Actin is a ubiquitous protein in eukaryotic cells and plays an important role in cell structure, cell motility, and the generation of contractile force in both muscle and nonmuscle cells. Multiple genes encoding muscle or nonmuscle actins have been isolated from several species of ascidians and their expression patterns have been investigated. Sequence and expression analyses of muscle actin genes have shown that ascidians have at least two distinct isoforms of muscle actin, the larval muscle and body-wall isoforms. In the ascidian Halocynthia roretzi, two clusters of actin genes are expressed in the larval muscle cells. The HrMA2/4 cluster contains at least five actin genes and the HrMA1 cluster contains a pair of actin genes whose expression is regulated by a single bidirectional promoter. cis-Regulatory elements essential for muscle-specific expression of a larval muscle actin gene HrMA4a have been identified. The adult body-wall muscle actin is clearly distinguished from the larval muscle actin by diagnostic amino acids. The adult muscle actin genes may be useful tools to investigate the mechanisms of muscle development in ascidian adults. The evolution of chordate actin genes has been inferred by comparing the organization and sequences of actin genes and performing molecular phylogenetic analysis. The results suggest a close relationship between ascidian and vertebrate actins. The chordate ancestor seems to have evolved the “chordate-type” cytoplasmic and muscle actins before its divergence into vertebrates and urochordates. The phylogenetic analysis also suggests that the vertebrate muscle actin isoforms evolved after the separation of the vertebrates and urochordates. Muscle actin genes have been used to investigate the mechanism of muscle cell regression during the evolution of anural development. The results suggest that the regression of muscle cell differentiation is mediated by changes in the structure of muscle actin genes rather than in the trans-acting regulatory factors required for their expression. Actin genes have provided a unique system to study developmental and evolutionary mechanisms in chordates.
Ascidians (subphylum Urochordata, class Ascidiacea) are chordates with a life cycle containing both larval and adult phases (reviewed in Satoh, 1994). Within about a day after fertilization, the ascidian egg develops into a tadpole larva, which consists of approximately 2600 cells. The ascidian tadpole larva exhibits the hallmarks of a chordate, including a motile tail containing a notochord, a dorsal nerve cord, and striated muscle cells. In contrast, the adult ascidian is a sessile organism with little resemblance to other chordates, except for the presence of pharyngeal gill slits and an endostyle, which is considered to be homologous to the vertebrate thyroid gland. Ascidian embryos have been favored for developmental research because they exhibit low cell numbers, contain only a few different tissue types, develop rapidly, and have a well-known cell lineage (Nishida, 1987; Swalla et al., 1993; Satoh, 1994; Jeffery, 1994; Satoh et al., 1996). Ascidians are also favorable for evolutionary research because of their phylogenetic position near the vertebrates and the radical changes in embryonic development exhibited by certain species (Jeffery and Swalla, 1990, 1992a; Jeffery, 1994; Satoh and Jeffery, 1995).
Actin is a major structural component of the contractile system, both in muscle cells and in nonmuscle cells (Pollard and Cooper, 1986; Sheterline and Sparrow, 1994). Actins are highly conserved proteins found in all eukaryotes from yeast to vertebrates. Most organisms exhibit multiple actin isoforms which are encoded by a small gene family (Sheterline and Sparrow, 1994). Each actin isoform shows a distinct expression pattern specific to a tissue or different developmental stages. In mammals, for example, there are at least four muscle isoforms (α-skeletal, α-cardiac, α-vascular, and γ-enteric) and two nonmuscle isoforms (β- and γ-cytopSasmic) (Vandekerckhove and Weber, 1979). Because of their widespread distribution among eukaryotes, their highly conserved sequences, and the presence of tissue- and developmental stage-specific isoforms, actin genes have been used as probes to trace the ontogeny and phylogeny of various organisms (Davidson, 1986; Raff et al., 1987; Miwa et al., 1991; Cox and Buckingham, 1992; Jeffery, 1994; Bhattacharya and Stickel, 1994; He and Haymer, 1995; Satoh et al., 1996). Here I de-scribe recent progress in studies of ascidian actin genes and discuss them in the context of evolution and development.
2. MUSCLE ACTIN GENES
Ascidians have three types of muscle tissues: larval tail muscle, adult body-wall muscle, and adult heart muscle (for review, see Satoh, 1994). The larval muscle consists of mononuclear striated muscle cells. Adult muscle tissues are morphologically distinguished from the larval muscle. The body wail (mantle) muscle cells are multinucleate smooth muscle cells (Shinohara and Konishi, 1982; Nevitt and Gilly, 1986; Terakado and Obinata, 1987), whereas the heart muscle consists of unicellular striated muscle cells (Kalk, 1970).
