea urchin embryos, termed SpF1, that by several criteria is similar to the human transcription protein called upstream stimulatory factor (USF). A single SpF1-binding site is located within the conserved sequence block of each Spec gene, but because of the unusual organization of the sequences flanking these genes, the SpFl-binding site is positioned very differently in each gene with respect to the site of transcriptional initiation. SpFl-binding activity was detected only in nuclear extracts from embryonic ectoderm cells; no activity was observed in endoderm or mesoderm cells despite the fact that these latter extracts contained other DNA-protein binding activities. Unexpectedly, mutations in any of the three Spec genes that abolished SpF1–DNA binding in vitro had no effect on expression of Spec-CAT reporter genes when these genes were microinjected into L. pictus eggs. Although SpF1 is clearly an embryonic ectodermspecific DNA-binding protein, its ability to function as a transcription factor remains to be determined.

S. purpuratus and L. pictus were obtained from Marinus (Venice, CA). Gametes were collected from gravid adults by intracoelomic injection of 0.5 M KC1. Eggs were washed in artificial sea water and fertilized with dilute suspensions of sperm. Embryos were cultured at 15°C with constant stirring at concentrations of 6 – 8000 embryos ml^-1.

Ectodermal cells were separated from endodermal/ mesodermal cells according to the procedures of McClay (1986).

The following oligonucleotides were synthesized in an Applied Biosystems 380B DNA synthesizer:

CTB corresponds to the sequence recognized by SpF1. CT1+2 and CT3+4 border and overlap six nucleotides on each side of CTB. CTC-1 and CTC-2 were used in the in vitro mutagenesis experiments for Sped. CTD-1 and CTD-2 were used in thé in vitro mutagenesis experiments for Spec2a. The underlined portions of CTC-1 and CTD-1 indicate the changed nucleotides, each of which contains a BglII sequence (AGATCT).

Clone 191H-S is a 191 bp HinPI – SalI fragment from – 58 to + 133 of the Sped gene. Clone 298N-S is a 298bp Nsil-Sah fragment from – 164 to +133 of Sped and clone 255R-N is a 255bp EcoRI-NszI fragment from – 420 to – 165 of Sped. Clone 102H-N is a 102 bp fragment from – 73 to +18 of Spec2a plus 11 bp of polylinker from pUC18. Sped RNA was probed with a 390 bp cDNA insert extending from exons 2 through 5 of Sped (Carpenter et al. 1984). Endoderm/ mesoderm RNA was probed with a 720 bp BamHI – EcoRI DNA fragment representing a portion of the codogenic region of the cytoplasmic actin gene CyIIIa (Shott et al. 1984). Fragment B is a 151 bp HpaII-BstNI fragment that contains the SM50 factor B binding site (Davidson, 1989). The ‘P6’ clone is a 210 bp Rsal fragment that contains the P6 binding site of the Cyllla actin gene (Thèzè et al. 1990).

Nuclei were isolated from embryos at the indicated stages of development as described previously (Tomlinson and Klein, 1990). Nuclear extracts were prepared from the purified nuclei (Parker and Topol, 1984; Morris et al. 1986). The nuclei were suspended in 5 vol of buffer A (25 mM Hepes, pH 7.8, 5mM MgCl₂, 1 mM EDTA, 1 mM DTT, and 0.1 mM PMSF and homogenized with two strokes of a Dounce B pestle. The homogenate was transferred to a tube containing 0.1 vol of 4 M (NH4)₂SO₄, gently rocked under ice for 30 min and then centrifuged at 136000g for 60min. The supernatant was transferred to a new tube containing 0.25 g (NH4)₂SO4ml^-1 extract and centrifuged at 136000g for 15 min at 4°C. The nuclear pellet was suspended in 1ml of buffer C (15% glycerol, 25mM Hepes, pH7.8, 40mM KC1, 0.1mM EDTA, 1mM DTT, and 0.1mM PMSF) with two strokes of a 2ml Teflon pestle. The nuclei were dialyzed against 300 vol of buffer C for 1 h. The extract was cleared by a 2 min spin in a microfuge and stored at – 70°C in small volumes.

SpF1 proved to be a heat-stable protein, as is the case for USF. SpF1 was purified approximately 10-fold by utilizing a heat-treatment purification scheme performed as described by Sawadogo et al. (1988).

