Heterotrimeric G proteins play crucial roles as mediators of signaling by many extracellular stimuli. The receptors that activate G proteins constitute the largest and most diverse family of cell surface molecules involved in signal transmission of metazoan cells. To investigate G protein signaling in the central nervous system (CNS) of chordates, we isolated cDNA fragments encoding five different G protein α subunits (CiGαx, CiGαq, CiGαi1a, CiGαi1b, and CiGαi2) from larvae of the ascidian, a simple chordate, Ciona intestinalis. In situ hybridization analysis revealed that each isoform had distinct patterns of spatial distribution in embryos. Among them, CiGαi1a and CiGαi1b mRNAs were specifically expressed in the CNS of the larva, whereas CiGαq transcripts were expressed in small parts of the trunk epidermis and the tip of the tail, but not in the CNS. The CiGαx expression was widely observed throughout the trunk and tail of the embryos, and the signals were stronger in the epidermis, mesenchyme, and tail muscle cells. Comparison of cDNA sequences and the exon-intron organization indicate that CiGαi1a and CiGαi1b are produced by alternative splicing of transcripts from a single gene, CiGαi1. In the cleavage and gastrula stages, transcripts of CiGαi1 were widely distributed in embryos, and the expression then became restricted to the CNS of tailbud embryos and larvae. An exhaustive search has failed to find transducin-type α subunits in C. intestinalis. Since CiGαi1 is expressed in the ocellus, CiGαi1 may mediate signals from Ci-opsin1, a visual pigment of the ocellus photoreceptor cells.
Ascidians, or sea squirts, are lower chordates, and their simple, tadpole-like larvae share a basic body plan with vertebrates (Corbo et al., 2001). Ascidian embryos have been favored for developmental research because they have low cell numbers, contain only a few different tissue types, develop rapidly, and have a well-known cell lineage (Corbo et al., 2001; Satoh, 2001). The larva of Ciona intestinalis, the cosmopolitan ascidian species, has 2,600 cells, including only 40 notochord cells and 36 muscle cells. The larva has a remarkably simple central nervous system (CNS) with about 330 cells, of which less than one-third are neurons, the remainder being glial cells (Meinertzhagen and Okamura, 2001). An anterior brain vesicle contains two sensory organs, an eyespot (ocellus) and a gravity sense organ (otolith). These two sensory organs are responsible for the swimming behavior of the larva.
It was shown that the larvae were induced to swim upon a step-down of light, and the action spectrum of photic behavior of ascidian larvae was similar to the absorption spectrum of human rhodopsin (Nakagawa et al., 1999; Tsuda et al., 2001). Localization of rhodopsin in the ocellus was shown by the retinal protein imaging method (Ohkuma and Tsuda, 2000). Ci-opsin1, the opsin of Ciona intestinalis, has been identified and shown to be closely related to the vertebrate retinal and pineal opsins (Kusakabe et al., 2001).
Opsins are apoproteins of visual pigments and a model of G protein-coupled receptors (GPCRs). Among signaling receptors, GPCRs are especially important because they constitute the largest and most diverse families of receptor proteins. More than 1000 GPCRs identified in the human genome are involved in the regulation of virtually all physiological processes (Marinissen and Gutkind, 2001). In a project analyzing the expressed sequence tags (EST) of C. intestinalis larvae (Kusakabe et al., 2002), we identified several cDNA clones encoding GPCRs. It is expected that tremendous numbers of GPCRs will be discovered in the whole genome sequencing projects of Ciona intestinalis, which are currently in progress at the National Institute of Genetics, Japan and Joint Genome Institute, USA.