Muscle actin cDNA and genomic clones have been isolated from several species of ascidians (Table 1) (Tomlinson et al., 1987; Kusakabe et al., 1991, 1992, 1995, 1996; Beach and Jeffery, 1992; Kovilur et al., 1993; Swalla et al., 1994). Isolation and characterization of cDNA clones from ascidian embryos showed that multiple actin genes are expressed in larval muscle (Kusakabe et al., 1991,1995; Beach and Jeffery, 1992). cDNA clones for muscle actin were also isolated from adult body-wall libraries (Tomlinson et al., 1987; Kovilur et al., 1993; Swalla et al., 1994). Analysis of sequences and expression patterns of ascidian muscle actin genes has shown that ascidians have at least two distinct muscle actin isoforms, the larval muscle and body-wall isoforms (Kusakabe, 1995; Kusakabe et al., 1997). Little is known about muscle actin genes expressed in adult heart muscle, although they seem to be different from actin genes expressed in the larval tail muscle (Kusakabe, 1995; Kusakabe et al., 1995).
Actin genes identified in ascidians
(i) Organization and expression of larval muscle actin genes
In Halocynthia roretzi, at least five muscle actin genes (HrMA2, HrMA4a, HrMA4b, HrMA5 and HrMAS) form a cluster (HrMA2/4 cluster) in a 30 kb region of genomic DNA (Fig. 1A; Kusakabe et al., 1992). These five genes are aligned in the same direction. The nucleotide sequences of the five HrMA2/4 cluster genes, including their 51 flanking regions are highly conserved, suggesting that the expression of these genes is controlled coordinately (Kusakabe et al., 1992). Microinjection of fusion gene constructs in which the 5′ flanking region of each HrMA2/4 cluster gene is fused with lacZ gene into H. roretzi embryos suggests that these genes are coexpressed in larval muscle cells of H. roretzi embryos (Kusakabe et al., 1995). None of the HrMA2/4 cluster genes are expressed in body wall muscle and heart muscle of mature adults (Kusakabe, 1995).
The HrMA4a gene is the most extensively studied gene in the HrMA2/4 cluster. HrMA4a transcripts first appear in B6.2 (the progenitor of B7.4) at the 32-cell stage, in B7.8 at the 64-cell stage, and in B7.5 around the 76-cell stage, respectively, suggesting that the transcription of this gene is initiated prior to developmental fate restriction in the B7.4-sublineage (Satou et al., 1995). HrMA4 expression is restricted to developing muscle cells and transcripts accumulate as development proceeds (Kusakabe et al., 1991; Satou et al., 1995) (Fig. 2). Promoter function of the HrMA4a gene has been analyzed in detail, and c/s-regulatory sequences required for the muscle-specific gene expression have been identified (Hikosaka et al., 1994; Satou and Satoh, 1996; see below).
The H. roretzi genome contains another cluster of muscle actin genes (HrMA1 pair) in which the HrMAIa and HrMAlb are closely linked in a head-to-head arrangement on opposite DNA strands and share a 340-bp 5′ flanking sequence (Fig. 1B; Kusakabe et al., 1995). The nucleotide sequences of HrMAIa and HrMAlb, including their first introns, are quite similar to each other, although the 3′ half of HrMAlb has not been isolated. Examination of the promoter function of the HrMA1 pair revealed that they share promoter activity and are coexpressed in larval muscle cells (Kusakabe et al., 1995; Satoh et al., 1996). HrMA1 transcripts were first detected at the 64-cell stage (Satoh et al., 1996). The tandem cluster of HrMA2/4 genes and the bidirectional promoter of the HrMA1 pair could expedite utilization of muscle-specific frans-acting factors. The organization of larval-muscle actin genes in the genome may play an important role in the synthesis of a large amount of actins during the process of rapid differentiation of the muscle cells (Kusakabe et al., 1995).
Expression of multiple muscle actin genes encoding similar or identical proteins in larval muscle cells seems to be a common feature in ascidians. In Styela ciava, at least four different muscle actin genes (ScTbl, ScTb24, ScTb30, and ScTb12/34) encoding the same actin isoform are expressed in the larval tail muscle cells (Beach and Jeffery, 1992). These muscle actin genes are expressed in different temporal and spatial expression patterns during development. ScTbl transcripts were detected in eggs at very low level and disappear shortly after fertilization. On the other hand, maternal mRNA of ScTbl2/34, 24, 30 was not detected. Zygotic transcription of ScTb genes first appears during gastrulation, and transcripts gradually accumulate during subsequent neurulation and muscle cell differentiation. ScTb24 is expressed primarily in tail muscle cells of the developing tailbud embryos, whereas ScTbl is expressed in a variety types of cells, including the muscle, mesenchyme, epidermal, and neural cells.
