Structure and Promoter Activity of the LpSl Genes of L ytechinus pictus DUPLICATED EXONS ACCOUNT FOR LpSl PROTEINS WITH EIGHT CALCIUM BINDING DOMAINS*

The LpSl genes of the sea urchin Lytechinus pictus are activated early in development in aboral ectoderm cells. They therefore have ontogenic properties similar to their counterparts in Stronglyocentrotus purpura- tus, the Spec genes. Both gene families encode proteins belonging to the calmodulin superfamily as evidenced by the presence of distinct EF-hand (helix-loop-helix) domains. The presence of eight EF-hand domains in LpS 1 proteins suggests that the LpS 1 genes arose from a duplication of an ancestral Spec-like gene. The LpS 1 genes were further analyzed to increase our under-standing of the mechanisms underlying their evolution and activation in aboral ectoderm cells. Genomic DNA blot analysis showed two LpSl genes, LpSla and LpSlj3, which did not appear to be closely linked. LpSl genomic clones were isolated by screening an L. pictus genomic library with an LpSl cDNA clone, and partial gene structures for both LpSla and LpSlj3 were constructed. These revealed internal duplication of the LpSl genes that accounted for the eight EF-hand domains in the LpSl proteins. Duplication of exon l in both genes suggested four different LpSl proteins could be derived from the LpSl genes. Primer extension to map the transcriptional initiation sites of the LpSl genes and sequencing analysis showed there was little in common among the 5”flanking regions of the LpSl and Spec genes except for the

The LpSl genes of the sea urchin Lytechinus pictus are activated early in development in aboral ectoderm cells. They therefore have ontogenic properties similar to their counterparts in Stronglyocentrotus purpuratus, the Spec genes. Both gene families encode proteins belonging to the calmodulin superfamily as evidenced by the presence of distinct EF-hand (helix-loop-helix) domains. The presence of eight EF-hand domains in LpS 1 proteins suggests that the LpS 1 genes arose from a duplication of an ancestral Spec-like gene. The LpS 1 genes were further analyzed to increase our understanding of the mechanisms underlying their evolution and activation in aboral ectoderm cells. Genomic DNA blot analysis showed two LpSl genes, LpSla and LpSlj3, which did not appear to be closely linked. LpSl genomic clones were isolated by screening an L. pictus genomic library with an LpSl cDNA clone, and partial gene structures for both LpSla and LpSlj3 were constructed. These revealed internal duplication of the LpSl genes that accounted for the eight EF-hand domains in the LpSl proteins. Duplication of exon l in both genes suggested four different LpSl proteins could be derived from the LpSl genes. Primer extension to map the transcriptional initiation sites of the LpSl genes and sequencing analysis showed there was little in common among the 5"flanking regions of the LpSl and Spec genes except for the presence of a binding site for the transcription factor USF. A sea urchin gene-transfer expression system showed that 762 base pairs (bp) of 5"flanking DNA and 17 bp of 5"untranslated leader sequence of the LpSlj3 gene were sufficient for correct temporal and spatial expression of reporter chloramphenicol acetyltransferase and lacZ genes in sea urchin embryos. Deletions at the 5' end to either 511 or 368 bp resulted in a 3-4 fold decrease in chloramphenicol acetyltransferase activity and disrupted the exclusive activation of the lac2 gene in aboral ectodermal cells. Based on a lineage analysis among the LpSl and Spec gene families and other related genes, we propose a model in which LpS 1 genes evolved from a series of duplications of an ancestral Spec-like gene.
* This study was supported by National Institutes of Health Grant HD22619 (to W. H. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in thispaper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) M62848.
Calmodulin and its many relatives form a large, complex superfamily with highly diverse functions and widespread distribution. The superfamily consists mainly of low molecular weight intracellular calcium-binding proteins related to each other through a conserved helix-loop-helix motif, the EF-hand (Kretsinger et al., 1988). It is thought that the superfamily evolved from a primordial EF-hand-containing polypeptide by tandem gene duplication, sequence divergence, and dispersion (Demaille, 1982). Many of these proteins play important roles in modulating intracellular calcium levels and in regulating cellular events via calcium (Norman et al., 1987). Much is already known about their evolution, expression, and function (Moncrief et al., 1990).
