Introduction ORIGINAL ARTICLE. selor&:jessica C. Kissinger Rudolf A. Raff

Transcription

1 Dev Genes Evol (1998) 208:82 93 Springer-Verlag 1998 ORIGINAL ARTICLE selor&:jessica C. Kissinger Rudolf A. Raff Evolutionary changes in sites and timing of actin gene expression in embryos of the direct- and indirect-developing sea urchins, Heliocidaris erythrogramma and H. tuberculata csim&:received: 27 July 1997 / Accepted: 30 December p&:Abstract We describe an evolutionary comparison of expression of the actin gene families of two congeneric sea urchins. Heliocidaris tuberculata develops indirectly via a planktonic feeding pluteus that forms a juvenile rudiment after a long period of larval development. H. erythrogramma is a direct developer that initiates formation of a juvenile rudiment immediately following gastrulation. The developmental expression of each actin isoform of both species was determined by in situ hybridization. The observed expression patterns are compared with known expression patterns in a related indirect-developing sea urchin, Strongylocentrotus purpuratus. Comparisons reveal unexpected patterns of conserved and divergent expression. Cytoplasmic actin, Cy- III, is expressed in the aboral ectoderm cells of the indirect developers, but is an unexpressed pseudogene in H. erythrogramma, which lacks aboral ectoderm. This change is correlated with developmental mode. Two CyII actins are expressed in S. purpuratus, and one in H. erythrogramma, but no CyII is expressed in H. tuberculata despite its great developmental similarity to S. purpuratus. CyI expression differs slightly between Heliocidaris and Strongylocentrotus with more ectodermal expression in Heliocidaris. Evolutionary changes in actin gene expression reflect both evolution of developmental mode as well as a surprising flexibility in gene expression within a developmental mode. dwk&:key words Actin gene family Sea urchins Molecular evolution Gene expression&bdy: Edited by D. Tautz J.C. Kissinger 1 Rudolf A. Raff ( ) Indiana Molecular Biology Institute, and Department of Biology, Indiana University, Bloomington, IN 47405, USA Present address: 1 Laboratory of Parasitic Diseases, National Institute of Allergy and Infectious Disease, NIH, Bethesda, MD , USA&/fn-block: Introduction Actin genes display unique temporal and spatial patterns of expression in sea urchin embryos (Cox et al. 1986; Wang et al. 1994; Fang and Brandhorst 1996). In the sea urchin Strongylocentrotus purpuratus, it has been shown that the expression of different cytoplasmic actin isoforms is tightly linked to cell lineage and that this linkage may result from functional requirements for specific isoforms (Cox et al. 1986; Cameron et al. 1989; Davidson et al. 1985; Davidson 1989). We undertook a study of the actin gene family of Heliocidaris erythrogramma, a direct-developing sea urchin to help us understand evolutionary transformations in morphology and cell fates in this highly modified embryo. This species can be compared to its congener, H. tuberculata, which exhibits standard indirect development via a pluteus larva. Indirect development is the ancestral and most common pattern of sea urchin development. It begins with a relatively small egg that cleaves and proceeds through gastrulation to form a feeding pluteus larva. This larva feeds for several weeks, while an adult rudiment forms within it, and then settles and metamorphoses into a juvenile urchin. However, about 20% of sea urchins, including H. erythrogramma, have independently evolved direct development and do not form feeding larvae. In these species, a large egg cleaves and undergoes a truncated gastrulation. Subsequently, instead of forming functional larval feeding structures, the larva subsists on egg stores and begins the production of a miniature adult that will metamorphose in as few as 4 days. Intermediate forms between direct and indirect development also exist. These species may exhibit larger egg size, facultative larval feeding capacity, partial loss of larval features, and variation in time to metamorphosis (Emlet 1990; Wray and Raff 1991; Raff 1996). H. tuberculata and H. erythrogramma diverged from each other about 10 million years ago and from S. purpuratus about million years ago (Smith et al. 1990; McMillan et al. 1992) (Fig. 1). As cleavage, cell lineages, cell fate, and morphogenesis are all substantially modified

2 83 Fig. 1 Inferred relationships among camarodont sea urchins (shown as embryos) and actin family members (after Kissinger et al. 1997). Phylogeny and times of divergences are based on Raff et al. (1988), Smith et al. (1990), McMillan et al. (1992), and Littlewood and Smith (1995). The known actin genes for each species are indicated above the embryos, suspected missing members are identified by a?. The Lytechinus pictus gene designations of Fang and Brandhorst (1994) are given, along with their inferred identities in the nomenclature of Lee et al. (1984) in parentheses (CyU unique cytoplasmic actin). Evolutionary changes are mapped on the phylogeny 1 CyI gene duplication and gene conversion among Lytechinus CyI 3 UTR s, 2 duplication of CyII and CyIII genes in Strongylocentrotus, 3 reduction in the number of cytoplasmic actin genes expressed in early development in Heliocidaris, 4 loss of CyII embryonic expression in H. tuberculata, 5 evolution of direct development, 6 loss of CyIII gene expression and conversion to pseudogene in H. erythrogramma, myr million years&/f :c.gi in H. erythrogramma (Raff 1996), we sought conserved cell-type gene markers to homologize regions of the H. erythrogramma