In contrast to the H. roretzi larval muscle actin genes, some of the ScTb genes (ScTbl, 12, 24) are also expressed in adults (Beach and Jeffery, 1992). in situ hybridization analysis showed that ScTb24 transcripts were most abundant in vascular tissues (blood sinuses) of the body wall and branchial sac; much lower level of transcripts were detected in body-wall muscle cells (Beach and Jeffery, 1992). On the other hand, ScTb30 expression was only detected in embryos (Beach and Jeffery, 1992).
(ii) Transcriptional regulation of larval muscle actin genes HrMA4a gene.
The 5′ flanking sequences of the five HrMA2/4 cluster genes resemble each other. As shown in Fig. 3A, the 5′ upstream region close to the transcription start site of HrMA4a contains several consensus sequences, which include a TATA box at -30, an E-box at -71, a CArG box at -116, and a cluster of three E-boxes between -150 and -190 (Kusakabe et al., 1992).
Promoter function of an ascidian muscle actin gene was first investigated by microinjection of pHrMA4aCAT, a recombinant plasmid in which about 1.4 kb of the 5′ upstream region of HrMA4a had been fused with the bacterial chloramphenicol acetyltransferase gene (CAT). When pHrMA4aCAT was introduced into H. roretzi fertilized eggs and the appearance of CAT protein was examined later by the anti-CAT antibody, CAT expression was restricted to muscle cells of the larval tail (Hikosaka et al., 1992). Specific promoter function of the 1.4 kb upstream region was further confirmed by microinjection of another fusion gene construct, pHrMA4a-Z, in which the same upstream region was fused with the coding sequence of a bacterial (3-galactosidase (lacZ) gene. When this construct was introduced into fertilized eggs and the lacZ expression was examined by histochemical detection of the enzyme activity, expression was present in larval muscle cells (Fig. 4A) (Hikosaka et al., 1993, 1994).
Deletion constructs of the 5′ upstream region of HrMA4a fused with lacZ were microinjected into H. roretzi fertilized eggs. No changes in the level or tissue specificity of expression were evident when the upstream sequence was deleted down to -103, suggesting that the 103-bp upstream region is sufficient for appropriate spatial expression of the gene (Hikosaka et al., 1994). However, the reporter gene was not expressed in muscle cells when a 72-bp upstream region was used in the assay (Hikosaka et al., 1994). Satou and Satoh (1996) further narrowed down the 5′-flanking region of HrMA4a required for muscle-specific expression by an exhaustive analysis. They analyzed expression of promoter-lacZ fusion constructs containing a series of 3′-deletions of the promoter, a mutated proximal promoter, or a partial SV40 promoter sequence in place of HrMA4a proximal promoter sequence. The results demonstrated that the 38-bp long sequence between -66 and -103 is sufficient for the muscle-specific expression of the gene (Satou and Satoh, 1996). Within this 38-bp region, two short sequences, 5′-TCGCACTTC-3′ and 5′-GTGATAACAACTG-3′, were shown to be essential to the muscle-specificity of HrMA4a promoter by analyzing a series of mutation constructs in which two- or three-base substitutions were introduced into the 38-bp region (Fig. 3A) (Satou and Satoh, 1996).
The reporter gene was also expressed in larval muscle cells when pHrMA4a-Z was injected into Ciona savignyi eggs (Hikosaka et al., 1993; see Fig. 4B). Ciona and Halocynthia are representative of the two major subgroup of ascidians, and the genetic circuitry underlying muscle differentiation seems to be conserved between the two species. This may be a great advantage to study mechanisms of muscle cell differentiation of ascidian embryos since muscle determinants have been investigated using Ciona (Nishikata et al., 1987; Marikawa et al., 1994,1995), while Halocynthia embryos have been used to study control mechanisms of muscle-specific gene expression (Hikosaka et al., 1994; Araki et al., 1994; Kusakabe et al., 1995; Satou et al., 1995; Satou and Satoh, 1996; Araki and Satoh, 1996; Satoh et al., 1996).
Myogenic regulatory genes, MyoD, myogenin, Myf-5, and MRF-4, are essential for vertebrate myogenesis (Hasty et al., 1993; Rawls et al., 1995). They encode transcription factors that belong to the basic helix-loop-helix (bHLH) protein family. These myogenic bHLH proteins bind to the E-box motif (CANNTG) as heterodimers with E2A proteins (Lassar et al., 1991). The AMD1 gene, encoding an ascidian homolog of vertebrate myogenic bHLH proteins, was isolated and characterized (Araki et al., 1994). Zygotic expression of AMD1 begins at the 64-cell stage and was confined to the muscle lineage cells during embryogenesis (Satoh et al., 1996). There is an E-box sequence at -71 of HrMA4a (Fig. 3A). We examined the significance of this motif for muscle-specific expression of the reporter gene. Mutations in the proximal E-box sequence did not diminish the muscle-specific expression of the reporter gene, although frequency of embryos with p-ga-lactosidase activity decreased to about two-thirds of the control (Hikosaka et al., 1994). Therefore, it is unlikely that AMD1 is required for and is closely associated with muscle-specific expression of HrMA4a. However, AMD1 may indirectly regulate HrMA4a expression via interaction with other DNA-bind-ing factors and maintain the differentiation state by enhancing the expression of muscle-specific differentiation genes.