In sea urchins, the calcium-binding Spec proteins represent interesting examples of the evolution and specialization of superfamily members. These proteins accumulate in embryos and larvae, but not in adults, and are found only in cell lineages giving rise to the aboral ectoderm, a squamous epithelium covering the surface of the late stage embryo and larva (Bruskin et al., 1982;Lynn et al., 1983;Carpenter et al., 1984;Klein et al., 1990). The exact function of the Spec proteins is unknown, but they may play a role in elevating calcium ion concentration in the aboral ectoderm, perhaps by enhancing transport of calcium from the sea water to the blastocoele, where high calcium concentrations are required for skeletogenesis (Klein et al., 1991).
We have been studying the Spec proteins and their genes from two different sea urchin species, Strongylocentrotus purpuratus and Lytechinus pictus, with the notion that a comparative analysis would be useful for gaining insights into Spec protein function. It should also be helpful for elucidating the mechanisms by which the Spec genes are activated specifically in aboral ectoderm cells.
In S. purpuratus, there are seven or eight Spec genes encoding approximately 10 polypeptides as visualized by twodimensional gel analysis (Bruskin et al., 1982;Hardin et al., 1988). The Spec proteins range in size from 14,000 to 17,000 Da and contain four helix-loop-helix domains per molecule (Bruskin et al., 1982;Carpenter et al., 1984;Klein et al., 1991). Recently, we initiated experiments to map the transcriptional regulatory elements for three Spec genes using a sea urchin embryo gene-transfer system. Promoter regions from all three genes are capable of activating a CAT' reporter gene at the appropriate time in development, but only the Spec 2a promoter yields aboral ectoderm specificity when fused with a lac2 reporter gene (Gan et al., 1990a, 199Ob). In the case of The abbreviations used are: CAT, chloramphenicol acetyltransferase; bp, base pair(s); kb, kilobase pair(s); Pipes, 1,4-piperazinedieth-anesulfonic acid; X-Gal, 5-bromo-4-chloro-3-indoyl (3-D-galactoside.

10524
Specea, 1516 bp of 5"flanking DNA plus 18 bp of 5"untranslated leader sequence are sufficient for aboral ectoderm expression, while 5.6 kb of Spec 1 5"flanking DNA plus 120 bp of 5"untranslated leader sequence result in preferential expression in mesenchymal cell types (Gan et al., 1990a).
All the characterized Spec genes contain an enhancer-like element in their 5'-flanking DNA that appears to be partly responsible for the activation of transcription at the correct time but does not confer aboral ectoderm specificity by itself (Gan et al., 1990b).' The Spec gene promoter regions also contain a high affinity binding site for USF, a transcription factor originally found associated with the upstream region of the major late genes of adenovirus (Sawadogo and Roeder, 1985;Tomlinson et al., 1990). In sea urchins, USF binding activity is found only in ectodermal cells; posttranslational inactivation may occur in other cell types .3 In contrast to S. purpurutus, L. pictus has only three Speclike proteins, with molecular sizes of 34,000 rather than 14,000-17,000 Da (Xiang et al., 1988). A cDNA clone, called LpS1, encodes at least one of these proteins and contains eight helix-loop-helix domains rather than four, suggesting that, in L. pictus, the Spec-like proteins are tandemly duplicated versions of their S. purpuratus counterparts (Xiang et al., 1988). The LpSl proteins are highly diverged from the S.
purpurutus Spec proteins, showing no more similarity to them than to other seemingly distant members of the superfamily (Xiang et al., 1988;Moncrief et al., 1990;Klein et al., 1991). Nevertheless, LpSl genes are activated at the same developmental stage as Spec genes, and LpSl messages accumulate only in aboral ectoderm cells, implying an identical functional role for LpSl and Spec proteins in the two species (Xiang et ul., 1988). We employed the LpSl cDNA to ask several specific questions regarding the evolution and expression of the Spec-like genes in the two species. We show that there are two LpSl genes in the L. pictus genome, resulting from a recent gene duplication. The two LpSl genes could result in as many as four different LpSl proteins. The coding exons of the LpSl genes have been duplicated, explaining the eight helix-loophelix domains in the LpSl proteins. We also show that 762 bp of 5"flanking and 17 bp of 5"untranslated leader DNA are sufficient for activating the LpSl genes at the appropriate time and in the correct cell type. In addition, sequence comparisons among LpSl and Spec 5'-flanking regions show that the USF binding site is a major conserved element.