embryo to that of indirect-developing sea urchins. Actins potentially offered such markers. In vertebrates actin isoforms and their 3 untranslated regions (UTR) are conserved, and the tissue-specific patterns of expression are conserved as well (Yaffe et al. 1985; Alonso et al. 1986; Rubenstein 1990). In the sea urchin S. purpuratus, the entire actin gene family has been cloned, and the expression patterns for each have been determined (Shott et al. 1984; Cox et al. 1986; Lee et al. 1986). The temporal and spatial expression patterns of each isoform are unique, and no single isoform is ubiquitous. We have characterized the Heliocidaris actin gene families (Hahn et al. 1995; Kissinger et al. 1997). The genomes of both species contain a muscle actin and one of each of the three cytoplasmic actin isoform types, but family representation differs. Using the isoform class names defined for S. purpuratus, H. erythrogramma has CyI and CyII cytoplasmic actin genes, and muscle actin, as well as a CyIII pseudogene, whereas H. tuberculata has CyI, CyII, CyIII and muscle actin genes (Fig. 1). Heliocidaris isoform-specific 3 untranslated region (UTR) probes were used to characterize the temporal and in situ expression pattern of each actin isoform in both H. erythrogramma and H. tuberculata. The results of this study confounded our original expectations of conservation. The sea urchin actin gene family has proven to be evolutionarily fluid over a relatively short time (Fang and Brandhorst 1994; Kissinger et al. 1997; Fig. 1). In this paper we show that actin isoform expression patterns in early development have also evolved rapidly. Thus, it is possible to test the functionality of actins by observing whether or not they have evolved in concert with the developmental changes that occurred in the their respective species. Changes relating to both phylogenetic branching and developmental mode have occurred. Materials and methods RNA gel blot analysis and 3 RACE Five micrograms of total RNA from developmental stages of both Heliocidaris species were loaded on a 0.8% agarose formaldehyde gel. RNA was transferred and fixed to Nytran (S&S) by UV crosslinking. The blots were hybridized sequentially, first with isoformspecific probes, and then with conserved coding region probes (Kissinger et al. 1997). Between hybridizations, the blots were stripped and probe removal was confirmed. Sea urchin culture and histology S. purpuratus were obtained from Marinus (Long Beach, Calif.); H. erythrogramma and H. tuberculata were collected off the coast of Sydney, Australia. Spawning was induced by intracoelomic injection of 0.55 M KCl. Embryos were cultured in ASW (artificial

3 84 Fig. 2A F Comparison of Heliocidaris developmental modes. A C H. tuberculata development; D F H. erythrogramma development. A H. tuberculata mid-gastrula: an almost fully extended archenteron has formed. Secondary mesenchyme cells form from the tip of the elongating archenteron. Primary mesenchyme cells (pmc) have begun to construct the skeleton (sk). D H. erythrogramma gastrula: invagination of the vegetal plate has produced the archenteron (ar). The mesenchyme cells (mc) have begun to migrate along the interior wall of the blastocoel (bc). B E Divergent later stage embryos of the same age. B Lateral view of H. tuberculata late prism: the skeleton is extending, and the embryo takes on a more pyramidal shape as skeletal arm rods elongate. The gut begins to differentiate into foregut (fg), stomach (st), and hindgut (hg). E H. erythrogramma late gastrula: the embryo is beginning to elongate and form structures of the juvenile rudiment. Both coelomic pouches have formed (lc and rc), and the opening at the base of the archenteron has closed. The ectoderm opposite the left coelomic pouch begins to thicken and bulge inward, eventually lining the vestibule. C A simplified view of a fully formed pluteus that has begun rudiment formation. Two additional pairs of arms will form later in pluteus development. The juvenile rudiment (jr) forms from the left coelomic (at the eight-armed pluteus stage). F An H. erythrogramma larva (approx. 53 h in age). The coelomic cavities are visible, tube feet (tf) have formed by interaction of vestibule floor and the hydrocoel (hy), which was derived from the left coelom. The embryo has elongated creating a larger blastocoel (bc). Internal structures include the stomach (st), left and right coelomes (lc and rc), and skeletal elements (sk). The embryos in D, E, and F have not been rotated and are shown in the same orientation [abec aboral ectoderm, ar archenteron, bc blastocoel, ec ectoderm, fg foregut, hg hindgut, jr juvenile rudiment, lc left coelomic pouch, lec left side ectoderm, mc mesenchyme cell, oec oral ectoderm, pmc primary mesenchyme cell, rc right coelomic pouch, rec right ectoderm, sk skeletal rods (fenestrated adulttype skeleton in F), smc secondary mesenchyme cell, st stomach, tf tube foot] (For detailed descriptions of pluteus, rudiment and H. erythrogramma see von Ubisch 1913; Hyman 1955; Okazaki 1975; Williams and Anderson 1975)&/f :c.gi sea water), or filtered sea water (Heliocidaris spp.) with gentle stirring at 14 C (S. purpuratus) or C (Heliocidaris). Embryos were fixed by one of two methods: 1% glutaraldehyde, following Angerer and Angerer (1991), or 2% glutaraldehyde and a longer fixation time. Embryos were dehydrated and stored at 4 C. Embryos were embedded as described in Angerer and Angerer (1991), and stored at 4 C until use. Embryos were sectioned to a thickness of 6 µm and ribbons were floated onto TESPA (3-aminopropyltriethoxysilane)-coated slides. RNA probe preparation and quantitation RNA probes were synthesized from linearized templates containing cloned isoform-specific 3 UTR or fourth exon coding region