HrMAIa and HrMAIb genes.
As shown in Fig. 3B, the transcription initiation sites of HrMAIa and HrMAIb genes are only 340-bp apart and a TATA box is located at -30 in each promoter (Kusakabe et al., 1995). Nucleotide sequences of the 51 untranslated region and untranscribed region up to the TATA boxes are highly conserved between the two genes, whereas the nucleotide sequence between the two TATA boxes showed no distinct symmetry except for the presence of two CArG box-like sequences (Minty and Kedes, 1986) around position -80 (Fig. 3B). One E-box sequence and one MEF2 binding site (Gossett et al., 1989) are located in the middle of the 5′ flanking region of the genes (Fig. 3B). When constructs in which the shared upstream region of HrMA1 pair was fused with lacZ in either direction were microinjected into eggs, the reporter gene was expressed in muscle cells of the larval tail, suggesting a bidirectional promoter that regulates muscle-specific transcription of the HrMA1 pair (Kusakabe et al., 1995). Mutations in the E-box sequence did not diminish the muscle specific expression of the reporter gene, although the frequency of embryos with (3-galactosidase activity decreased to some extent (Satoh et al., 1996). Therefore, as in the case of the clustered actin-genes, it is unlikely that AMD1 is required for muscle-specific expression of HrMAIa and HrMA 1b (Satoh et al., 1996).
The promoter activity of deletion constructs of HrMAIa and HrMAIb was also examined (Satoh et al., 1996). Deletion constructs of the 190-bp upstream region of HrMAIa and of the 139-bp upstream region of HrMA 1b lack the MEF2 binding site (Fig. 3B). When these deletion constructs were micro-injected into fertilized eggs, the reporter gene was expressed in muscle cells of tailbud embryos. Deletion constructs of the 85-bp upstream region of HrMAIa and of the 89-bp upstream region of HrMAIb lack the CArG box-like binding site (Fig. 3B). When these deletion constructs were microinjected into fertilized eggs, the reporter gene expression was not detected in most of the injected embryos. Therefore, it is likely that rather short sequences including CArG box-like sequence are essential for the muscle-specific expression of HrMAIa and HrMAIb (Satoh et al., 1996).
(iii) Actin genes expressed in adult muscle
As described above, neither the HrMA2/4 cluster genes nor the HrMA1 genes are expressed in body-wall muscle and heart muscle of H. roretzi mature adult (Kusakabe, 1995; Kusakabe et al., 1995). The adult muscle tissues seem to express actin genes different from those expressed in larval muscle (Kusakabe, 1995). Muscle actin genes distinct from those expressed in larval muscle cells have been identified from Styela plicata and Molgula citrina adult body-wall cDNA libraries and a Molgula oculata genomic library (Tomlinson et al., 1987b; Kovilur et al., 1993; Swalla et al., 1994; Kusakabe et al., 1997) (Table 1). These genes are expressed in adult body-wall muscle but not in larval tail muscle (Jeffery et al., 1990; Swalla et al., 1994).
The temporal and spatial expression patterns of the adult muscle actin genes were investigated during postembryonic development of several ascidian species, including S. plicata, M. citrina, M. oculata, and Molgula occulta (Jeffery et al., 1990; Swalla et al., 1994). In S. plicata, M. oculata, and M. occulta, which are oviparous species, expression of the adult muscle actin gene begins after the larva settles on a substrate and begins metamorphosis (Jeffery et al., 1990). On the other hand, in M. citrina, which is an ovoviviparous species, the expression of adult muscle actin gene McMA1 was first detected exclusively in the mesenchyme cells of the late tailbud embryos. The McMA1 transcripts persist in mesenchyme cells after metamorphosis. Thus, the adult muscle actin gene shows a heterochronic shift of expression into the larval phase in M. citrina (Jeffery et al., 1990; Swalla et al., 1994). M. citrina has evolved adultation (Jagersten, 1972), in which development of the branchial sac, siphons, and heart is shifted into the larval phase (Grave, 1926). Thus, the heterochronic expression of McMA 1 accompanies adultation in M. citrina (Jeffery et al., 1990; Swalla et al., 1994).