EXPERIMENTAL PROCEDURES
Southern Blot Analysis of Genomic DNAs-Genomic Southern blot analysis was done as described (Hardin et al., 1988). L. pictus sperm DNA was digested with an appropriate restriction enzyme, separated on an agarose gel, transferred to N-Hybond filter, and hybridized to the 1.1-kb EcoRI fragment of the LpSl cDNA clone (Xiang et al., 1988). Final wash conditions were 68 "C and 0.15 M NaCl.
Pulsed Field Gel Electrophoresis-L. pictus sperm were washed twice in ice-cold phosphate-buffered saline, 1 mM dithiothreitol, and resuspended in phosphate-buffered saline to a concentration of 1.6 X 10' sperm/ml. An equal volume of 1% low-melting agarose (FMC Bioproducts) was added to the sperm for a total volume of 3 ml. The temperature of the sperm suspension and agarose was adjusted to 42 "C, and the two were mixed completely, pipetted into a plus mold (Bio-Rad), and incubated at 4 "C for 5 min. The plugs were transferred to 50 ml of 0.5 M EDTA, pH 8.0, 1% lauroylsarcosine, and 1 mg/ml proteinase K (ESP buffer) and incubated at 50 "C for 48 h. The ESP buffer was decanted and the plugs washed twice for 1 h with 50 ml of TE (10 mM Tris-HC1,l mM EDTA, pH 7.5) containing ' L. Gan and W. Klein, unpublished results. M. Kozlowski and W. Klein, unpublished results. 1 mM phenylmethanesulfonyl fluoride at room temperature. The plugs were washed an additional time in 50 ml of TE without phenylmethanesulfonyl fluoride for 30 min and stored at 4 "C in 10 ml of 0.5 M EDTA. The DNA was prepared for restriction enzyme digestion by washing the plugs two times with 50 ml of TE for 30 min at room temperature and 30 min at 4 "C in 3 volumes of restriction enzyme buffer. The buffer was removed and replaced with 3 volumes of fresh restriction enzyme buffer. Approximately 40 units of the appropriate restriction enzyme were added, and digestion was carried out for 24 h at the appropriate temperature.
The plugs were loaded directly into the wells of a 1% agarose gel (Bio-Rad CHEF-DR I1 Megabase DNA pulsed field electrophoresis system) and electrophoresed for ramp times of 30-40 s at 200 V. The gels were stained, photographed, and blotted as described by Sambrook et al. (1989). Hybridizations and washes were performed as in Tomlinson and Klein (1990).
Screening of an L. pictus Genomic Library-An L. pictus XEMBL3 genomic library  was screened and rescreened using the 0.52-, 0.44-, and 1.1-kb EcoRI fragments of the LpSl cDNA clone (Xiang et al., 1988) at a final wash condition of 68 "C, 0.15 M NaCl. Sixteen positive clones were obtained and subsequently mapped by restriction digestion.
Sequence Analysis of LpSl Genomic Clones-All positive clones were hybridized to the three EcoRI fragments of the LpSl cDNA clone. The smallest hybridizing fragments were subcloned into Bluescript. Oligonucleotides corresponding to various exons of the LpSla (i.e. the gene with sequence corresponding to the LpSl cDNA) were synthesized and used to sequence both LpSla and LpSlp hybridizing fragments by the chain termination method (Sanger et al., 1977).
Primer Extension Analysis-A 17-bp primer complementary to +lo6 to +90 in the LpSla or LpSlp gene was synthesized and endlabeled with [Y-~'P]ATP to a specific activity of -1 X lo9 cpmlpg.
Then, 20 pg of L. pictus total ectodermal RNA from pluteus stage embryos was annealed to 15 ng of labeled primer in 5 p1 of annealing buffer (40 mM Pipes, pH 6.4, 1 mM EDTA, and 0.4 M NaCl) for 4 h at 42 "C. The reaction was then carried out for 2 h at 42 "C in a 50-p1 solution containing 50 mM Tris-HC1 (pH 8.3), 75 mM KCl, 3 mM MgCl', 10 mM dithiothreitol, 0.5 mM each dATP, dCTP, dGTP, and dTTP, 80 units of RNasin (Promega Biotec, Madison, WI), and 200 units of Moloney murine leukemia virus reverse transcriptase (BRL). The reaction products were precipitated with 2.5 volumes of ethanol and separated on an 8% sequencing gel.