4 85 sequences ( bp; Kissinger et al. 1997), using the Maxiscript kit (Ambion). Probes were labeled with a 33 P UTP to a specific activity of Template DNA was removed by DNase digestion and incorporation was monitored by trichloroacetic acid (TCA) precipitation. Probes were hydrolyzed to an average size of 150 bp (Angerer and Angerer 1991). Test hydrolyses were performed and time points were separated on native glycerol/acrylamide gels and analyzed by autoradiography to confirm probe length. Hybridization and washing conditions Paraffin sections were rehydrated, pretreated, hybridized, washed, stained and mounted (Angerer and Angerer 1991), with the following changes. Paraffin sections were treated with three complete changes of xylene, rehydrated through an ethanol series and then treated with the appropriate amount of Proteinase K for their level of fixation: 1.8 mg/ml for embryos fixed with 2% glutaraldehyde, 0.1 mg/ml for embryos fixed with 1%. Levels were empirically determined with a proteinase K concentration series experiment and an actin coding region antisense probe. A saturating concentration of probe, 0.3 µg/ml/kilobase of probe sequence complexity, was added to each slide. Hybridization temperature was stringent at 47 C and the length of hybridization varied from h. All post-hybridization washes (2 SSC) were performed using a volume of 1 l/25 50 slides, and the final wash was performed in 1 l 0.5 SSC at 50 C (20 SSC = 3.3 M sodium chloride and 0.3 M sodium citrate) In situ data capture Slides were coated with photographic emulsion and stored at 4 C in light-proof containers in the presence of silicone desiccant packets. Optimal exposure times ranged from 6 14 days for HeM, HeCyI, HtM and HtCyIII, and from days for HeCyII and HtCyI probes. Slides were stained with eosin and toluidine blue during the final ethanol dehydration series. Acid-washed coverslips were put in place with Permount for viewing and permanent storage. Bright and darkfield microscopic images were viewed on a Zeiss Axioplan microscope equipped with a Sony 3ccd digital camera. Images were captured with the RasterOps 24XLTV video capture system. No adjustments were made to collected images other than scaling, contrast, brightness and JPEG data compression. In the in situ figures, the majority of the sections are presented such that they are comparable to Fig. 2. A few panels are crosssections through the dorsal-ventral axis of the embryo as indicated by line B in Fig. 2F, or are en face sections through the embryo that look in from the vestibular cavity as illustrated by line C in Fig. 2F. Results Timing of actin expression Figure 3 shows the temporal pattern of actin expression in embryos of both Heliocidaris species. Maternal CyI mrna in H. erythrogramma eggs is barely detectable upon long exposure (data not shown; Hahn and Raff 1996). In contrast, there is a low level of CyI mrna in the egg of H. tuberculata (Hahn and Raff 1996), as is also reported for S. purpuratus (Shott et al. 1984). HeCyII mrna is present in unfertilized egg mrna in H. erythrogramma, (also recovered from a cdna library made from unfertilized eggs), and continues through the latest stages of H. erythrogramma examined (Kissinger 1995). HtCyII actin gene expression was not detected even after long exposures. Fig. 3 RNA gel blot analysis of actin expression during Heliocidaris development. Total RNA isolated from embryos was electrophoresed, transfered to Nytran (S&S) and sucessively probed with the indicated isoform-specific 3 untranslated region (UTR) probes. The H. erythrogramma stages are: unfertilized egg (lane 1), 13.5-h blastula (lane 2), 18-h gastrula (lane 3), 26.5-h gastrula (lane 4), 30-h early larva (lane 5), 36-h larva (lane 6), 53-h larva (lane 7), 73-h late larva (lane 8), and 89-h metamorphosing larva/juvenile (lane 9). H. tuberculata stages are: unfertilized egg (lane 1), morula (lane 2), mesenchyme blastula (lane 3), early gastrula (lane 4), mid-gastrula (lane 5), late gastrula/early prism (lane 6), and pluteus (lane 7). A coding sequence probe is also shown for H. tuberculata to better illustrate the two different actin transcript sizes. Arrows indicate the positions of ribosomal bands. Reduced levels of actin expression are seen in lane 8 for all isoforms. This reduction results from some, as of yet poorly understood, pre-metamorphosis phenomenon in which all mrna levels appear to be transiently reduced, and mrna is no longer Poly A+ selectable in H. erythrogramma (Popodi and Raff, unpublished)&/f :c.gi CyIII expression in H. tuberculata is first detected in the early gastrula and it is seen through the pluteus stage. CyIII in H. erythrogramma is a pseudogene with a stop codon inserted in the coding region, and is not detectably expressed (Kissinger et al. 1997). Muscle actin expression is first detectable in the pluteus stage of H. tuberculata, coordinately with development of larval pharyngeal muscles. Expression in H. erythrogramma is first detectable around 53 h when the juvenile tube feet are forming and peaks around 89 h, just prior to metamorphosis. Transcripts of two different lengths are detected in H. tuberculata, approximately 2.2 (HtCyI and HtM) and 1.9 kb (HtCyIII; Fig. 3). In S. purpuratus, the CyIIIa transcript is also shorter than other actin transcripts, 1.8 as opposed to 2.1 kb. This reflects differences in lengths of the 3 UTR (Flytzanis et al. 1989). In H. erythrogramma, all actin transcripts are kb in length. In situ patterns of actin gene expression Heliocidaris embryos have proven to be refactory to whole-mount in situ techniques. For this reason, we