In H. roretzi, some larval muscle actin genes are expressed in trunk ventral ceils, which are derived from B7.5 blastomeres of the 64-cell stage embryo (Nishida, 1987; Kusakabe et al., 1995; Y. Satou et al., unpublished data). A recent cell-lineage study has shown that the trunk ventral cell is a progenitor of adult body-wall and heart muscle cells (Hirano and Nishida, personal communication). Therefore, the larval muscle actins may also be a component of body-wall and heart myofibrils in the early stages of adult development. The expression patters of muscle actin genes during adult muscle development are still largely unknown. Future studies on muscle actin gene expression during adult development would provide important information to help understand development of body wall and heart muscles as well as their evolutionary relationship with vertebrate musculature.
3. NONMUSCLE ACTIN GENES
Actin genes expressed in nonmuscle cells encode isoforms different from those in muscle cells (Sheterline and Sparrow, 1994). cDNAs coding for nonmuscle actin (cytoplasmic or cytoskeletal actin) have been isolated from the ascid-ians S. plicata, S. ciava, and H. roretzi (Beach and Jeffery, 1990; Kovilur et al., 1993; Araki et al., 1996) (Table 1). ScCA15 is a cDNA clone encoding a cytoplasmic actin of S. ciava (Beach and Jeffery, 1990). The ScCA 15 transcripts are present in eggs and cleaving embryos and disappear before gastrula-tion. Zygotic ScCA15 mRNA accumulation begins after neu-rulation and continues during tail formation, in situ hybridization shows that the zygotic transcripts accumulate primarily in epidermis and neural tube (Beach and Jeffery, 1990). These tissues continue to divide after cell division has ceased in other embryonic cells. The ScCA15ls also expressed in specific tissues of the adult: notably the digestive tract and the germinal layers of the testis and ovary, each of which also contains populations of rapidly dividing cells (Ermak, 1975,1976). Thus, the ScCA15 actin may function in cell proliferation (Beach and Jeffery, 1990). A cDNA clone SpCA8 was isolated from an S. plicata body-wall library and encodes a cytoplasmic actin resembling ScCA15 (Kovilur et al., 1993). The expression patterns of SpCA8 have not been reported to date.
In H. roretzi, the expression patterns of a cytoplasmic actin gene (HrCA1) have been investigated by northern and in situ hybridizations (Araki et al., 1996). As in the case of ScCAW, HrCA1 transcripts are present in eggs and disappear after fertilization. The zygotic expression of HrCAt begins at the late gastrula stage. The first sign of expression was detected in primordial muscle cells. Weak expression was then observed in notochord cells and this expression continued to be present up to the hatching stage. At the neural-plate and later stages, HrCA1 is expressed predominantly in mesenchyme cells and some neuronal cells. In addition, cleavage-arrested embryos exhibited HrCA 1 expression only in mesenchyme-cell lineages. Therefore, HrCA1 expression may be used as molecular marker for monitoring the differentiation of mesenchyme cells (Araki et al., 1996). In adults, HrCA1 expression was found in every tissue examined, including the gill, body-wall muscle, gonad, digestive gland, and intestine. The expression patterns of HrCA 1 are different from those of ScCA15 (Beach and Jeffery, 1990; Araki et al., 1996). The deduced amino acid sequences are also considerably different between HrCA1 and ScCA15 (Kusakabe et al., 1997). Ascidians probably have at least two types of nonmuscle actins whose sequence and expression patterns are different from each other (Araki et al., 1996; Kusakabe et al., 1997).
4. ACTIN GENES AND CHORDATE EVOLUTION
(i) Phylogenetic analysis of chordate actins
To investigate the origin and evolution of chordate actin isoforms, the sequences and exon-intron organization of as-cidian actin genes were compared with those of other invertebrate and vertebrate actin genes (Kovilur et al., 1993; Kusakabe, 1995; Kusakabe et al., 1997). Sequence comparisons and molecular phylogenetic analyses suggested a close relationship between the ascidian and vertebrate actin isoforms (Kovilur et al., 1993; Kusakabe et al., 1997). In vertebrates, the muscle actin is clearly distinguished from the nonmuscle actin by about 20 diagnostic amino acid positions (Vandekerckhove and Weber, 1978,1979). On the other hand, nonchordate-invertebrate muscle actins, including echinoderm, arthropod, nematode actins, are much more similar to the vertebrate cytoplasmic actin than to the vertebrate muscle actin (Vandekerckhove and Weber, 1984). As shown in Figs. 5 and 6, ascidian muscle actins are distinguished by diagnostic amino acids and group with the vertebrate muscle actins in the phylogenetic tree. Similarly, the ascidian cytoplasmic actin SpCA8 is similar to the vertebrate cytoplasmic actin (Figs. 5 and 6; Kovilur et al., 1993; Kusakabe et al., 1997). The chordate ancestor seems to have evolved the “chordate-type” cytoplasmic and muscle actins before its divergence into vertebrates and urochordates.
The chordate muscle actin clade consisted of three branches, each supported by relatively high bootstrap values (Fig. 5). One of the branches contained the vertebrate muscle actins and the other two branches contained the ascidian larval and ascidian body-wall muscle actins. The presence of two distinct lineages of ascidian muscle actins is consistent with our findings that the larval and adult muscle actins are distinguished by diagnostic amino acids (Kusakabe, 1995). The molecular phylogenetic trees indicated that the ascidian larval muscle actin is more closely related to the vertebrate muscle actin than the ascidian adult actin (Fig. 5; Kusakabe et al., 1997). This suggests that the chordate ancestor had at least two muscle actin isoforms: the ancestral adult-muscle actin and ancestral vertebrate/larval muscle actin. However, an alternative possibility is that the adult isoform appeared in the urochordate lineage after the vertebrate lineage diverged and that the ancestral adult actin evolved rapidly.
The molecular phylogenetic analyses showed that the four muscle actin isoforms in vertebrates (α-skeletal, α-cardiac, α-vascular, and γ-enteric) are more closely related to each other than to the ascidian isoforms (Fig. 5; Kusakabe et al., 1997). Similarly, the [β- and γ-cytoplasmic actins of vertebrates show a closer relationship to each other than to the ascidian and echinoderm nonmuscle isoforms (Fig. 5). These results suggested that the vertebrate actin gene family was established by the duplication of one ancestral muscle actin gene and one ancestral cytoplasmic actin gene after the divergence of the vertebrate and urochordate lineages (Kusakabe et al., 1997). The diversification of multigene families is thought to have played an important role during vertebrate evolution (Miyata et al., 1994; Iwabe et al., 1996) and may have coincided with the evolution of the complex vertebrate body plan.
The length and sequence of the amino-terminal regions are highly variable among actin isoforms and in different species (Fig. 6). While vertebrate muscle actins and most invertebrate actins have a Met-Cys sequence followed by a cluster of acidic amino acids (Glu and/or Asp), the vertebrate cytoplasmic actins lack a Cys residue next to the first Met. The comparison of the amino-terminal sequences and molecular phylogenetic analyses showed that the HrCAl cytoplasmic actin is closely related to the echinoderm cytoplasmic actin. In contrast the SpCA8 cytoplasmic actin is more distantly related to the echinoderm cytoplasmic actin, and its amino-terminal sequence is similar to that of the vertebrate cytoplasmic actins and lacks a Cys residue next to the first Met. These results suggest that at least two types of nonmuscle actins are present in ascidians. Multiple nonmuscle actin genes in ascidians were suggested by genomic Southern hybridization (Beach and Jeffery, 1990) and the expression pattern of a cytoplasmic actin gene (Beach and Jeffery, 1990; Araki et al., 1996). Since a vertebrate-type cytoplasmic actin lacking a Cys residue in the amino-terminal region has not been reported in echinoderms, the vertebrate-type cytoplasmic actin genes may have arisen from an ancestral actin gene by losing the Cys residue (Kusakabe et al., 1997).
(ii) Exon-lntron organization
The positions of introns in ascidian muscle actin genes were shown to be identical to those of vertebrate muscle actin genes (Kusakabe et al., 1992,1995,1997). However, the number of introns in the ascidian larval muscle actin genes is smaller than that in other deuterostome actin genes. An extreme case is the M. oculata muscle actin gene MocuMAI, which contains no introns (Kusakabe et al., 1996). The primitive situation in deuterostomes, however, seems to be muscle actin genes with introns. The ascidian larval muscle actin genes may have lost their introns to expedite the processing and cytoplasmic accumulation during the relatively short interval of muscle cell differentiation during larval development (Kusakabe et al., 1996).
Six intron positions (41/42, 121/122, 150, 204, 268, 328/329) are present in both muscle and nonmuscle actin genes in deuterostomes, but they are not conserved in protostomes (Kusakabe et al., 1997). The ancestral deuterostome may have had a single prototypic actin gene that contained six or more introns. Since the number of introns varies from zero to seven in the extant deuterostome actin genes, different introns seem to have been lost during the evolution of each lineage. The conservation of intron positions in both the deuterostome cytoplasmic and muscle actin genes suggests that the ancestral vertebrate-type muscle actin gene appeared early during chordate evolution, and its characteristic amino acid sequence was established in a relatively short time.
The intron at position 308 in the HrCA 1 cytoplasmic actin gene is unique among deuterostome actin genes but is also present in Drosophila melanogaster muscle actin genes 79B and 88F (Fyrberg et al., 1981; Kusakabe et al., 1997). Common intron positions of distantly related species are known for plant and vertebrate-muscle actin genes (position 150) and for plant actin genes and a Caenorhabditis elegans actin gene (position 18/19). The presence of these conserved intron positions supports the hypothesis that the ancient eukaryotic actin gene had a large number of introns (Doolittle, 1978; Kusakabe et al., 1997).
5. EVOLUTIONARY CHANGES IN MUSCLE ACTIN GENES IN ANURAL ASCSDIANS
(i) Anural (tailless) development of ascidians
Most ascidian species show indirect development in which the embryo develops into a tadpole larva. The tailed (or urodele) larva consists of a head, containing a brain with a neural sensory organ(s), and a tail, containing a notochord and flanking bands of striated muscle cells (Katz, 1983). The larval tail is formed by coordinated morphogenetic movements and differentiation of the prospective notochord, muscle, and posterior epidermal cells (Swalla, 1993; Satoh, 1994). Anural development is an alternate mode of development in which the embryo develops into a tailless (or anural) larva (reviewed in Jeffery and Swalla, 1990). Anural embryos lack typical urodele larval features, including the neural sensory organ, notochord, and differentiated muscle cells. Fewer than 20 ascidian species have been described with anural development, and most of these species are classified in the family Molgulidae (Berrill, 1931; Jeffery and Swalla, 1990). Urodele development is thought to be ancestral in ascidians, a viewpoint supported by the expression of vestigial urodele features in some anural species (Berrill, 1931; Whittaker, 1979; Swalla and Jeffery, 1990, 1992; Bates and Mallett, 1991) and molecular phylogenetic analysis (Hadfield et al., 1995).
The mechanisms underlying the transition from urodele to anural development have been investigated in the closely related ascidians M. oculata, which has a urodele larva, and M. occulta, which has an anural larva (Swalla and Jeffery, 1990, 1996; Jeffery and Swalla, 1991; Swalla et al., 1993; Kusakabe et al., 1996). M. occulta produces notochord and muscle precursor cells but they remain undifferentiated in the posterior region of the larva. The ancestral urodele features, including the neural sensory organ, the notochord, and some aspects of muscle cell differentiation, are restored in interspecific hybrids produced by fertilization of M. occulta eggs with M. oculata sperm (Swalla and Jeffery, 1990; Jeffery and Swalla, 1991, 1992b). The restoration of urodele features in hybrid larvae suggests that anural development is mediated by ioss-of-function mutations in zygotic genes. Recently, the uro genes, which encode potential regulatory factors expressed in M. oculata but not in M. occulta, have been identified and characterized (Swalla et al., 1993). The uro gene Manx, which encodes a zinc finger protein, is required for restoration of urodele features in hybrid embryos and may play an important role in the specification of the chordate body plan (Swalla and Jeffery, 1996).
(ii) Muscle actin gene expression and its regulation in anural embryos
The muscle actin genes were used to investigate the mechanism of muscle cell regression of anural development (Kusakabe et al., 1996). MocuMAI is a single-copy, larval muscle actin gene in the urodele ascidian M. oculata. The accumulation of muscle actin mRNA in M. occulta anural embryos was investigated by in situ hybridization using the coding region of MocuMAI as a probe. The results showed that M. occulta embryos do not produce detectable amounts of muscle actin mRNA during embryogenesis (Kusakabe et al., 1996). The MocuMA 1 probe can detect muscle actin mRNA in embryos of the urodele ascidian species Molgula occidentalis and Styela clava, which are more distantly related to M. oculata than is M. occulta. Therefore, the inability of the MocuMAI probe to detect muscle actin transcripts in the M. occulta embryos seems due to lack of muscle actin gene expression but not to sequence divergence between the M. occulta and M. oculata muscle actin genes. The larval muscle actin gene(s) is probably silent or downregulated in M. occulta embryos. The sister species M. occulta and M. oculata are capable of interspecific hybridization (Swalla and Jeffery, 1990). The results of in situ hybridization experiments with the MocuMAI coding region probe showed that muscle actin mRNA accumulates in the vestigial muscle cells of hybrid embryos produced by fertilizing M. occulta eggs with M. oculata sperm (Kusakabe et al., 1996). Thus, muscle actin gene expression is restored in the vestigial muscle cells of hybrid embryos.
pMocuMA1-Z is a fusion gene construct consisting of the 702-bp 5′ upstream region of the MocuMAI gene fused with the lacZ gene that is expressed in the tail muscle cells of urodele embryos. To investigate whether frans-acting factors required for MocuMAI expression are present in M. occulta embryos, pMocuMA1-Z was microinjected into M. occulta eggs and the resulting larvae were assayed for p-galactosidase activity (Fig. 7; Kusakabe et al., 1996). Expression of pMocuMA1-Z was observed in a few vestigial muscle cells in the posterior region of M. occulta larvae, suggesting that transacting factors responsible for expression of the MocuMA 1 gene in the muscle cell lineage have been retained in M. occulta embryos (Fig. 4C).