Plasmid Constructs for Microinjection Assays-All the LpSlp-CAT constructs were derived from pSVo. CAT(S), which was a derivative of pSVo. CAT (Gorman et al., 1982;Flytzanis et al., 1987). The 762-CAT construct was obtained by subcloning the HincII-EcoRI fragment of LpSlp from +17 to -762 bp into pSVo.CAT(S) in a forward orientation. A corresponding construct in the reverse orientation was also obtained. Deletion mutants were obtained following the En0111 deletion protocol (Promega, Erase-a-base kit). The 511-CAT construct contained LpSlp upstream sequences from +17 to -511 bp, and the 368-CAT contained sequences from +17 to -368 bp, The LpSlp-lacZ constructs were generated from the pNL vector (Gan et al., 1990a). To create the 762-lacZ, the LpSlp HincII-EcoRI fragment containing 5"flanking sequences from +17 to -762 bp was subcloned into the pNL vector in a forward orientation through Sal1 linker ligation. To generate the 3700-lac2 construct, an upstream 2.9kb EcoRI-PstI fragment of LpSlp containing sequences from -763 to about -3700 bp was inserted into the PstI site of the 762-lacZ construct in a forward orientation through PstI linker ligation. The deleted HincII-EcoRI fragments were also subcloned into the pNL vector in a forward orientation to generate the 511-lac2 and 368-lacZ constructs.
Microinjection of Sea Urchin Eggs or Zygotes-The collection of sea urchin gametes and microinjection of eggs was the same as previously described (McMahon et al., 1985). Injections were performed with either unfertilized or fertilized egges. Each egg was injected with approximately 2 pl of a 40% glycerol solution containing 2000 molecules of the linearized plasmid constructs plus sea urchin sperm carrier DNA at a molar ratio of 1:5. After microinjection, L. pictus embryos were cultured at 18 "C, and Lytechinus uariegatus embryos at room temperature. Sperm carrier DNA was isolated from either S. purpuratus or L. pictus as described (Lee et al., 1984) and digested with PstI. The CAT constructs were linearized with BglI. The 3700-lacZ construct was linearized with BamHI, and the 762-, 511-, and 368-lacZ constructs with PstI.
CAT Assays and DNA Determinations-Microinjected embryos were harvested at the desired developmental stages and mixed with approximately 1.500 uninjected embryos. One-half' of these embryos were used to measure CAT activity (McMahon et d., 1985); the other half were used to determine the plasmid DNA levels. Slot, blots were prepared (Flytzanis ct a/., 1987), hybridized with a CAT DNA probe, and washed in 0.15 M NaCl at 68 "C. X-Go/ Staining and Microscopy--X-Gal staining for /j-galactosidase activity was performed as described by Gan et nl. (1990a). Several hundred L. cnrirgatus or L. pictus fertilized eggs were injected with the LpSld-lnci! constructs, harvested at gastrula through prism stages, and fixed with lri glutaraldehyde in Ca'+-free sea water at room temperature for ahout 15 min. After fixation, the embryos were rinsed twice with staining solution (10 mM NaPO., (pH S.0), 150 mM NaCI, 1 mM MgCI,, 5 mM K.,Fe(CN),;, 5 mM K.,Fe(CN),;, and 0.2:; ?(-Gal) and stained for periods ranging from overnight to 2 days at room temperature. Stained embryos were observed by Nomarski interference optics on an inverted Nikon Diaphot microscope and photographed on Kodak Gold 100 film (ASA 100).

RESULTS
Two LpSl Genes-In previous work we characterized an LpSl cDNA clone isolated from an L. pictus gastrula cDNA library using an oligonucleotide probe corresponding to a consensus helix-loop-helix motif (Xiang et al., 1988). Three EcoRI fragments of 0.52, 0.44, and 1.1 kb, generated from the cDNA insert, represented the 5"untranslated leader sequence plus the first half of the protein coding sequence, the second half of the protein sequence, and the 3' end of the coding sequence plus the 3"untranslated trailer sequence, respectively. Preliminary experiments using these EcoRI fragments as probes for genomic Southern blot analysis suggested that L. pictus contained only a single LpSl gene unlike the multiple Spec gene family in S. purpuratus .
To determine the number of LpSl genes more precisely, we performed a series of quantitative slot blot and genomic Southern blot experiments with the EcoRI fragments of LpS1. Slot blot analysis was consistent with the notion of one LpSl gene but could not eliminate the possibility of a second (data not shown). However, genomic Southern blots with the LpSl 1.1-kb EcoRI fragment clearly indicated a second LpSl gene (Fig. 1). The 1.1-kb EcoRI fragment contained the 3"untranslated trailer sequence and corresponded to the final exon of the LpSl gene, based on comparison with the Spec genes of S. purpuratus. Below we demonstrate this also to be the case with genomic LpSl clones. The number of hybridizing bands therefore indicates the number of 3' LpSl exons in the genome.