Radiolabeled DNA was prepared by the polynucleotide kinase procedure. Terminal phosphate groups were removed by incubating 2 – 4 μ g DNA in 50 μl 10mM Tris, pH8.0, and 1 mM EDTA with 0.2 unit of bacterial alkaline phosphatase at 65°C for 1h. Following extraction and precipitation, one hundred to three hundred nanograms of DNA was end-labeled in a total volume of 50 μ l at 37°C for 30min. The reaction mix contained 50 mM Tris, pH 8.0, 10 mM MgCl₂, 15 mM DTT, 0.33 μ M ATP, 150 μ CI [gamma-³²P]ATP (ICN; Irvine, CA) and 5 units of T4 polynucleotide kinase.

Protein-binding reactions were done on ice in a total volume of 20 μ l. Nonspecific protein binding was eliminated by incubating 2 – 8 μ g crude nuclear extract in 5mM Hepes, pH7.8, 87mM NaCl, 2.5mM KC1, and 1mM MgCl₂ with 100 μ g ml^-1 dI-dC for 15min. Competition reactions were performed by adding the indicated amounts of competitor to the reaction tubes for an additional 15 min. One to two nanograms of the appropriate radioactively labeled DNA (20 – 40 000 cts min^-1) was then added and incubated for 15 min. The reactions were loaded onto 5 % acrylamide gels in 0.25 × TBE (1 × TBE:89mM Tris, 89mM boric acid, and 2mM EDTA) and run at 4°C in 0.25 × TBE gel-running buffer. The gels were fixed in 10 % acetic acid and 10 % methanol and set up for autoradiography.

The protein-binding reactions were carried out as described above for the band shift gels, except the reaction buffer also contained ImM DTT and 0.2mM PMSF. At the end of the binding reaction, 3 μ g DNase I was added to the tube while on ice. After Imin, the DNase I was stopped with 30 μl of 0.6M NaCl, 0.4% SDS, 20 mM EDTA, 33 μ g ml^-1 tRNA, and 100 μ g ml^-1 proteinase K. The footprint reactions were extracted once with phenol/chloroform and precipitated with 0.5 vol 7.5 M ammonium acetate and 2 vol ethanol. The DNA pellet was washed once with 70% ethanol, dried and loaded onto an 8 % acrylamide urea denaturing gel (Maniatis et al. 1982).

Between 0.05 and 1.25 units of bacterial alkaline phosphatase were added to 20 μ l of blastula stage nuclear extract and incubated at ambient temperature for 10min. The radiolabeled 191 bp HinPI-SalI fragment was then added and incubated at room temperature for an additional 15 min.

Total RNA was denatured at 55 °C for 15 min in a total volume of 100 μ l containing 40mM MOPS, pH 7.0, 10mM sodium acetate, ImM EDTA, 50% formamide, and 17.5% formaldehyde solution (Maniatis et al. 1982). 5 μ g of RNA per slot was loaded onto Hybond-N filter paper (Amersham; Arlington Heights, IL). Filter hybridization and washes were performed as described previously (Tomlinson and Klein, 1990).

4 μ g of 0.8 Spec1-CAT and Spec2a-CAT plasmid DNA were digested with PvuII, which cuts at a unique site in the CAT gene. The opened plasmids were digested at 37°C with 450 units of exonuclease III in 50 μ l of 25 mM Tris, pH 7.8, 50 mM NaCl, 10mM MgCl₂, 100 μ g μ l^-1 BSA, and 1mM DTT for 30s intervals from 30 s to 2 min. The reaction was stopped with 2 μ l of 250 mM EDTA and extracted once with phenol/ chloroform and once with chloroform. The DNA was precipitated with ammonium acetate and ethanol. The DNA hybridization reaction contained 40 ng of phosphorylated primer and mutated oligonucleotides and 1 μ g of exonuclease Ill-digested Spec-CAT DNA in a total volume of 10 μL containing 25 mM Tris, pH 8.0,10 mM MgCl₂,100 Mg ml^-1 BSA, and 1 mM DTT. The hybridization reactions were incubated at 65 °C for 10 min and then cooled slowly to 30°C. The DNA was repaired by adding 4 μ l 2.5 mM dNTPs, 2 μ l 10 mM ATP; 1 μ l of a solution of 250mM Tris, pH8.0, 100mM MgCl₂, 1mgml^-1 BSA, and 10 mM DTT; 1 μ l Klenow; and 1 μ l T4 ligase in a total volume of 20 μ l-The reaction tubes were incubated at 37°C for 2h. 1 μ l of T4 ligase was added, and the tubes were incubated at 37 °C for an additional 30 min. Competent Escherichia coli HB101 cells were used for the transformation reactions. Altered sequences were confirmed by DNA sequence analysis.