5 86 Fig. 4A J In situ CyI, CyIII and muscle actin expression in Heliocidaris tuberculata. A J H. tuberculata embryos probed with sense or anti-sense actin probes. The upper rows show the images in brightfield viewed with DIC optics. The lower rows show the same images in darkfield to illuminate the silver grains. A D CyI anti-sense probe; G CyI sense probe; E F CyIII anti-sense probe; and H J muscle anti-sense probe. The age and exposure time for the embryos in each panel are: A 23-h early gastrula, 30 day exposure; B 29-h late gastrula, 30 day exposure; C 36-h early prism, 30 day exposure; D 86-h pluteus, 20 day exposure; E 71-h pluteus, 18 day exposure; F 71-day pluteus, 18 day exposure; G mixed developmental stages, 30 day exposure; H 43-h prism, 14 day exposure; I 71-h pluteus, 14 day exposure; J 71-h pluteus, 14 day exposure (cp coelomic pouch, scale bars 50 µm)&/f :c.gi

6 87 Fig. 5A F Localization of CyI actin mrna in H. erythrogramma. A F Developmental series of H. erythrogramma sections hybridized with a CyI 3 UTR anti-sense probe. Age and exposure time for the embryos in each panel are: A 18-h mesenchyme blastula, 22 day exposure; B 21.5-h gastrula, 22 day exposure; C 24-h mid-gastrula, 20 day exposure; D 36-h larva, 12 day exposure; E 36-h larva, 12 day exposure; F 94-h pre-metamorphosis larva, 22 day exposure. B E are oriented such that the vegetal end of the embryo is to the right, and the presumptive oral side of the juvenile is facing down. White arrows indicate areas of increased or decreased signal referred to in the text (lcp left coelomic pouch, rcp right coelomic pouch, g gut, ve vestibule ectoderm, scale bars 50 µm)&/f :c.gi chose to employ an in situ technique that used 33 P and serially-sectioned embryos. Three criteria were used to judge the reliability of in situ patterns: (1) the pattern had to correspond to the temporal pattern detected by RNA gel blot analyses; (2) the pattern had to be reproducible with a different batch of embryos; and (3) the pattern had to be different from the signal (if any) observed with the corresponding sense probe. In most cases, several stages of embryos were hybridized on the same slide with the same probe, or on different slides with aliquots of the same probe. CyI expression CyI actin mrna is observed in all cells of H. tuberculata by the mesenchyme blastula stage, with the strongest signal observed in secondary mesenchyme cells (Fig. 4A). As development proceeds, the CyI signal slowly fades from ectoderm, but remains strong in mesenchyme, especially secondary mesenchyme (Fig. 4B). CyI signal is strong in the archenteron by the early prism stage and faint in the presumptive oral ectoderm (Fig. 4C). In the

7 88 Fig. 6A G Localization of CyII mrna in H. erythrogramma. A G Developmental series of H. erythrogramma sections hybridized with a CyII 3 UTR anti-sense probe. Age and exposure time in each panel are: A unfertilized egg, 40 day exposure; B 18-h mesenchyme blastula, 40 day exposure; C 21.5-h gastrula, 40 day exposure; D 32.5-h early larva, 20 day exposure (viewed in plane B as shown in Fig. 2F); E 32.5-h early larva, 20 day exposure; F 44-h larva, 40 day exposure; G 55-h larva, 20 day exposure. Embryos in C and E are oriented as in Fig. 4. (Scale bars 50 µm)&/f :c.gi pluteus, CyI is expressed in oral ectoderm, esophagus and weakly in the stomach. There is no aboral or hindgut signal, and the oral-aboral ectodermal boundary in the ectoderm is clearly seen at the base of the pluteus arm (Fig. 4D). Cells of the aboral ectoderm are squamous epithelium, whereas those of the oral ectoderm are cuboidal (Cameron et al. 1994; Fig. 4D F). No signal is detected with the CyI sense strand probe (Fig. 4G). In H. erythrogramma, CyI mrna is first detected in the vegetal plate of 15.5-h-old mesenchyme blastula stage embryos (Kissinger 1995) and by 18 h the signal is

8 89 Fig. 7A D Localization of muscle actin expression in H. erythrogramma. A D Late developmental series of H. erythrogramma sections embryos hybridized with a muscle actin 3 UTR anti-sense probe. The left column shows the brightfield images viewed with DIC optics, right column same images viewed in darkfield. The exposure time for all images is 14 days. The age of the embryos in each panel is: A 55- h larva; B 69-h late larva; C 94- h pre-metamorphosis larva; D 94-h pre-metamorphosis larva. The embryos in A and C are oriented with the oral side down. In B and D the oral side of the embryo is facing the reader and the vegetal side is to the right (viewed in plane C of Fig. 2F; dlm developing lantern and lantern muscles, ec ectoderm, es esophagus, mxc M actin-expressing coelom cells, scale bars 50 µm)&/f :c.gi concentrated on the periphery but not the center of the gastrula vegetal plate (Fig. 5A). CyI is strongly expressed in all mesenchyme cells (Fig. 5B, C), reaching a plateau around 36 h of development. At this time, CyI mrna is detected weakly in the ectoderm of the embryo, except in the vegetal plate and one other small ectodermal patch, which may represent the cells that will form the madroporite on the future dorsal side (Fig. 5D). There is little or no signal in the archenteron until 32 h of development, when a strong signal is seen in this region (Figs. 2E, 5D, E). A low level of CyI mrna is detected in the coelomic pouches that will form part of the developing juvenile (Fig. 5D, E). CyI is detected