Sperm DNA from two individuals was probed with the 1.1kb EcoRI fragment. In a representative blot, all lanes from both individuals showed two hybridizing bands, one strong and one weak, with the exception of the lane labeled X in individual 1 (Fig. 1). The simplest interpretation is that the strong band corresponds to the LpSl gene from which the cDNA clone was derived and the weaker band to a second LpSl gene. For clarity, we call these two genes LpSla and LpSlP, respectively. In lane X of individual 1, a restriction fragment length polymorphism can be seen in the LpSla gene. This individual appeared to be heterozygous at the L p S l a locus; one allele contained the 3"untranslated trailer exon within an 8.5-kb XbaI fragment, and a second allele, not present in individual 2, contained the exon within a 6.0-kb XbaI fragment. Both the 8.5-and 6.0-kb XbaI fragments were cut by HindIII to produce the 4.5-kb HindIII-Xbal fragment visualized in the X/H lanes of both individuals (Fig. 1). The experiments shown in Fig. 1 strongly implied the existence of two LpSl genes.
We used pulsed field gel electrophoresis to determine whether the LpSla and LpSlP genes were physically linked. Sperm DNA from three individuals was digested with either SalI, NotI, or SfiI and subjected to pulsed field gel electrophoresis. Hybridization with the 0.52-kb EcoRI fragment of the LpSl cDNA clone, which represents the 5'-half of the cDNA, showed two or three hybridizing bands in all digestions, ranging in size from 40 to 450 kb (Fig. 2). These results are consistent with the presence of two polymorphic LpSl genes in the L. pictus genome. From the intensity of the hybridizing fragments, we could predict which contained the LpSla or LpSlB gene (Fig. 2). However, since we did not observe both genes on a single SalI, NotI, or SfiI fragment, it was not possible to determine conclusively whether the genes were linked. NotI and SfiI are enzymes with rare cutting sites appearing on an average of every 200-300 kb in the L. pictus genome (data not shown). The probability of both NotI and SfiI cutting between two linked LpSl genes is low, and the fact that they did suggested that the LpSla and LpSlP genes were not closely linked.
Partial Structures of the LpSla and LpSlg Genes-We used the 0.52-, 0.44-, and 1.1-kb EcoRI fragments to screen an L. pictus genomic library for the LpSl genes. DNA isolated from the positive phage was mapped with several restriction enzymes, and genomic fragments hybridizing to different regions of the LpSl cDNA were subcloned and sequenced using oligonucleotide primers generated from the LpSl cDNA sequence. The genomic clones fell into three nonoverlapping regions as shown in Fig. 3. Two regions corresponded to the 5'-and 3'-most ends of the LpSla gene, judging by sequence identity between the LpSl cDNA clone and the genomic fragments (data not shown; Xiang et al. (1988)). The third group of genomic clones corresponded to all but the final two exons of the second gene, LpSlp. Its sequence was different from LpSla (see below). Despite repeated attempts, we were unable to isolate the middle portion of the LpSla gene or the 3' exons of the LpSlp genes. Their putative positions are shown parenthetically in Fig. 3. Several noticeable features emerged from analysis of the LpSla and LpSlp gene structures. Both genes contained two exon Is, and the downstream exon 1s (labeled 1* in Fig. 3) were truncated by six codons at their 3' ends, allowing each LpSl gene to encode two slightly different proteins by differential promoter utilization. The LpSlp gene, for which most of the reading frame exons have been cloned, contained exons 2-5 and shared 93% sequence identity with the 5' half of the LpSla cDNA clone. The LpSlp gene had similar exon/intron junctions to those found in the S. purpuratus Spec genes (Hardin et al., 1985(Hardin et al., , 1988. Following these exons were exons corresponding to the second half of the LpSl cDNA. We had previously shown that the LpSl cDNA contained a duplicated reading frame encoding a protein with twice the molecular weight of the Spec proteins (Xiang et al., 1988). Because these latter exons have sequence similarity with exons 2-5 and because they have similar exon/intron junctions, we labeled them 2', 3', 4', and 5' (exon 5' sequence was obtained from the LpSla gene) to indicate that they were duplicated versions of exons 2, 3, 4, and 5 , respectively (Fig. 3). It can be readily seen from the partial structures of LpSla and LpSlp that the LpSl duplication event was the result of an internal duplication of coding exons 2-5.