Most cells contain GPCRs that choose between multiple G proteins to regulate a host of intracellular signaling processes (Gilman, 1987; Tsuda, 1987; Neer, 1995; Hamm and Gilchrist, 1996; Offermanns, 2001). Heterotrimeric G proteins are composed of α, β, and γ subunits, and there are multiple isoforms of each subunit (Nürnberg et al., 1995; Hildebrandt, 1997). In response to extracellular signals, the GPCRs exert guanine nucleotide exchange factor activity that substitutes GTP for GDP on the α subunit (Gα), resulting in dissociation of the αβγ trimer into an active Gα-GTP monomer and a Gβγ dimer. The activated G proteins, in turn, regulate the activity of a variety of effector proteins, including intracellular enzymes and ion channels. A given GPCR can activate more than one G protein subtype, leading to divergent signaling pathways. The importance of G proteins has been suggested in the regulation of developmental events, such as those regulated by Wnt and Hedgehog signaling (Hammerschmidt and McMahon, 1998; DeCamp et al., 2000; Liu et al., 2001; Malbon et al., 2001; Knust, 2001).
In this study, as an initial step to understand specific coupling between GPCRs and G proteins in ascidian embryos, we isolated and characterized cDNA clones encoding five different Gα subunits from the ascidian Ciona intestinalis. Each Gα isoform showed distinct expression patterns during embryogenesis, suggesting the involvement of G protein signaling in a variety of physiological and developmental processes in ascidian embryos and larvae. We also report that two CNS-specific Gα isoforms are produced from a single gene by alternative splicing. The CNS-specific Gα isoforms may couple with Ci-opsin1 in the photoreceptor cells of the ocellus.
MATERIALS AND METHODS
Animals and embryos
Mature adults of C. intestinalis were collected from harbors in Murotsu and Aioi, Hyogo, Japan. The adults were maintained in indoor tanks of artificial seawater (Marine Art BR, Senju Seiyaku, Osaka, Japan) at 18°C. The embryos were prepared using gametes obtained from the gonoducts, as described previously (Nakagawa et al., 1999).
Isolation and sequencing of cDNA clones encoding G protein α subunits
The total RNA was prepared from the C. intestinalis larvae, and a cDNA library was constructed with a λZAP vector (Stratagene, La Jolla, USA), as described previously (Iwasa et al., 2000). The cDNA library was directly used as a template to amplify the cDNA fragments of about 500 bp encoding the G protein α subunits by polymerase chain reaction (PCR), using a pair of degenerate oligo-nucleotide primers corresponding to two conserved amino acid sequences, KQM(K/R)IIH and KWI(H/Q)CF, respectively. The 5′-and 3′- portions of cDNAs were amplified from the cDNA library by PCR using a gene-specific primer and a vector primer, as described previously (Iwasa et al., 2000). Full-length cDNA clones were amplified by PCR using a thermostable DNA polymerase bearing proofreading activity (Takara LA Taq; Takara Shuzo, Japan), with primers corresponding to the 5′- and 3′-untranslated region (UTR) sequences of cDNAs (5′-ATACGAGCAAGCACAGCGGGAA-3′ for 5′-UTR, and 5′-TATGCATGCGATGACGTCAC-3′ for 3′-UTR). The PCR products were subcloned into plasmid vectors and sequenced on both strands with an automatic DNA sequencer (Shinadzu DSQ 1000L, Shimadzu, Kyoto, Japan).
Molecular phylogenetic analysis
The deduced amino acid sequences of Gα encoded by the 500-bp cDNA fragments amplified by PCR from C. intestinalis larvae were aligned with the amino acid sequences of Gα from other animals. A neighbor-joining tree was constructed with the alignment using the Clustal W program (Thompson et al., 1994). The evolutionary distances were estimated using Kimura's empirical method. The sequences used were: eleven Homo sapiens Gα isoforms (Gαi XM_011603, Gαq NM_002072, Gαs X04408, Gαolf L10665, Gαz XM_009867, Gα11 AF011497, Gα12 L01694, Gα13 L22075, Gα15 XM_009220, cone transducin D10377, rod transducin X63749), two Rattus norvegicus isoforms (Gαo M17526, gustducin X65741), four Octopus vulgaris Gα isoforms (Gαi AB025780, Gαq AB025782, Gαo AB025781, Gαs AB025783), and Drosophila melanogaster Gαf L09700.