in the ectoderm of the tube feet by 55 h of development and from then onward (Figs. 2F, 5F). Prevalent expression of CyI continues in the mesenchyme of the blastocoel and in the area where the madroporite is forming on the dorsal/aboral surface of the embryo (Fig. 5F). CyII expression Although the CyII gene is present in the H. tuberculata genome (Kissinger et al. 1997), no expression was observed even under low stringency conditions. CyII mrna is detectable in the H. erythrogramma egg (Fig. 6A). CyII expression continues during gastrulation, and is primarily expressed in mesenchyme cells (Fig. 6B, C). Around 30 h of development CyII is prevalent in the central region of the archenteron, and continues in this region, which will later form the gut of the juvenile (Fig. 6D G; Williams and Anderson 1975). No signal is detected in the coelomic pouches or developing rudiment. Small thick ectodermal patches on the dorsal vestibule region of the embryo show a CyII signal at h of development (Kissinger 1995). This same ectodermal region does not express CyI mrna during this time (Fig. 5D). This ectodermal signal may represent insertion of secondary mesenchyme cells that have differ-

9 90 entiated into pigment cells, but this has not been confirmed. CyIII expression CyIII mrna is detected in H. tuberculata, but not in H. erythrogramma. CyIII is first detected in presumptive aboral ectoderm by the start of gastrulation (Kissinger 1995) but the signal is not strong until the late gastrula stage. There is no signal in gut, mesenchyme, or oral ectoderm (Fig. 4E, F). Muscle actin expression Muscle actin mrna is first detected in H. tuberculata at the late prism stage before the arms have extended (thus between Fig. 2A and B), when the pharynx is forming. Signal is limited to endo-mesodermal cells at the tip of the archenteron which will give rise to the pharynx/foregut. Muscle actin mrna is detected before the cells have differentiated into muscle cells (Fig. 4H). After the gut fully differentiates, muscle actin mrna is seen in the foregut of the pluteus and in the surrounding coelomic pouches, but nowhere else (Fig. 4I, J). In H. erythrogramma, no muscle actin signal is detected until after 55 h of development when the tube feet and hydrocoel system are more fully formed (Fig. 7A). By 69 h of development, muscle actin mrna is prevalent in the mesodermal lining of the tube feet (Fig. 7B). These cells have not yet developed typical muscle cell morphology. By 94 h of development, a complex pattern of muscle actin expression is observed in the pre-metamorphosis larva. In cross-section (Figs. 2F plane B, and 7C), muscle actin mrna is seen in the tube feet, the hydrocoel, the esophagus, and in cells lining the coelom. The strong ovals of muscle actin mrna signal at each end of the hydrocoel mark the forming jaw muscles of the developing lantern. If a 94-h embryo is examined in a plane through the middle-oral side of the embryo (Fig. 2F, plane C), muscle actin mrna expression reveals the pentameral symmetry of the embryo and developing hydrocoel system (Fig. 7D). The esophagus is ringed by the hydrocoel canal which extends into each of the five primary podia. The circumference of the coelom expresses muscle actin. Discussion A number of significant changes have occurred in the actin gene family of Heliocidaris, making the use of actin expression patterns to homologize embryo regions difficult. Of the six functional actin genes present in S. purpuratus, only two, CyI and muscle, are expressed in both Heliocidaris species. The remaining members, CyII and CyIII, are complicated by gene duplications in the Strongylocentrotus lineage and gene loss (CyIII) or non- Fig. 8 Comparative summary of gastrula expression patterns by species and actin isoform. Mid-gastrula stage embryos are shown for each of three sea urchin species, S. purpuratus, H. tuberculata and H. erythrogramma.the actin isoform being compared is indicated on the left of each horizontal row. Stiples represent lower levels of expression than solid colors. S. purpuratus has two CyII genes, CyIIa is shown in purple and CyIIb is shown in green. S. purpuratus has two CyIII genes, CyIIIa and CyIIIb. At this stage of development their expression patterns are identical (shown in blue). Expression patterns for S. purpuratus are based on Cox et al. (1986)&/f :c.gi expression (CyII) in H. erythrogramma and H. tuberculata, respectively (Fig. 1). Cytoplasmic actin expression patterns for the three species at the mid-gastrula stage are summarized in Fig. 8. The indirect-developing species strongly express CyI in primary and secondary mesenchyme cells (Fig. 4B, C; Cox et al. 1984). H. erythrogramma expresses CyI in all mesenchyme cells, as do the others, even though no secondary mesenchyme cells are generated from the tip of the archenteron (Fig. 5B, C; Parks et al. 1988). Both Heliocidaris species, unlike S. purpuratus, have low level CyI ectodermal expression at this stage (Figs. 3B, 4B, C; Cox et al. 1984). In S. purpuratus, there are two CyII genes, CyIIa and CyIIb. CyIIb shares a CyI expression pattern and CyIIa is restricted to secondary mesenchyme (Cox et al. 1986). CyII is not expressed during H. tuberculata development. CyII expression in H. erythrogramma is restricted to a subset of the mesenchyme cells at this stage (Fig. 6C). CyIII is similarly expressed in presumptive aboral ectoderm in both S. purpuratus and H. tuberculata, but is absent from H. erythrogramma. As was the case with CyII, S. purpuratus has two isoforms, CyIIIa and CyIIIb (Lee et al. 1984; Akhurst et al. 1987; Flytzanis et al. 1989), but both isoforms have identical spatial expression in the gastrula (Cox et al. 1984). The CyIII expression patterns are