A partial sequence of the LpSlp gene is shown in Fig. 4A. In most genes belonging to the calmodulin superfamily, including the S. purpuratus Spec genes, the first exon ends with the initiator methionine codon (Hardin et al., 1985;1988;Perret et al., 1988). However, in the LpSla and LpSlp genes, all four first exons contain coding sequence beyond the AUG codon. The upstream exon 1s of both LpSla and LpSlp contained codons that matched the LpSla cDNA clone precisely (Fig. 4 B ) , while the downstream exon Is were truncated at the second codon, a serine. If LpSl proteins were generated from these downstream exon Is, they would be missing amino acid residues 3-8. Fig. 4B shows the predicted amino acid sequence for LpSla (derived from the cDNA clone; Xiang et al. (1988)) and LpSlp (derived from the genomic sequence). Although the proteins were obviously similar, it is interesting to note that the most significant divergence existed outside of the calcium binding helix-loop-helix domains (underlined in Fig. 4 B ) .
In order to map the 5' ends of the LpSla and LpSlp genes, we generated an oligonucleotide primer within the first exon (bracketed in Fig. 4 A ) and performed a primer extension analysis. Two major extended fragments could be observed, 105 and 106 bp in length (Fig. 5 ) . The last base of the longer product corresponded to an A residue 99 bp from the initiator codon, and this base was arbitrarily assigned +1 on Fig. 4A.
Since there were two exon Is for both LpSl genes and all four of these exons have nearly identical sequences, we could not distinguish among them with the primer extension technique. Sequences upstream of the 5' end of all four exons showed more than 99% identity for 326 bp; then exons 1 and 1* of the LpSla gene diverged completely from exons 1 and 1* of the LpSlp gene. In the latter gene, sequences upstream from exons 1 and 1* were nearly the same for 762 bp from their 5' ends, and then they diverged. Thus, it was possible that transcriptional elements sufficient for appropriate temporal and spatial expression laid upstream of both first exons of both LpSl genes. The 5"flanking sequences contained a putative TATA box at -17 to -22 bp and a USF binding site at -526 to -531 bp (Fig. 4 A ) . The USF site was found to bind USF specifically in uitro, albeit with weaker affinity than the corresponding USF sites within the Spec gene promoter^.^ A comparison of 5"flanking DNA among the LpSl and Spec genes showed no other convincing similarities. In particular, there was no homology with the S. purpuratus RSR element, a highly conserved 600-bp region flanking the Spec genes and containing an enhancer-like element within it.     I I I I I I I I I I I I I I I I I I I I I  I1  1  1 I I I I I I I I I l I I I 1 1 1 1 I I I I1 I l l I I I I l I l I I I

LPS1c L T N A E H S Q 1 L N R N L P G Q Y S E~1 l N E M I S R V D L N D D G R V Q~G E~L~A Q~L S K~D l K~Q F~l D K D X N G I ( l
gene on the pSVo plasmid. The LpSlP-CAT plasmid was injected into L. pictus or L. variegatus eggs and CAT activity monitored at different developmental stages. We have shown previously that CAT activity resulting from pSVo is virtually absent in L. pictus and L. variegatus eggs and embryos without appropriate promoter elements (Gan et al.,199Ob). LpSlP-CAT activity produced in hatched blastulae and later stage embryos of L. pictus was quantitatively comparable with what we had found previously with Spec promoters (Fig. 6; Gan et al. (199Ob)). Less than one-tenth of this activity could be seen when the LpSl fragment was inserted into pSVo in the reverse orientation (data not shown). Moreover, no detectable activity was observed with cleavage stage embryos. Though the amount of plasmid DNA was not as high in cleavage stage embryos as in later stage embryos, owing to less DNA amplification, sufficient DNA was present that any expression should have been detected. In the experiment shown in Fig.  6, the ratio of LpSlP-CAT DNA at the hatched blastula stage to that in the cleavage stage was about 10. Assuming that the levels of DNA in these experiments were nonsaturating and thus proportional to CAT activity, we should have observed at the cleavage stage 10% of the level of CAT activity of the hatched blastula stage. However, we could detect no more than 0.1% (Fig. 6A). In Fig. 6B, it can be seen that CAT activity remained fairly constant through the remainder of L. pictus embryogenesis. Similar results were obtained with L. uariegatus embryos (data not shown). Earlier work using nuclear run-on analysis has shown that the LpSl genes are activated at the end of the cleavage stage and remain active . The results presented in Fig. 6 were in good agreement with these nuclear run-on experiments and suggested that the LpSlP-CAT plasmid behaved appropriately with regard to temporal activation.