(ii) Loss-of-functiors mutations of muscle actin genes in anural ascidians
If trans-acting factors regulating MocuMAI gene expression are retained in M. occulta embryos, then lack of muscle actin gene expression must be due to other evolutionary changes. Two muscle actin genes MoccMA 1a and MoccMA 1b, which are orthologous to MocuMAI, were isolated from an M. occulta genomic library (Kusakabe et al., 1996). The coding regions of the MoccMA 1a and MoccMA 1b genes contains deletions, insertions, and codon substitutions that would make their encoded polypeptide nonfunctional as actins. The frame-shifts in the MoccMAIa and MoccMAlb genes would generate stop codons, causing premature termination of translation. These features suggest that muscle actin gene expression has been altered in M. occulta by mutations in the MoccMAIa and MoccMAlb genes.
The MoccMA1a gene shows strong similarity to MocuMA1 in its 5′- and 3′-flanking sequences, including the putative promoter region. The 5′- and 3′-flanking regions of the MoccMA 1b gene are less conserved relative to the MocuMAI gene. Promoter activity in the 5′ upstream regions of the MoccMAIa and MoccMA 1b genes was investigated by microinjecting pro-moter-/acZ fusion gene constructs into eggs of the urodele ascidian Ciona intestinalis (Kusakabe et al., 1996). Interestingly, expression of MoccMAIa/lacZ and MoccMA1b/lacZ fusion constructs showed that they both retain muscle-specific promoter activity, although it is reduced in MoccMAlb.
Our analysis on muscle actin genes suggest that the regression of muscle cell differentiation in M. occulta embryos is mediated by loss-of-function mutations in muscle actin genes, which have become pseudogenes, rather than by changes in frans-acting regulatory factors required for expression of these genes (Fig. 8; Kusakabe et al., 1996). Other changes leading to lack of muscle cell differentiation in anural embryos, such as the loss of myosin gene expression (Swalla and Jeffery, 1990), may be due to similar changes in gene structure. Loss of function of tissue-specific structural genes by cis-mutation may be an important mechanism of regression of certain cell types as an evolutionary modification of development.
Actin genes have provided a unique system to study developmental and evolutionary mechanisms in ascidians.
Multiple muscle actin genes are expressed during larval muscle cell development. In the genome these genes form clusters which may have evolved to synthesize a large amount of actin during rapid differentiation of larval muscle cells. Further studies on the transcriptional regulation of the actin gene clusters may provide information on the role of genomic structure in the regulation of transcription.
Expression patterns and cis-regulatory elements of the larval muscle actin genes have been analyzed in detail. Rather short upstream sequences of the 5′ flanking region of muscle actin genes are responsible for the tissue-specific expression of the genes. Investigating frans-acting factors controlling expression of larval muscle actin genes may lead to identification of muscle determinants localized in egg cytoplasm.
Ascidians have distinct actin isoforms for adult muscle. The adult muscle actin genes have been used to study heterochrony in ascidian development and may be useful tools to investigate adult muscle development.
Ascidians probably have multiple nonmuscle actin isoforms whose sequences and expression patterns are distinct. They can be used as molecular markers for certain cell types such as mesenchyme cells.
Phylogenetic analysis of actin sequences suggests a close relationship between ascidian and vertebrate actins. The chordate ancestor seems to have evolved the “chordate-type” cytoplasmic and muscle actins before its divergence into vertebrates and urochordates. The phylogenetic analysis also suggests that vertebrate actin isoforms evolved after the separation of the vertebrates and urochordates.
Muscle actin genes have been used to investigate the mechanism of muscle cell regression during the evolution of anural development. The results suggest that this process is mediated by changes in the structure of muscle actin genes rather than in the trans-acting regulatory factors required for their expression.
Future studies using ascidian actin genes may further elucidate fundamental problems in development and evolution. In addition to ascidian actin genes, future studies on actin genes in hemichordates and other chordates, including am-phioxus, salps, larvaceans, lampreys, hagfish, and jawed fish would provide important information about chordate evolution as well as evolution of gene regulation and development.
I thank Drs. Noriyuki Satoh, William R. Jeffery, Kazuhiro W. Makabe, Akira Hikosaka, Isato Araki, Billie J. Swalla, and Yutaka Satou for their collaboration and encouragement in this research. I am also grateful to Dr. William R. Jeffery for critical reading of the manuscript. Our research described here was supported in part by Grants-in-Aid from the Ministry of Education, Science, Sports and Culture of Japan to T.K. and Dr. Noriyuki Satoh, and by NIH and NSF grants to Dr. William R. Jeffery.