Microinjections of sea urchin eggs were done as described previously (McMahon et al. 1985; Gan et al. 1990b). Plasmids derived from pSVo were linearized with BglI, which cuts within the pBR322 portion of the plasmid, and mixed with sperm DNA that had also been digested with BglI. Injected eggs were fertilized in situ, cultured at 18°C, and collected at blastula stage, 15 h after fertilization, or gastrula stage, 24h after fertilization.

CAT assays were performed using extracts from embryos that had been injected with the 0.8 Spec1 – CAT plasmid extending from – 800 to +133 of Sped and the Spec2a-CAT plasmid extending from – 1500 to +18 of Spec2a (Gan et al. 1990b). Injected embryos were collected along with approximately 1500 uninjected embryos of the same stage. One-half of the lysate was assayed for CAT activity as described by Gorman et al. (1982). The other half of the lysate was used for CAT DNA determination. The lysate was probed with a 500 bp fragment that contains the CAT gene coding region (Flytzanis et al. 1987).

The three Spec genes that have been analyzed in detail, Sped, Spec2a and Spec2c, share about 800bp of a strongly conserved sequence upstream of their translational initiation sites (Hardin et al. 1988; Gan et al. 1990b). The sequence match in this region is greater than 90%. In Spec2a, the conserved sequences are continuous and include an RSR repetitive sequence element. Elsewhere, we have provided evidence that there is a transcriptional enhancer-like element contained within the RSR sequence (Gan et al. 1990b). In Sped and Spec2c, the conserved sequence block is interrupted by 2.4 or 2.9 kb of nonhomologous DNA. In addition, the transcriptional initiation site for Sped is positioned within the inserted element. These features alter the relative placement of the RSR element among the three genes as well as other 5′ flanking DNA. In the case of Sped, homologous DNA that is 5′ flanking in Spec2a and Spec2c is part of the first exon of Sped (Hardin et al. 1988).

To identify proteins that bind to the conserved and nonconserved regions of the Spec genes, we undertook an in vitro analysis of DNA-protein interactions by incubating various Spec1 gene fragments with nuclear extracts from S. purpuratus blastula stage embryos. One DNA fragment, a 191 bp HinPI – SalI fragment covering the region from – 58 to +133 bp of the Spec1 gene (Fig. 1A), showed a strong-protein binding site by both band shift gel assays and DNasel footprinting. The protein factor interacting with this site was called SpFl for Spec gene Factor 1.

Protein binding was evident by the two major bands (SpFl complex A and B) shown on a band shift gel (Fig. IB). Protein binding was specific because competition assays showed that 10- to 500-molar ratios of homologous competitor DNA (Fig. 1C, lanes 3 – 5) and a 300-fold molar ratio of the 298 bp Ndel-SaR fragment (Fig. 1C, lane 6), a clone that encompasses the 191 bp HinPI-SalI fragment, inhibited the binding of proteins, but a 375-fold molar ratio of a 255 bp EcoRI-Ndel fragment, a DNA fragment of comparable size located immediately upstream of the 191 bp fragment, did not (Fig. 1C, lane 7). Protein-binding specificity was further verified by a DNase I footprint gel (Fig. 1D). Both the coding (CS) and noncoding (NCS) strands clearly showed that as the amount of blastula nuclear extract was increased (Fig. 1D, lanes 2 – 5 and 7 – 10), a single footprint region became observable at +39 to +60 bp, yet again, protein binding was inhibited by the homologous 191 bp HinPI-SalI fragment (Fig. ID, lanes 5 and 10). As discussed below, the sequence – CACGTG – protected by SpF1 is the core sequence motif for the human transcription factor USF (Fig. 1A).

To test whether the sequence identified by the DNase I footprint gel was the correct SpFl recognition site, a double-stranded complementary 22 bp oligonucleotide (CTB) corresponding to the footprint region was used in competition experiments (Fig. 1E). Two additional double-stranded oligonucleotides used as negative controls flank both ends of the region, a 23-mer homologous to the region immediately downstream (CT1+2), and a 21-mer immediately upstream (CT3+4). Both CT1+2 and CT3+4 overlap CTB by six nucleotides. Band shift gel analysis with blastula stage nuclear extracts incubated with the labeled 191 bp HinPI-SalI fragment showed that a molar ratio of 3.8 × 10⁴ of CTB to labeled fragment inhibited SpF1 binding, but the same molar ratio of the bordering oligonucleotides CT1+2 and CT3+4 did not (data not shown). Further verification of the SpF1-binding site was shown by the DNase I footprint gel in Fig. 1E. Lanes 3 and 8 show that while 1.5pmoles of CTB had little competitive effect, 15 pmoles (lanes 4 and 9) and 150pmoles of CTB (lanes 5 and 10) completely inhibited the SpF1 footprint. These experiments demonstrated the presence of a strong DNA-protein interaction in the 5′ untranslated leader region of the Spec1 gene.