To monitor the embryonic cell types capable of utilizing the LpSlP promoter elements, we fused either the same DNA fragment (-762 to +17 bp) or one containing from about -3700 to +17 bp to lacZ. Injections were performed on L. pictus or L. variegatus fertilized eggs, and late gastrula/early prism stage embryos were harvested and stained for P-galactosidase activity with X-Gal. It should be noted that the injected DNA was incorporated into only a fraction of the cleaving nuclei during development, and although the incor-i E FIG. 5. Determination of the 5' end of the LpSl genes. The primer was radiolabeled at its 5' end with "P and hybridized with 20 pg of total ectodermal RNA isolated from pluteus stage embryos. The products of primer extension (I") and sequencing reactions (G, A , T, C ) obtained from the same primer were analyzed on a n 8% sequencing gel as described under "Experimental Procedures." Shown on the left is the DNA sequence surrounding the base (indicated by the asterisk) corresponding to the 106-bp major primer extension product. The complementary sequence is shown in poration event is random with respect to cell type, the DNA is distributed in nonuniform patches (Hough-Evans et al., 1988). We observed preferential aboral ectoderm staining with both the -762 and -3700 bp LpSlp promoter constructs. Examples of some L. uariegutus embryos are shown in Fig. 7. We routinely found larger patches of staining with the -3700 bp construct (Fig. 7, A-C) than with the -762 bp construct ( Fig. 7, D-F), although it was not clear why. We analyzed 129 stained embryos injected with the -3700 bp plasmid and 239 stained embryos with the -762 bp plasmid (Table I), and more than 90% of the embryos showed only aboral ectoderm labeling. We conclude that both constructs were capable of conferring proper aboral ectoderm expression and that a minimum of 762 bp of 5"flanking DNA plus 17 bp of 5"untranslated leader sequence was sufficient for the proper temporal activation and spatial expression of the LpSlp gene.
We generated two 5' deletions to determine whether smaller regions of 5'-flanking DNA would allow for the same levels of CAT activity and the same cell type specificity of lac2 expression. Fig. 8 shows the results of deleting an additional 251 bp (-511 deletion) or 394 bp (-368 bp deletion) from the parental -762 bp LpS10-CAT plasmid. The -511 bp deletion resulted in a 3-4-fold drop in CAT activity (Fig. 8, A and B ) , but further deletion to -368 bp showed no additional loss of activity; in fact, a slight increase was observed in two experiments (Fig. 8, A and B ) .
We have analyzed the region between -762 and -511 bp for DNA-protein interactions by conventional band shift analysis and methylation interference protection assays. One prominent element was the USF binding site at -531 to -526 bp discussed above. Site-directed mutagenesis of this USF site by inserting a Hind111 linker within the . . . CACGTG . . . sequence in the -762 bp LpSlp-CAT plasmid did not affect CAT activity: Thus, the loss of the USF site could not explain the 3-4-fold drop in CAT activity observed in the -511 bp deletion. A second region at -726 to -721 bp contained a stretch of C residues (Fig. 4A) that was a high affinity binding site for an unidentified L. pictus nuclear protein (data not shown). This C stretch was repeated in an inverted orientation (i.e. a G stretch) at -75 to -70 bp (Fig. 4A). We have not yet tested whether mutating these regions affects the CAT activity of the -762 bp LpSlP-CAT plasmid.
We also tested whether the drop in CAT activity with the -511 or -368 bp deletions correlated with an effect on spatial expression by cloning these deletions into the lac2 reporter gene. Although aboral ectoderm labeling was still preferred with both constructs, strict aboral ectoderm expression was lost. Staining of nonaboral ectoderm cells, particularly primary and secondary mesenchyme cells, was observed with both the -511 and -368 bp lac2 constructs, but endoderm and oral ectoderm were rarely stained (Fig. 9). Aboral ectoderm labeling was preferred with both constructs; about 70% of the stained embryos had only aboral ectoderm cells labeled (Table I). If random, no more than 40% of the stained embryos would be labeled solely in aboral ectoderm. Since both plasmid constructs yielded identical results, we conclude that at least one spatial element exists between -762 and -511 bp that prevents expression of the LpS1B gene in primary and secondary mesenchyme cells.