10 comparable in the prism stage of H. tuberculata, and remain so in the pluteus, with the exception of a shift downward in the aboral/oral ectoderm boundary toward the base of the arm in H. tuberculata. This boundary difference probably reflects differences in position of the dorsal-ventral axis of first cleavage in H. tuberculata relative to to that of S. purpuratus (Henry et al. 1992). We thus posed the following questions: 1. Do species that develop via the same mode, e. g. H. tuberculata and S. purpuratus, have conserved actin expression patterns despite their evolutionary distance, as among the vertebrate actin isoforms? 2. Is the evolution of a novel developmental pathway accompanied by new patterns of actin gene expression? 3. Is actin expression tied to specific cell lineages despite changes in morphology and cell fate? Lack of actin expression pattern conservation in species that develop via the same mode Each of the three indirect developers examined contains at least one gene related to CyII on the basis of sequence, linkage, or expression pattern. As was the case for CyIII, the 3 UTR s are quite divergent, especially the LpC3 gene of L. pictus. Two CyII isoforms are expressed in S. purpuratus. CyIIb appears to be expressed in the same pattern as CyI, but at lower levels (Cox et al. 1986). CyIIa expression in S. purpuratus is confined to secondary mesenchyme and gut, particularly the stomach and hindgut. In the pluteus, CyIIa mrna is detected in the coelomic pouches evaginated from the sides of the esophagus, and forms a crucial part of the adult rudiment (Cox et al. 1986). H. tuberculata also forms its coelomic mesoderm from an outpouching from the esophagus wall; however, H. tuberculata does not express CyII during embryonic development. L. pictus expresses a gene in presumed secondary mesenchyme cells, but this gene, LpC3, appears to be more closely related to CyI isoforms on the basis of phylogenetic analyses (Fang and Brandhorst 1994; Kissinger et al. 1997). Expression beyond the late gastrula has not been described for LpC3. CyI expression has been described in four different genera of indirect-developing sea urchins, S. purpuratus (Cox et al. 1986), H. tuberculata (this work), L. pictus (Fang and Brandhorst 1996), and Tripneustes gratilla (Wang et al. 1994; Fig. 1). Although the gene and its 3 UTR sequence are highly conserved among these species, sites of expression in the embryo vary. All species express CyI in the vegetal plate of the blastula, secondary mesenchyme at the tip of the archenteron, and in the oral ectoderm of the pluteus. The broadest expression pattern detected, in terms of numbers of cells and tissue types, was in T. gratilla. All the species express CyI in gut. However H. tuberculata differs from the others in that this expression is strongest in foregut and stomach (Fig. 4D), rather than in hindgut and stomach. Some generalizations are possible from these and our previous studies (Kissinger et al. 1997): 1. Muscle actin is conserved in gene sequence, copy number, and tissue-specific expression pattern. 2. CyIII actin varies in gene number, and its 3 UTR sequence is not conserved. However, CyIII actin is conserved at site of expression in aboral ectoderm, correlated with conservation of expression in a particular squamous epithelial tissue. 3. Actin gene family member numbers are not conserved with developmental mode. 4. No particular number of cytoplasmic actin genes is needed for development of equivalent morphologies and cell types. 5. Expression patterns of orthologous cytoplasmic actin genes can vary. The evolutionary variability of actin gene expression in early development of sea urchins over short evolutionary time spans is unexplained. It is consistent with an exchangeability of function among some cytoplasmic actin isoforms (CyI and CyII), and suggests that some aspects of evolutionary variability among related species may have more to do with regulation of expression than with differences in function. This idea can explain only a part of what is observed. Clearly, muscle actin is highly conserved in function regardless of developmental mode. New patterns of actin gene expression accompany the evolution of a novel developmental pathway 91 The differences between H. erythrogramma embryos and those of indirect-developing sea urchins might be expected to have profound effects on the cytoskeleton and on use of actin isoforms. Two distinctions in tissue types and their origins illustrate the point. First, cells of the aboral ectoderm of indirect-developing sea urchins are highly differentiated as a rigid, squamate, ciliated monolayer. Aboral ectodermal cells form an integral part of the pluteus, but they do not contribute to the formation of the juvenile (Akhurst et al. 1987; Flytzanis et al. 1989). This tissue type is not seen in adults, and S. purpuratus CyIII, an aboral ectoderm marker, is not detected in juveniles or adults. All H. erythrogramma epithelium is columnar. The loss of the CyIII gene correlates with change in ectodermal territories evolved with the change in developmental mode in H. erythrogramma. It is also interesting to note that this gene did not become co-opted and expressed in other cell types, suggesting that CyIII is functionally different from the other isoforms. Second, the origin of mesenchyme in H. erythrogramma differs from that of S. purpuratus or H. tuberculata. In indirect-developing species, much of the secondary mesenchyme arises from the tip of the invaginating archenteron. In H. erythrogramma, there is no archenteron tip mesenchyme. All mesenchyme arises from the vegetal plate (Wray and Raff 1990). Nearly 2000