USF is a transcription factor that binds the major late promoter of adenovirus (Sawadogo and Roeder, 1985) as well as a variety of other mammalian genes containing a core sequence of CACGTG (Sawadogo et al. 1988; R. Roeder, personal communication). Inspection of the SpF1-binding site in the Spec1 gene suggested that SpF1 may be similar to USF. USF has been purified to near homogeneity from HeLa cells (Sawadogo et al. 1988) and was found to be in two forms with apparent molecular weights of 43000 and 44000. Both forms show identical DNA-binding properties, though it is not clear what their relationship is to each other (Sawadogo, 1988).

To determine whether USF could recognize the SpF1-binding site on the Spec1 gene, purified USF was incubated with the 191 bp HinPI-SalI fragment and resolved on band shift gels (Fig. 2). Two complexes were apparent with very similar mobilities as observed for sea urchin embryo nuclear extracts. Thus, SpF1 complexes A and B may correspond to the same two molecular weight forms of USF. Increasing amounts of CTB oligonucleotide efficently inhibited USF binding (Fig. 2A, lanes 3 – 7); molar ratios of 200 totally eliminated USF binding (lane 4). USF-binding specificity was further demonstrated using the CT1+2 and CT3+4 oligonucleotides as negative controls (Fig. 2B), in that molar ratios of 200 had no discernable competitive effect (lanes 3 and 5) and molar ratios as high as 3.8 × 10⁴ of CT1+2 and CT3+4 to labeled fragment (lanes 4 and 6) did not completely inhibit USF binding.

Sawadogo et al. (1988) have shown that USF is heat stable, as heating at 70°C did not significantly alter USF-binding activity. The sea urchin embryo nuclear extracts used in our experiments were subjected to a heat treatment purification protocol precisely the same as Sawadogo et al. (1988), implying that SpF1 was also heat stable. Thus, based on three criteria - sequence specificity, electrophoretic mobility of the DNA-protein complexes, and heat stability – SpF1 is similar to human USF.

The Spec1 gene is transcribed in vitro by nuclei prepared from ectoderm cells, but not by nuclei from endoderm/mesoderm cells (Tomlinson and Klein, 1990) . These results suggested that ectoderm-specific transcription factors might exist that are involved in Spec1 gene transcription. We asked whether SpF1 was present only in the ectodermal cells of the embryo. Ectodermal and endodermal/mesodermal cells from S. purpuratus pluteus stage embryos were separated and purified (McClay, 1986). The degree of cross contamination was tested by isolating total RNA from each cellular fraction and spotting the RNA on two identical blots (Fig. 3A and B). The extent of ectodermal contamination in the endoderm/mesoderm fraction was monitored with a Spec1 codogenic probe and the extent of endodermal/mesodermal contamination in the ectoderm fraction with the endoderm/mesoderm-specific probe Cylla (Cox et al. 1986). As shown in Fig. 3A and B, there was minimal contamination in both fractions.

Nuclear extracts prepared from the ectoderm and endoderm/mesoderm fractions were incubated with the 191 bp HinPI-SalI fragment of Spec1, with a 151 bp HpaII-BstNI fragment upstream of the SM50 gene, and with a 210 bp Rsal fragment upstream of the CyHIa actin gene. The SM50 gene is expressed only in the primary mesenchyme lineage and thus was used as a control for detection of protein binding of endodermal/ mesodermal origin. At least one protein, termed factor B, is known to bind the 151 bp fragment (Davidson, 1989). The 210 bp Rsal fragment from the Cyllla actin gene was used as an additional control. It contains a ‘P6’ site that binds a protein known to be active at approximately equal levels in ectoderm and endoderm/ mesoderm extracts (Thèzè et al. 1990; E.H. Davidson, personal communication). Equivalent amounts of heat-treated ectoderm and endoderm/mesoderm extracts were used in the band shift gels in Fig. 3C – F. SpFl was greatly enriched and probably resides exclusively in ectodermal cells as shown in Fig. 3C. Both complexes, A and B, were observed. Increasing amounts of ectodermal extract resulted in increasing amounts of complexes (Fig. 3C, lanes 2 – 4). However, endoderm/ mesoderm extracts revealed no complex formation, even at the highest extract level (Fig. 3C, lanes 5 – 7). Nevertheless, the nuclear proteins of the endoderm/ mesoderm cells have binding capacity as shown in Fig. 3D. The 151 bp Hpali-BstNl fragment from the SM50 gene was capable of forming several protein complexes with the endoderm/mesoderm fraction (Fig. 3D, lanes 5 – 7).