DISCUSSION
The LpSl Gene Pair of L. pictw-We have presented evidence that there are two closely related LpSl genes in L. pictus that could encode up to four distinct LpSl proteins. The genes do not appear to be closely linked, and each gene has the exon order 1, 1*, 2, 3, 4, 5, 2', 3', 4') 5') 6, implying that an internal duplication has occurred. In addition, we have shown that 762 bp of 5'-flanking DNA from the LpSlp gene plus 17 bp of 5"untranslated leader sequence were sufficient for activating reporter genes at the appropriate developmental time and in aboral ectoderm cells.
The principal reason for characterizing the LpSl genes was to make comparisons with the S. purpuratus Spec genes in the hopes of identifying conserved regions important for expression and function. There are seven or eight Spec genes, and four of them (Spec 1, 2a, 2c, and 2d) have been cloned and characterized (Hardin et al., 1985(Hardin et al., , 1988. Spec 2a and 2c are the most similar in sequence, and Spec 1 is more similar to them than is Spec 2d (Carpenter et al., 1984;Moncrief et al., 1990). The highly diverged Spec 2d is only weakly expressed and appears to encode a nonfunctional product (Hardin et al., 1988;Klein et al., 1991). All four genes have an 800-bp conserved sequence block, designated RSRA, upstream of their translational start sites, we have  56 24 67 f 6.6 (n = 2) 'Embryos were scored as correct spatial expression if only aboral ectoderm cells were stained.
bEmbryos were scored as incorrect if unambiguous staining was seen in cell types other than aboral ectoderm, regardless of whether aboral ectoderm cells were stained or not.
'If applicable, the average value is expressed with a standard deviation. n is the number of experiments.
suggested it contains cis elements important for Spec gene activation (Gan et al.,199Ob). In Spec 2a, this block is continuous but in Spec 1, 2c, and 2d, inserted DNA interrupts the conserved block, displacing the RSR portion and parts of the A region a few kilobases upstream (Gan et al.,199Ob). The 5' half of the RSR region contains a transcriptional element that appears to be a temporal enhancer, and the A region contains the above mentioned USF binding site, which binds sea urchin USF with high affinity (Gan et al., 1990b;Tomlinson et ale, 1990). We have shown USF activity is present only in ectoderm cells but have not yet demonstrated a role for USF in the transcriptional activation of the Spec genes (Tomlinson et aL, 1990). The two LpSl genes have diverged in sequence from the Spec genes. Sequence comparisons within the protein-coding regions show only weak similarity, mostly in the calcium binding domains (Xiang et al., 1988). In addition, the 5'flanking and 5'-and 3"untranslated regions are totally dissimilar. Within the limits of the sequence analysis, LpSla or LpSlP are no more closely related to one Spec gene than to another, suggesting that the LpSl gene pair in L. pictus arose separately from the Spec gene family. We hypothesize that a common echinoid ancestor to S. purpuratus and L. pictus contained a single Spec-like gene. Most likely, this gene arose from a duplication event of an ancestral calmodulin gene, since calmodulin is the most ancient member of the superfam- ily yet shares a similar exon/intron structure with the Spec genes. This ancestral Spec gene somehow obtained or evolved the regulatory elements required for aboral ectoderm expression. Sometime after the branching of the Strongylocentrotus and Lytechinus genera, subsequent duplications produced the Spec gene family and LpSl gene pair. However, an alternative view, difficult to rule out, is that the Spec and LpSl genes did not have a shared ancestral Spec-like gene, but that each family arose and acquired the appropriate regulatory elements independently. Though independent acquisition of aboral ectoderm specificity would seem a less probable scenario, the A actions, approximately 10-15 for each gene.5 In several instances the same proteins appear to be interacting with fragments from both genes. These latter proteins are likely candidates for transcription factors specifically involved in controlling the expression of the Spec and LpSl genes in aboral ectoderm cells. However, only after many of the specific cis elements and trans factors are precisely defined will a picture emerge regarding the relatedness of the two gene promoter regions. Ultimately, comparisons of the regulatory features between the Spec 2a and LpSlP genes required for aboral ectoderm specificity should provide useful information on how cis regulatory elements on ancestrally related genes have changed over evolutionary time and how trans factors have evolved in response to these changes.