11 92 cells ingress in H. erythrogramma compared to fewer than 100 cells in indirect developers (Parks et al. 1988). H. erythrogrammahas primary mesenchyme that produces the skeleton, and secondary mesenchyme cell types that are clearly distinguishable later in development (for example, the pigment cells, which later insert themselves into the ectoderm; Wray and Raff 1990). Both H. erythrogramma cytoplasmic actin genes are expressed in mesenchyme cells. One, CyI, is expressed in H. tuberculata, but the other, CyII, is not. In the H. erythrogramma mesenchyme blastula stage, CyII is detected in a subset of newly ingressed, possibly secondary mesenchyme. Actin expression, cell lineages, and the evolution of early development The most striking finding is that H. tuberculata forms a pluteus, with all of the differentiated cell types present, using only two cytoplasmic isoforms, as compared to the five found in S. purpuratus (Fig. 1). In H. tuberculata CyI appears to have been co-opted to substitute for CyII in mesenchyme and gut and a single CyIII gene is sufficient to form aboral ectoderm. Sea urchin actin genes seem particularly prone to duplication events (Lee et al. 1984; Akhurst et al. 1987; Durica et al. 1988). Such events could lead to the divergence of one gene, or a regulatory evolution event that alters its site of expression. In both S. purpuratus and H. tuberculata, the pattern of actin gene expression is cell lineage-specific, but these cell lineages are founded differently relative to the first cleavage plane (Henry et al. 1992), a change most apparent at the boundary between oral and aboral ectoderm in the pluteus. In S. purpuratus the first two cleavage planes are oriented at a 45 angle to the oral-aboral axis. The consequence is that, in an eight-cell embryo, only one of the four cells of the upper tier and one of the lower tier will give rise to oral ectoderm. In H. tuberculata this boundary appears to extend further up and over the tips of the arms. Because of the shift in axis relative to the planes of cleavage, the oral ectoderm in H. tuberculata arises from portions of two adjacent cells instead of one cell of the four-cell embryo. Cell fate maps of H. tuberculata have not been made, and so a precise statement of cell fate boundaries is not possible. Although changes in the actin family and its expression within a conserved developmental mode are striking, they are not unique. Wray and McClay (1989) showed that both heterochronies and heterotopies exist for other genes expressed in typical indirect-developing sea urchin embryos. Ascidians, which are also deuterostomes, have evolved patterns of actin utilization in development that differ substantially from sea urchins (Beach and Jeffery 1992; Kusakabe 1995), but as in H. erythrogramma, the evolutionary loss of larval features in an ascidian species has been accompanied by conversion of a muscle actin to a non-expressed pseudogene (Kusakabe 1995; Kusakabe et al. 1996). The tolerance of variation within developmental mode is related to a second surprising aspect of evolution of development. Different early developmental trajectories can lead to the same phylotypic stage (Elinson 1987; Raff 1996). The shared phylotypic stage of sea urchins is the pentameral juvenile rudiment. The numerous changes in cell lineage and morphogenesis observed in H. erythrogramma illustrate that a substantially different new trajectory in early development can evolve rapidly and among closely related lineages. In some cases, gene expression territories are conserved between H. erythrogramma and indirect developers, as in the case of primary mesenchyme (Klueg et al. 1997). However, the ectodermal territories of H. erythrogramma do not correspond directly with those of indirect developers (Haag and Raff, submitted). Sea urchin actin genes exhibit an analogous mix of conservation and flexibility in expression during early development. 2.p&:Acknowledgements This work was supported by NIH Grant RO1 HD21337 to R.A.R. and an NIH genetics training grant (T32GM07757) to J.C.K. We thank Dr. Eric Haag for critically reading the manuscript, and Drs. Robert Burke and R. A. Cameron for helpful discussions. References Akhurst RJ, Calzone FJ, Lee JJ, Britten RJ, Davidson EH (1987) Structure and organization of the CyIII actin gene subfamily of the sea urchin, Strongylocentrotus purpuratus. J Mol Biol 194: Alonso S, Minty A, Bourlet Y, Buckingham M (1986) Comparison of three actin coding sequences in the mouse; evolutionary relationships between the actin genes of warm-blooded vertebrates. J Mol Evol 23:11 22 Angerer LM, Angerer RC (1991) Localization of mrnas by in situ hybridization. Methods Cell Biol 35:37 71 Beach RL, Jeffery WR (1992) Multiple actin genes encoding the same alpha-muscle isoform are expressed during ascidian development. Dev Biol 151:55 66 Cameron RA, Britten RJ, Davidson EH (1989) Expression of two actin genes during larval development in the sea urchin Strongylocentrotus purpuratus. Mol Reprod Dev 1: Cameron RA, Courtney Smith L, Britten RJ, Davidson EH (1994) Ligand-dependent stimulation of introduced mammalian brain receptor alters spicule symmetry and other morphogenetic events in sea urchin embryos. Mech Dev 45:31 47 Cox KH, Angerer LM, Lee JJ, Davidson EH, Angerer RC (1986) Cell lineage-specific programs of expression of multiple actin genes during sea urchin embryogenesis. J Mol Biol 188: Davidson EH (1989) Lineage-specific gene expression and the regulative capacities of the sea urchin embryo: a proposed mechanism. Development 105: Davidson EH, Flytzanis CN, Lee JJ, Robinson JJ, Rose SJ, Sucov HM (1985) Lineage-specific gene expression in the sea urchin embryo. Cold Spring Harbor Symp Quant Biol 50, pp Durica DS, Garza D, Restrepo MA, Hryniewicz MM (1988) DNA sequence analysis and structural relationships among the cytoskeletal actin genes of the sea urchin Strongylocentrotus purpuratus. J Mol Evol 28:72 86 Elinson RP (1987) Changes in developmental patterns: Embryos of amphibians with large eggs. In: Raff RA, Raff EC (eds) Development as an evolutionary process. Alan R. Liss, New York, pp 1 21