In situ hybridization
Digoxigenin-labeled RNA probes were synthesized using a DIG RNA labeling kit (Roche, Japan), according to the manufacturer's protocol. For CiGαi2, CiGαq, and CiGαx, the 500-bp cDNA fragments obtained by PCR were used as templates to synthesize the probes. For CiGαi1a and CiGαi1b, the RNA probe was synthesized from the 136-bp coding region (nt 607–742 for CiGαi1a and nt 497–632 for CiGαi1b) corresponding to the alternatively-spliced exon of each transcript. To detect all transcripts from the CiGαi1 gene, an RNA probe was also synthesized from a cDNA fragment containing 3′-coding and untranslated regions (spanning from the nt 603 to the 1925 of CiGαi1a).
Whole-mount in situ hybridization was carried out basically according to the protocol by Wada et al. (1995). After the coloring reaction, the embryos and larvae were dehydrated in an ethanol series and incubated in ethanol for 5–10 min. Following rehydration with PBST, the embryos were incubated in 25% glycerol in PBST for 5 min, and then transferred to 50% glycerol in PBST. The embryos were photographed and stored in 50% glycerol/PBST.
Analysis of partial genomic structure of the CiGα i1 gene
The genomic DNA of C. intestinalis was extracted from the sperm of one individual, according to a standard method (Sambrook et al., 1989). A genomic DNA fragment containing exons for both CiGαi1a and CiGαi1b was amplified by PCR using LA Taq with a pair of gene-specific primers 5′-TGGGAGACTGCATGAAACGAAT-3′ (corresponding to nt 520–541 of the CiGαi1a cDNA), and 5′-GCTACACAGAAGATGATAGCAG-3′ (corresponding to nt 808–829 of the CiGαi1a cDNA). The PCR products were cloned into pBluescript II SK (+) (Stratagene). The nucleotide sequence was determined on both strands with an automatic DNA sequencer (Shimadzu DSQ 1000L, Shimadzu).
Isolation and characterization of C. intestinalis cDNAs encoding Gα subunits
Five DNA fragments, each encoding a central part (164–166 aa) of Gα with a distinct amino acid sequence, were amplified from a C. intestinalis larval cDNA library by PCR with the degenerate primers. The Gα isoforms encoded by the cDNA fragments were designated as CiGαx, CiGαq, CiGαi1a, CiGαi1b, and CiGαi2. The deduced amino acid sequences of the five isoforms were shown in Fig. 1.
To investigate the structural and evolutionary relationships among CiGαx, CiGαq, CiGαi1a, CiGαi1b, CiGαi2, and known metazoan Gα isoforms, phylogenetic analysis was performed by the neighbor-joining method (Saitou and Nei, 1987; Fig. 2). CiGαq is most closely related to vertebrate Gαq/Gα11 isoforms. CiGαx is also a member of the Gq class. Within the Gq class, however, CiGαx is fairly diverged from both the Gαq/Gα11 subfamily and the Gα15/Gα16 subfamily. Therefore, CiGαx may represent a novel subfamily, the members of which have not been identified in other animals. CiGαi1a, CiGαi1b, and CiGαi2 were closely related to vertebrate and invertebrate Gαi isoforms. Among these, the C. intestinalis Gαi isoforms CiGαi1a and CiGαi1b are most closely related to each other.
Messenger RNA encoding CiGαi1a and CiGαi1b are produced by alternative splicing from the CiGαi1 gene
Between CiGαi1a and CiGαi1b, only 12 residues were different out of 165 amino acids encoded by the cDNA fragment described above. The diverged positions were limited within a 33-amino acid portion of the polypeptides, and outside this variable region there were few synonymous nucleotide substitutions between their cDNA sequences. These features of CiGαi1a and CiGαi1b indicate the possibility that transcripts for these isoforms are splicing variants of a single gene. To assess this possibility, we first determined the entire cDNA sequences of CiGαi1a and CiGαi1b. The full-length cDNAs encode a 354-amino acid polypeptide, and the predicted amino acid sequences are identical between CiGαi1a and CiGαi1b, except for the 12 positions found by the analysis of the PCR fragments (Fig. 3A). The 5′ and 3′UTR sequences are almost identical between the CiGαi1a and CiGαi1b transcripts (identity >98%, data not shown).
To further confirm that the CiGαi1a and CiGαi1b tran-scripts originate from a single gene, we then examined the structure of the gene. Genomic DNA containing the coding regions that had different sequences between CiGαi1a and CiGαi1b were amplified from the C. intestinalis sperm DNA by PCR. A 2.2-kb genomic DNA fragment was amplified and cloned. The entire nucleotide sequence of the genomic DNA fragment revealed that two homologous exons were tandemly aligned on the same DNA strand (Fig. 3B). One of the exons that locates upstream to the other encodes the variable region of CiGαi1a, and the downstream exon encodes that of CiGαi1b. The positions of the introns adjacent to the isoform-specific exons of CiGαi1a/b are conserved with respect to the Gα sequences in mammals and insects (Fig. 3A,B). The genomic organization of the CiGαi1 gene strongly suggests that the CiGαi1a and CiGαi1b mRNAs are produced by alternative splicing.
Expression patterns of Gα mRNAs in C. intestinalis embryos
The expression patterns of mRNAs for C. intestinalis Gα isoforms were examined in tailbud embryos by whole-mount in situ hybridization (Fig. 4). The CiGαx expression was widely observed throughout the trunk and tail of the embryos, and the signals were stronger in the epidermis, mesenchyme, and tail muscle cells (Fig. 4A,B). The CiGαq transcripts were expressed in the anterior and dorsal trunk epidermis as well as the dorsal side of the tip of the tail (Fig. 4C). The anterior CiGαq-expressing regions seem to contain the developing adhesive organ. On the dorsal trunk epidermis, the CiGαq-expressing regions were bilaterally located as two pairs of patches (Fig. 4D). Both of the two splicing variants of CiGαi1, CiGαi1a, and CiGαi1b, were specifically expressed in the brain and adhesive organ (Fig. 4E, F). A difference in expression patterns was not clear between CiGαi1a and CiGαi1b, although the hybridization signals were much weaker for CiGαi1b. We failed to detect clear hybridization signals for CiGαi2, even after observing the staining reaction for three days; only faint signals were observed in the brain (Fig. 4G).
The distinct expression of the CiGαi1 gene in the nervous system of tailbud embryos prompted us to further examine the expression patterns of this gene throughout embryonic development. Probably due to the small size (136 bp) of the splice variant-specific probes, the hybridization signals were very weak, especially for CiGαi1b (Fig. 4E,F). Therefore, we used a longer probe that can be hybridized with both CiGαi1a and CiGαi1b mRNAs (see Materials and Methods). The CiGαi1 transcripts were present in the eggs and cleavage stage embryos as maternal messages (Fig. 5B–F). The transcripts were ubiquitously distributed in the eggs and early embryos (Fig. 5B–G). The ubiquitous distribution of the CiGαi1 mRNA was observed until the gastrula stage (Fig. 5G). At the neurula stage, the CiGαi1 expression was restricted to the anterior ectoderm, especially the presumptive brain vesicle, although the hybridization signal was ambiguous (Fig. 5H). The CiGαi1 expression became restricted to the palps (adhesive organ), the central nervous system (CNS), including the brain vesicle and the visceral ganglion, and the dorso-distal part of the tail (Fig. 5I–K). At the mid tailbud stage, the hybridization signals were very strong and were restricted to the palps, the entire brain vesicle, and the visceral ganglion (Fig. 5L,M). The gene expression persisted to the CNS until the tadpole larva stage (Fig. 5N).
Diversity of Gα isoforms in ascidians
In the present study, we showed that mRNAs encoding at least five different Gα isoforms are present in ascidian embryos. Mammals have at least 17 functional Gα genes, several of which are spliced alternatively, that encode 23 distinct protein products (Nürnberg et al., 1995). The presence of multiple Gα isoforms has also been demonstrated in various invertebrates, including insects, nematodes, octopus, hydra, and sponges (Wilkie and Yokoyama, 1994; Suga et al., 1999; Iwasa et al., 2000). Metazoan Gα sub-units can be classified into Gs, Gi, Gq, and G12 classes (Simon et al., 1991; Suga et al., 1999). The present analysis identified two members of the Gq class and three of the Giclass in C. intestinalis embryos. Although we have not found Gαs isoforms in Ciona, we recently identified a cDNA clone encoding Gα closely related to vertebrate Gαs isoforms in another ascidian species, Halocynthia roretzi (Iwasa et al., 2001). To date, however, no Gα isoforms have been assigned to the G12 class in ascidians.
Among the Gq class members, CiGαq is closely related to the vertebrate Gαq and Gα11. However, the vertebrate Gαq and Gα11 are more closely related to each other than to CiGαq, suggesting that the vertebrate Gαq and Gα11 iso-forms originated by gene duplication during vertebrate evolution after the divergence between the vertebrate and the urochordate. The metazoan Gq-class isoforms are further classified into Gαq and Gα15/16 subfamilies. Another Ciona Gq-class isoform, CiGαx, is fairly diverged both from the Gαqand Gα15/16 subfamilies. Therefore, CiGαx may be a member of a novel Gα subfamily. It will be interesting to see whether Gα closely related to CiGαx, and is present in other animals, including vertebrates.
Based on their primary structure, Gi-class isoforms are further classified into four distinct subfamilies: Gαi, Gαo, Gαt, and Gαz (Suga et al., 1999). All three Gi class members of Ciona belong to the Gαi subfamily, and so far no ascidian Gα isoforms have been assigned to the Gαt, Gαo, and Gαz subfamilies. Since Gαo isoforms have been reported in diverse invertebrate phyla, including sponges (Suga et al., 1999), arthropods (Thambi et al., 1989; Horgan et al., 1995), and molluscs (Kojima et al., 1997; Iwasa et al., 2000), it is likely that ascidians also have this isoform. Go is abundant in the CNS both in vertebrates (Strathmann et al., 1990) and in insects (Thambi et al., 1989; Horgan et al., 1995), and play important roles in the function and development of the nervous systems (Offermanns 2001). Therefore, future studies are needed to clarify whether Gαo isoforms are present in ascidians.
Alternative splicing of CiGα i1 transcripts
A comparison of the CiGαi1a and CiGαi1b cDNA sequences and the genomic structure of the CiGαi1 gene strongly suggest that CiGαi1a and CiGαi1b are products of alternative splicing. In vertebrate, splice variants of Gαs, Gαi, and Gαo are known. Alternative splicing of the mammalian Gαo gene generates two different isoforms of Gα (Tsukamoto et al., 1991). These isoforms, Gαo1 and Gαo2, are identical to each other in the N-terminal 248 amino acids; the sequences thereafter diverge. Gαo1 and Gαo2 result from alternative splicing of exons 7 and 8. The diverged carboxyl-terminus contains an effector-interacting domain. Therefore, Gαo1 and Gαo2 exhibit different properties in signal transduction (Kleuss et al., 1991). There are four splice variants of mammalian Gαs (Kozasa et al., 1988). These Gαs variants directly activate adenylate cyclases and calcium channels (Mattera et al., 1989). The relative proportion and tissue distribution of the two variants of Gαs change during cellular differentiation, development, aging, and adaptive processes (Ihnatovych et al., 2001). Mammalian Gαi proteins are encoded by three different genes, Gαi1, Gαi2, and Gαi3, which are closely-related to each other (Itoh et al., 1988). Transcripts of the mammalian Gαi2 gene undergo alternative splicing, which in turn give rise to two distinct proteins with different carboxyl-terminal amino acid sequences (Montmayeur and Borrelli 1994). The Gαi2 vari-ants exhibit differential cellular localization and function. Multiple Gα isoforms are also produced by alternative splicing of a single Gαo gene in Drosophila (de Sousa et al., 1989). Therefore, alternative splicing seems to be a common mechanism to produce diversity of Gα isoforms in a wide variety of animals. Interestingly, the way alternative splicing occurs differs between CiGαi1 and vertebrate Gα genes, suggesting that these alternative splicing events evolved independently in ascidians and in vertebrates. In future studies, it will be very important to investigate and learn the functional differences between CiGαi1a and CiGαi1b in the regulation of signal transduction during ascidian development.
Spatially restricted expression of Ciona Gα genes and roles of G-protein signaling in ascidian embryos and larvae
The present study demonstrated that multiple Gα genes are expressed with distinct expression patterns in Ciona embryos. Although autonomous cell-fate specification is a dominant mechanism in ascidian embryos, the importance of cell-cell communications has increasingly become evident in the determination and differentiation of embryonic cells in ascidians, especially in their nervous systems (Meinertzhagen and Okamura 2001; Wada and Satoh 2001; Darras and Nishida 2001). It is quite probable that G proteins mediate signaling in these cell-cell interactions during ascidian development. It is also known that G proteins are located on intracellular membranes, and are involved in membrane trafficking and vesicular transport mechanisms of the cell (Nürnberg et al., 1995). Therefore, some of the Ciona Gα isoforms may participate in these cellular activities.
Among the four Gα genes identified in this study, CiGαq and CiGαi1 showed distinct and spatially restricted expression patterns. In the tailbud stage, CiGαq is expressed in the anterior and dorsal trunk epidermis and the tail tip. Trunk regions expressing CiGαq seem to include the future adhesive organ and siphon rudiments (Nakayama et al., 2001). The CiGαq protein may be involved in cell signaling during development of these organs.
From fertilization to the gastrula stage, CiGαi1 mRNA is present ubiquitously as maternal messages. Therefore, CiGαi1 may mediate signaling between blastomeres during the cleavage stages. Later in embryogenesis, the CiGαi1 expression is restricted to the CNS and the adhesive organ. This expression pattern suggests that CiGαi1 isoforms play important roles in the development and function of the nervous systems which require intracellular signaling in various aspects.
Recently, we have reported that Ciona larvae express a vertebrate-type opsin gene Ciopsin1 in the photoreceptor cells of the ocellus (Kusakabe et al., 2001). In vertebrates, Gt (transducin) are responsible for signal transduction in the photoreceptor cells of the retina by coupling with rhodopsin, while invertebrate opsins activate Gq and Go (Tsuda and Tsuda, 1990). However, our extensive search has failed to find Gαt in the Ciona EST and genome sequence databases. Interestingly, both Gt and Go belong to the Gi class and the vertebrate rhodopsin can activate Gi in vitro (Terakita et al., 2002). We have shown that CiGαi1 is expressed in the brain vesicle including photoreceptor cells. Therefore, CiGαi1 may interact with Ci-opsin1 in phototransduction of the ascidian larval ocellus. Since no Gαt has been reported in invertebrates to date, Gαt may have appeared during early vertebrate evolution after the separation between vertebrates and urochordates.
We thank Dr. Tatsuo Iwasa for useful discussion and critical reading of the manuscript. This work was supported in part by Grants-in-Aid for Scientific Research from the MEXT and JSPS to T. K. and M. T.