SpF1 specificity was further demonstrated by adding increasing amounts of oligonucleotide CTB to the ectodermal nuclear extract (Fig. 3E, lanes 3 – 6). Competition was complete with the addition of molar ratios of 200 of CTB to labeled fragment (Fig. 3E, lane 4), yet molar ratios of 2.5 × 10³ of the 151 bp SM50 fragment had no effect (Fig. 3E, lane 7). Conversely, increasing amounts of the 151 bp SM50 fragment inhibited any endoderm/mesoderm binding proteins from complex formation (Fig. 3F); e.g. 10pmoles of the 151 bp fragment showed total competition (Fig. 3F, lane 5), yet 300pmoles of CTB failed to effectively compete (Fig. 3F, lane 7). The lack of complex formation observed with the SM50 fragment and ectoderm nuclear extracts was different from that seen by Davidson and his co-workers, in that in their hands complexes were found in both fractions (E.H. Davidson, personal communication). The basis of this discrepancy may be that, in the experiment shown in Fig. 3C – F, extracts were heat treated before use. However, when nonheat-treated extracts were used, the same result was obtained with the Sped 191 bp HinPI-SalI fragment; that is, ectoderm but not endoderm/mesoderm nuclear extracts were capable of forming complexes, although with non-heat-treated extracts, the bands were not as distinct as with the heat-treated extracts (Fig. 3G). In this experiment, the 210 bp Rsal fragment from the Cyllla actin gene formed the predicted ‘P6’ complex with nuclear extracts from either fraction (Fig. 3H).

Because phosphatases are prominent in gut tissue of sea urchin embryos, it was possible that the lack of complex formation with the Sped fragment was due to dephosphorylation of SpF1, rendering it inactive. We incubated blastula extracts with varying amounts of bacterial alkaline phosphatase (BAP) and used these extracts in a band shift gel analysis. No decrease in the ratio of complexed to uncomplexed probe was observed, though the total amount of labeled probe decreased with increasing phosphatase due to déphosphorylation of the probe (Fig. 4, lanes 2 – 4). These experiments suggested that phosphorylated forms of SpFl were not required for SpFl binding. Similar conclusions have been drawn from experiments with the human USF (M. Sawadago, personal communication). Our results demonstrated that within the limits of our analysis, SpFl resided exclusively in the ectodermal cells of the embryo.

The SpF1-binding site found in the 5′ untranslated leader region of the Sped gene also occurs upstream of the Spec2a and Spec2c genes (Fig. 1A). The sequence is identical with the exception of a G → A change corresponding to position +40 of the Sped gene. As discussed above, these sequences are part of the conserved sequence block shared among the three Spec genes (Gan et al. 1990b). The CACGTG core sequence is located at positions – 66 to – 61 in Spec 2a and – 3016 to – 3011 in Spec 2c. We asked whether SpF1 had the ability to bind the SpF1-binding site of Spec 2a. A 102 bp HindIII-NsiI fragment from Spec2a containing the SpF1-binding site was used in this experiment. SpF1 binding was demonstrated using heat-treated nuclear extracts prepared from blastula stage embryos. Fig.5, lane 2 shows that two complexes similar to those observed with the Spec1 fragment were seen with the Spec2a fragment. Futhermore, these complexes were eliminated with the competitor CTB oligonucleotide (Fig. 5, lane 3). Thus, at least in vitro, SpF1 bound the Spec2a site as efficiently as it did Spec1.

The sea urchin species S. purpuratus and L. pictus diverged 35 million years ago (Smith, 1988). The Spec genes of S. purpuratus and the Spec-like gene, LpS1, of L. pictus have diverged to a degree such that within the protein-coding region only the calcium-binding domains are recognizably similar (Xiang et al. 1988). However, in common with the Spec genes, LpS1 is expressed exclusively in the aboral ectoderm (Xiang et al. 1988; Tomlinson and Klein, 1990). To demonstrate the conservation of SpFl in L. pictus, nuclear extracts were prepared from L. pictus blastulae and incubated with the 191 bp HinPI-SalI fragment of the 5. purpuratus Sped gene. The band shift gel displayed in Fig. 6A demonstrated that an SpFl-like factor resided in L. pictus. Complex formation with L. pictus extracts (Fig. 6A, lane 2) appeared to be similar to that with S. purpuratus extracts because increasing amounts of the CTB oligonucleotide (Fig. 6A, lanes 3 – 7) efficiently prevented binding. Binding specificity was further demonstrated by the fact that neither control oligonuc leotides CT1+2 nor CT3+4 inhibited binding (lanes 8 and 9). DNase I footprinting with L. pictus extracts showed that the L. pictus protein bound the same site as observed with S. purpuratus extracts (Fig. 6B). Binding specificity was again demonstrated by lanes 5 and 10 in which the CTB oligonucleotide effectively inhibited binding. In both the band shift analysis and the DNase I footprinting, it was apparent that the L. pictus protein did not bind the SpF1 site on the Sped gene as tightly as the 5. purpuratus protein. Nevertheless, these experiments demonstrated that L. pictus extracts contained SpF1 activity and strongly suggested that SpF1 is a conserved protein with identical function in both species.

Site-directed mutagenesis was employed to mutate the SpF1-binding site in 0.8 Spec1 – CAT and Spec2a – CAT reporter gene plasmids. The mutations and plasmids are shown schematically in Fig. 7A and B. In these mutants, the core – CACGTG – sequence plus an additional 4 bp upstream have been replaced with a different 10 bp sequence. Fig. 7C and D show a band shift analysis where the Sped and Spec2a mutant fragments were incubated with heat-treated S. purpuratus blastula stage extracts. The wild-type control fragments yielded the expected complexes, but no significant complex formation was observed with the mutant fragments. These results support the conclusion that SpF1 binds to a specific sequence containing the – CACGTG – motif.

The wild-type plasmids depicted in Fig. 7A and B have been shown to be active when injected into the heterologous L. pictus egg (Gan et al. 1990b). In an earlier study, CAT activity resulting from microinjecting Spec1 – CAT or Spec2a – CAT was seen in blastula, gastrula, prism and pluteus stage embryos but not in cleavage stage (Gan et al. 1990b). These data were consistent with nuclear run-on experiments showing the Spec genes are activated at the late cleavage - early blastula stage (Tomlinson and Klein, 1990). Moreover, many studies using L. pictus eggs have produced information on some of the elements needed for the correct expression of sea urchin histone genes for heterologous species. For example, Vitelli et al. (1988) have shown by SP6 RNA mapping that all five early histone genes of Psammechinus miliaris were properly expressed in L. pictus eggs. Colin et al. (1988), Lai et al. (1988,1989), and Di Liberto et al. (1989) have identified sequences needed for the correct temporal and quantitative expression of S. purpuratus histone genes. The studies of Lai et al. (1988) have shown that a single base pair change within the 6 bp histone Hl Spl-binding sequence can dramatically alter temporal and quantitative expression.

Because SpF1 activity was present in L. pictus extracts, we argued that L. pictus eggs were appropriate for testing the in vivo activity of the SpF1-binding site. Unexpectedly, however, as shown in Fig. 7E and F, no significant differences were ever observed between Sped or Spec2a mutant and wild-type plasmids when eggs were injected, fertilized, allowed to develop to the gastrula stage (Fig. 7E) or the blastula and gastrula stage (Fig. 7F), and subsequently analyzed for CAT activity and plasmid DNA content. We had previously made an internal deletion in the Spec2c 5′ flanking region that removed the SpF1-binding site plus substantial amounts of DNA both upstream and downstream from the binding site and found no differences in Spec2c-CAT activity (Gan et al. 1990b). These results do not imply a lack of general responsiveness of the expression system to mutations associated with cis transcriptional elements since several other regions including the 5′ half of the RSR could be deleted in these plasmids with drastic effects on Spec gene expression (Klein et al. 1990; Gan et al. 1990b). We have also repeated the microinjection experiments with the Spec1-CAT mutant using S. purpuratus eggs and have got the same result; that is, there was no effect on CAT activity when the SpF1 site was mutated (data not shown). From these results we conclude that the SpF1 protein-Spec DNA interaction is not essential for expression of the microinjected plasmids.

Our results clearly demonstrate the existence of a Spec–DNA binding protein whose activity at the pluteus stage is restricted to ectodermal cells. This protein binds to a known family of genes, the Spec genes and displays cell type specificity. A nuclear protein binding to Spec DNA might be expected to be an even earlier marker of aboral ectoderm than the Spec genes themselves. We have not proven that SpFl activity is localized solely to aboral ectoderm lineages, but its tight association with the Spec genes suggests that this may be the case. If SpFl synthesis precedes Spec gene activation, its temporal and spatial features should be highly interesting. In sea urchin development, the animal-vegetal axis is specified during oogenesis, and it is clear that more vegetal blastomeres can induce various fates in cells above them (Horstadius, 1973; Wilt, 1987; Davidson, 1989). Thus, in normal development, micromeres at the vegetal pole, which give rise to skeletal mesenchyme, probably induce gut, muscle and pigment tissue in the adjacent tiers. It is not clear what action vegetal blastomeres have on ectodermal differentiation. However, under certain experimental conditions, such as ectopic transplantation of micromeres (Horstadius, 1973) or treatment with vegetal inducing agents such as Li⁺ (Livingston and Wilt, 1989), gut, muscle, pigment and skeletal tissue can be induced in animal blastomeres normally fated to become oral or aboral ectoderm. For SpF1 activity to occur only in the ectoderm, there must be differential expression or activation at some point during embryogenesis. However, if SpF1 is an early ectoderm marker, its activity must be reversible, since cells destined to become ectoderm can be converted to non-ectoderm cell types.

Recent studies with lineage tracers have shown that the second major embryological axis in sea urchins, the oral-aboral axis, is specified by at least the 2-cell stage (Cameron et al. 1989). Davidson (1989) has suggested that a short time after fertilization, a reorganization of the cytoplasm is responsible for setting up the oral-aboral axis in a fashion similar to what has been observed for the dorsal-ventral axis in amphibian embryos. Once this event takes place, it is likely, based on the regulative ability of the ectodermal cells, that a series of intercellular interactions results in the differential activation and inactivation of transcription factors along the axis. This in turn leads to the activation of various cell-type-specific genes and the differentiation of the oral and aboral ectoderm cell types. We argue that SpFl activity appears only in the aboral ectoderm lineages and fits into the above scheme by being activated very early after oral-aboral axis specification in the aboral ectoderm region, or alternatively, by being inactivated in the oral and more vegetal regions. SpF1-binding activity appears to be absent in unfertilized eggs (C.R.T., unpublished results), suggesting that SpF1 must be activated in aboral ectoderm cells following fertilization.

Whatever its role in Spec gene activation, SpF1 has properties that suggest it may be a sea urchin equivalent of the human USF. USF plays a key role in activating the major late promoter of adenovirus (Sawadogo and Roeder, 1985; Lee et al. 1988). It binds approximately 60 bp upstream of the adenovirus major late start site. In vitro, USF has been shown to interact with the TATA binding factor, TFIID, which in adenovirus binds 30 bp downstream from USF (Sawadogo et al. 1988; Workman et al. 1990). It has been proposed that USF functions in stimulating transcription by associating with TFIID. However, its precise role is not clear. Workman et al. (1990) have shown that under conditions of in vitro chromatin assembly, the fold stimulation of transcription by USF was several-fold greater than in the absence of chromatin assembly. These authors suggested that an important role for USF is to establish the transcriptional potential of the promoter during chromatin assembly by facilitating the binding of TFIID. An additional role is to enhance utilization of the promoter in transcription by increasing the loading of RNA polymerase II and additional factors (Workman et al. 1990). USF has been shown to bind promoter regions from several mammahan genes besides adenovirus including those encoding mouse metallothionein, rat fibronectin, human growth hormone and several fiver-specific genes (Sawadogo et al. 1988). A USF-binding site occurs in the intragenic region of the chicken and duck histone H5 genes and USF appears to activate expression of histone H5 in duck erythrocytes (During et al. 1990). Moreover, a USF-like protein has been purified from Xenopus oocytes, and there is a USF-binding site 250 bp upstream from the Xenopus TFlIIa gene (Scotto et al. 1989; Hall and Taylor, 1989).

Human and Xenopus USF cDNA clones have been isolated and sequenced, and they show the presence in their open reading frame of a basic region followed by amphipathic helices, termed the HLH (helix-loop-helix) domain by Murre et al. (1989). Sequence comparisons relate this 60 amino acid stretch to those found in several nuclear proteins including: C-myc, N-myc, L-myc, MyoDl, myogenin, myf5, daughterless, twist, achaete-scute and the immunoglobulin kappa chain enhancer binding proteins, E12 and E47. Some of these proteins are known to have essential roles as transcription factors involved in determining cell fate. For example, MyoDl, myogenin and myf5 control muscle differentiation in mammals (Davis et al. 1987; Wright et al. 1989; Edmondson and Olson, 1989; Braun et al. 1989), and achaete-scute is associated with the formation of the central and peripheral nervous system in Drosophila (Villares and Cabrera, 1987). It is possible that SpF1 belongs to this group of proteins and plays some role in the specification of aboral ectoderm lineages.

We have presented circumstantial evidence that SpF1 is a transcription factor involved in activating Spec genes. There is a high affinity binding site for SpF1 in the Spec gene conserved sequence block, and the binding properties as well as the sequence of the binding sites are similar to USF (Sawadogo et al. 1988). However, we have not been able to demonstrate SpF1 activity in vivo. Mutations affecting the binding of SpF1 to Spec DNA have no effect in our expression system, either with L. pictus or S. purpuratus eggs. If SpF1 played a role in aboral ectoderm-specific expression, it would be expected to be a positive activator of transcription since its binding activity is found restricted to ectodermal cells where the Spec genes are active. However, SpF1 does not appear to be essential for expression of the microinjected Spec genes. Nevertheless it is still possible that SpF1 is actually required for correct endogenous Spec gene expression. For example, if the major role of SpF1 were to potentiate the Spec promoters by keeping them in an open configuration (similar to what may be occurring with USF), it might be that such a requirement is not necessary for the microinjected genes. The latter genes are present in the embryo in hundreds of copies per cell and the structural features of the chromatin may not be the same as exists in the endogenous genes. We have recently shown that the presence of an SpFl-binding site is not sufficient for conferring proper spatial expression from Spec promoters (Gan et al. 1990a). Spec-LacZ reporter genes containing 5′ flanking and 5′ untranslated leader sequences from Spec1, Spec2a and Spec2c were microinjected into Strongylocentrotus or Lytechinus eggs, and embryos were analyzed in situ for β -galactosidase activity. In the case of Spec2a, 1.5 kb of 5′ flanking DNA plus 18 bp of 5′ untranslated leader sequence was sufficient for correct aboral ectoderm expression (Gan et al. 1990a). Unaccountably, Spec1-LacZ and Spec2c-lacZ reporter gene constructs containing highly similar regions were expressed in other cell types besides aboral ectoderm (Gan et al. 1990a). Yet both of these latter constructs contained unaltered SpFl-binding sites.

The Spec genes have evolved in an unusual manner. Insertions or deletions normally would not be tolerated in the 5′ flanking DNA of genes transcribed by RNA polymerase II. However, in the Spec genes of 5. purpuratus, these events seemed to have occurred frequently. In addition, the Spec-related gene of L. pictus, LpSl, has undergone an internal duplication resulting in a protein of twice the normal molecular weight (Xiang et al. 1988). Whether the difference in placement of the SpF1-binding site (the 5′ untranslated leader of Spec1, the – 60bp region of Spec2a, and the – 3.0 kb region of Spec2c) have any effect on Spec gene expression is unknown. In this regard, another part of the Spec gene conserved sequence block, the RSR repeat, contains an enhancer-like element that has a much stronger effect on the expression of Spec2a, where the repeat is only 100 bp upstream of the start of transcription, than on Spec1 or Spec2c, where it is 2.4 to 2.9 kb upstream (Gan et al. 1990b). With respect to the SpF1-binding sites, Spec2a most resembles the adenovirus major late promoter in that the SpF1/USF-binding sites and TATA boxes are positioned in a similar fashion. However, as discussed in the Results section, elimination of the SpF1-binding site in Spec2a does not affect expression.

It is likely that the USF-binding site exists on genes other than those in the Spec family. The expression properties of such genes would be interesting. It is also possible that there is a family of USF-like proteins or even unrelated proteins that recognize similar binding sites. Recently, Beckman et al. (1990) have reported the existence of another helix-loop-helix protein, distinct in sequence from USF, that binds to the same -CACGTG-sequence motif. Purification of SpF1 and the eventual cloning of the SpF1 gene should provide powerful tools for further investigation. The most crucial problem at present is ascribing a function to SpF1.

We thank Dr W. Michael Perry for initially pointing out the relationship of USF to SpF1 and for his insight and advice. We also thank Dr M. Sawadogo for providing purified USF. The oligonucleotide synthesizer was supplied by NIH grant CA16672. This work was supported by an NIH grant (HD22619) to W.H.K.