12 93 Emlet RB (1990) World patterns of developmental mode in echinoid echinoderms. In: Hoshi M, Yamashita O (eds). Advances in invertebrate reproduction. Elsevier, Amsterdam, pp Fang H, Brandhorst BP (1994) Evolution of actin gene families of sea urchins. J Mol Evol 39: Fang H, Brandhorst BP (1996) Expression of the actin gene family in embryos of the sea urchin Lytechinus pictus. Dev Biol 173: Flytzanis CN, Bogosian EA, Niemeyer CC (1989) Expression and structure of the CyIIIb actin gene of the sea urchin Strongylocentrotus purpuratus. Mol Reprod Dev 1: Hahn J-H, Raff RA (1996) Expression of CyI cytoplamsic actin genes in sea urchin development. J Biochem Mol Biol 29: Hahn J-H, Kissinger JC, Raff RA (1995) Structure and evolution of CyI cytoplasmic actin-encoding genes in the indirect- and direct-developing sea urchins Heliocidaris tuberculata and Heliocidaris erythrogramma. Gene 153: Henry JJ, Klueg KM, Raff RA (1992) Evolutionary dissociation between cleavage, cell lineage and embryonic axes in sea urchin embryos. Development 114: Hyman LH (1955) The invertebrates: Echinodermata. The coelomate bilateria. McGraw-Hill, New York Kissinger JC (1995) Dynamic evolutionary changes in actin genes and their expression in camarodont sea urchins. Ph.D. dissertation, Indiana University Kissinger JC, Hahn J-H, Raff RA (1997) Rapid evolution in a conserved gene family. Evolution of the actin gene family in the sea urchin genus Heliocidaris and related genera. Mol Biol Evol 14: Klueg K, Harkey MA, Raff RA (1997) Mechanisms of evolutionary changes in timing, spatial expression, and mrna processing in the msp130 gene in a direct-developing sea urchin, Heliocidaris erythrogramma. Dev Biol 182: Kusakabe T (1995) Expression of larval-type muscle actin-encoding genes in the ascidian Halocynthia roretzi. Gene 152: Kusakabe T, Swalla BJ, Satoh N, Jeffery WR (1996) Mechanism of an evolutionary change in muscle cell differentiation in ascidians with different modes of development. Dev Biol 174: Lee JJ, Calzone FJ, Britten RJ, Angerer RC, Davidson EH (1986) Activation of sea urchin actin genes during embryogenesis: measurement of transcript accumulation from five different genes in Strongylocentrotus purpuratus. J Mol Biol 188: Lee JJ, Shott RJ, Rose SJ, Thomas TL, Britten RJ, Davidson EH (1984) Sea urchin actin gene subtypes. Gene number, linkage and evolution. J Mol Biol 172: Littlewood DTJ, Smith AB (1995) A combined morphological and molecular phylogeny for sea urchins (Echinoidea: Echinodermata). Philos Trans R Soc London Ser B 347: McMillan WO, Raff RA, Palumbi SR (1992) Population genetic consequences of developmental evolution in sea urchins (genus Heliocidaris). Evolution 46: Okazaki K (1975) Normal development to metamorphosis. In: Czihak G (ed) The sea urchin embryo. Springer, Berlin Heidelberg New York, pp Parks AL, Parr BA, Chin J-E, Leaf DS, Raff RA (1988) Molecular analysis of heterochronic changes in the evolution of directdeveloping sea urchins. J Evol Biol 1:27 44 Raff RA (1996) The shape of life. Genes, development, and the evolution of animal form. Chicago University Press, Chicago Raff RA, Field KG, Ghiselin MT, Lane DJ, Olsen GJ, Pace NR, Parks AL, Parr BA, Raff EC. (1988) Molecular analysis of distant phylogenetic relationships in echinoderms. In: Paul CRC, Smith AB (eds) Echinoderm phylogeny and evolutionary biology. Clarendon Press, Oxford, pp Rubenstein PA (1990) The functional importance of multiple actin isoforms. BioEssays 12: Shott RJ, Lee JJ, Britten RJ, Davidson EH (1984) Differential expression of the actin gene family of Strongylocentrotus purpuratus. Dev Biol 101: Smith MJ, Boom JDG, Raff RA (1990) Single-copy DNA distance between two congeneric sea urchin species exhibiting radically different modes of development. Mol Biol Evol 7: Ubisch L von (1913) Die Entwicklung von Strongylocentrotus lividus. Z Wiss Zool Abt A 106: Wang AVT, Angerer LM, Dolecki GJ, Lum R, Wang GVL, Carlos R, Angerer RC, Humphreys T (1994) Distinct pattern of embryonic expression of the sea urchin CyI actin gene in Tripneustes gratilla. Dev Biol 164: Williams DHC, Anderson DT (1975) The reproductive system, embryonic development, larval development and metamorphosis of the sea urchin Heliocidaris erythrogramma (Val.) (Echonoidea: Echinometridae). Aust J Zool 23: Wray GA, McClay DR (1989) Molecular heterochronies and heterotopies in early echinoid development. Evolution 43: Wray GA, Raff RA (1990) Novel origins of lineage founder cells in the direct-developing sea urchin Heliocidaris erythrogramma. Dev Biol 141:41 54 Wray GA, Raff RA (1991) The evolution of developmental strategy in marine invertebrates. Trends Ecol Evol 6:45 50 Yaffe D, Nudel U, Mayer Y, Neuman S (1985) Highly conserved sequences in the 3 untranslated region of mrnas coding for homologous proteins in distantly related species. Nucleic Acids Res 13: