Comprehensive analysis of the ascidian genome reveals novel insights into the molecular evolution of ion channel genes

Yasushi Okamura, Atsuo Nishino, Yoshimichi Murata, Koichi Nakajo, Hirohide Iwasaki, Yukio Ohtsuka, Motoko Tanaka-Kunishima, Nobuyuki Takahashi, Yuji Hara, Takashi Yoshida, Motohiro Nishida, Haruo Okado, Hirofumi Watari, Ian A. Meinertzhagen, Nori Satoh, Kunitaro Takahashi, Yutaka Satou, Yasunobu Okada, Yasuo Mori


Ion fluxes through membrane ion channels play crucial roles both in neuronal signaling and the homeostatic control of body electrolytes. Despite our knowledge about the respective ion channels, just how diversification of ion channel genes underlies adaptation of animals to the physical environment remains unknown. Here we systematically survey up to 160 putative ion channel genes in the genome of Ciona intestinalis and compare them with corresponding gene sets from the genomes of the nematode Chaenorhabditis elegans, the fruit fly Drosophila melanogaster, and the more closely related genomes of vertebrates. Ciona has a set of so-called “prototype” genes for ion channels regulating neuronal excitability, or for neurotransmitter receptors, suggesting that genes responsible for neuronal signaling in mammals appear to have diversified mainly via gene duplications of the more restricted members of ancestral genomes before the ascidian/vertebrate divergence. Most genes responsible for modulation of neuronal excitability and pain sensation are absent from the ascidian genome, suggesting that these genes arose after the divergence of urochordates. In contrast, the divergent genes encoding connexins, transient receptor potential-related channels and chloride channels, channels involved rather in homeostatic control, indicate gene duplication events unique to the ascidian lineage. Because several invertebrate-unique channel genes exist in Ciona genome, the crown group of extant vertebrates not only acquired novel channel genes via gene/genome duplications but also discarded some ancient genes that have persisted in invertebrates. Such genome-wide information of ion channel genes in basal chordates enables us to begin correlating the innovation and remodeling of genes with the adaptation of more recent chordates to their physical environment.

  • ascidian
  • homeostasis
  • embryogenesis

ion channels, membrane proteins that regulate ion fluxes, provide the molecular bases for diverse physiological phenomena, including neuron function, respiration, absorption, secretion, and the osmotic and volume regulation of body fluids. Intensive studies in the last two decades have led to the identification of the molecular nature and detailed biophysical mechanisms of a wide variety of ion channels. However, there still remains a large gap between the molecular properties of ion channels and their macroscopic physiological functions. Moreover, little is known about the evolutionary history of ion channel genes and how these may have become remodeled in the chordate lineage before assuming their vertebrate functions.

Adult ascidians are marine sessile invertebrates, but the initial stage in their metamorphic life cycle, the tadpole larva, has a body form with many deep similarities to that of vertebrates. The larval nervous system in particular is both dorsal and tubular, like the vertebrate neural tube, and similarly shows an inductive mode of formation during its development (42). The central nervous system (CNS) of the ascidian larva has been claimed to contain <100 neurons (48), the cell-lineage and positions of which have recently been precisely mapped (9). Later in its life cycle, providing an unusual biological feature that is characteristic of ascidians, the larva undergoes a radical metamorphosis to change its body plan and its behavior, transforming in a short period from a motile swimming form to a sessile filter feeder. In the process, it acquires many adult-specific organs, such as the heart, pharyngeal gill slits, and endostyle (16), that are thought also to reflect the evolutionary origins of vertebrate organs. The developmental programs for these organs are proposed to have become decoupled during evolution and in current forms to be represented at different stages of the ascidian life cycle, in the larva and adult (20).

Recently, the entire genome of one species of ascidian, Ciona intestinalis, has been sequenced (11). The draft Ciona genome contains ∼16,000 protein-coding genes, a number which is intermediate between those of protostome and vertebrate genomes. More than 10% of the genes are specific to chordates, whereas 63% are common to both chordates and protostomes. The Ciona genome is <160 Mb, which is more compact and dense (7.5 kb/gene) than the genomes of either Drosophila (9 kb/gene) or the human (100 kb/gene). This recommends ascidians as an excellent model to reveal gene networks by systematic analysis of promoters. Combined with the wealth of information recently obtained about gene expression profiles, provided especially from expressed sequence tag (EST) [Ref. 58 and Ghost Database (a C. intestinalis cDNA resource):] and in situ hybridization data (50), several groups of genes from the Ciona genome have been surveyed systematically, and their relations with vertebrate homologs have been analyzed. These genes include those related to the cytoskeleton, muscle contraction, signal transduction, cell interactions, and transcription (57). With the background provided by findings from classical physiology (16), when combined with this recent flood of genomic information, ascidians provide a unique opportunity to gain insights into the origin of vertebrate physiological functions.

Here we have systematically surveyed the putative ion channel genes in the genome of C. intestinalis and compared them with the comparable gene sets of Chaenorhabditis elegans, Drosophila melanogaster, and the human genomes. In addition, we have also referred to the genome of the Japanese puffer fish Fugu rubripes for some ion channel genes. Not surprisingly, we find that the Ciona genome contains a minimum set of prototype genes for neuronal signaling, voltage-gated cation channels and transmitter-gated ion channels. Channel genes related on the other hand to homeostatic control, such as transient receptor potential (TRP) and connexins, comprise not only vertebrate prototype genes but also multiple genes that are more specialized and belong to ascidian-specific clades. Ion channel genes for fine-tuning of excitability and pain-related channels are not present in the Ciona genome.


Retrieving Sequences from the C. intestinalis Genome and a cDNA/EST Database

The C. intestinalis protein sequences were tblastn searched against the draft genome sequence [Ref. 11 and DOE Joint Genome Institute (JGI) Ciona intestinalis v1.0 (JGI site for the complete C. intestinalis genome sequence and gene annotations):] and a cDNA/EST database (Ref. 58 and Ghost Database: using human, Drosophila, and C. elegans channel proteins (38). Channel proteins were identified using the following basic method (57). Briefly, when the corresponding cDNA sequence covering the diagnostic sequences for an ion channel molecule (such as a channel pore region or a transmembrane region) was available by InterPro search, the deduced protein sequence was used for the analyses. In surveying auxiliary subunits for ion channels, the protein sequence was analyzed when it showed significant overall homology with previously known subunits in the blast search. When the cDNA sequence was not available and grailEXP or genewise confidently predicted the gene encompassing the entire channel region, the peptide sequence deduced from the gene model was used. When the predicted gene model was not perfect, but the ESTs either covered the entire region or the region that the gene model lacked, the peptide sequence was deduced from the assembled sequence obtained using either a set of ESTs (5′ and 3′ EST pair), multiple sets of ESTs, or both an EST and the gene model. All analyzed Ciona genes are listed in Supplemental Table S1 (available at the Physiological Genomics web site).1

Molecular Phylogenetic Analysis

The sequences were aligned using the CLUSTALX 1.83 program (18). We also used another program, MUSCLE (12), which is based on the algorism with smaller possibility of unreliable alignment between distantly related sequences. Both alignment programs gave similar results. Alignments in nonconserved extracellular regions of channel proteins were carefully checked by eye, and regions with ambiguous alignments, in particular, at less conserved cytoplasmic region and extracellular region, were eliminated. Thus verified alignments were used to construct phylogenetic trees both through the neighbor-joining (NJ) method and maximum-likelihood (ML) analysis for the molecules depicted in Figs. 13. For the NJ method, the trees were calculated using the MEGA program (55). The same fasta files were analyzed with the ML method using TREE-PUZZLE 5.0 (59). For the analyses shown in Figs. 13, the NJ and ML trees were compared, and branch points were taken significant only when both analyses gave significant values >50%. Trees of ML analysis following sequence alignment using Clustal program are shown as Supplemental Figs. S4–S6. For the molecules depicted in Supplemental Figs. S1–S3, only NJ analysis was performed. The sequences used are designated in succession by: the accession number, the abbreviation of the species, and the gene name. Abbreviations of the species are as follows: HS for human, DM for D.melanogaster, CE for C. elegans, GG for Gallus gallus, AC for Aplysia californica, EE for Electrophorus electricus, LB for Loligo bleekeri, HR for Halocynthia roretzi, AG for Anopheles gambiae, HC for Haemonchus contortus, DR for Danio rerio, XL for Xenopus lavies, CC for Cyanea capillata, BT for Bos taurus, SP for Strongylocentrotus purpuratus, OB for Oceanobacillus, BH for Bacillus halodurans, AT for Arabidopsis thaliana, and Hal for Halobacterium sp. NRC-1.

Fig. 1.

Phylogenetic trees by the neighbor-joining (NJ) method for voltage sensor-containing channels. A: Nav channel α-subunit. B: Cav channel α1-subunit. C: Kv1–4 channel α-subunit. D: KCNQ channel α-subunit. E: cyclic nucleotide-sensitive cation channel genes. Branch points supported by both the NJ and maximum-likelihood (ML) methods are shown by red circles. Bootstrap values of NJ and ML analyses are indicated in black and red letters, respectively.

Best-Hit Gene Analysis

We compared the indicated Ciona proteins with the International Protein Index human proteome. At first, identified proteins were compared using the blastp program (1). The best-hit protein in the human proteome was then blastp searched against the Ciona version 1 proteome without the option of gapped alignment. When the best-hit sequence of the human protein corresponded to the region encoding the starting Ciona protein, the relationship between the two proteins was called the “bidirectional best-hit relationship.” Otherwise, it was called a “unidirectional best-hit relationship.” To define the orthology between the two proteins, either bidirectional best-hit relationship or a bootstrap value of >50% in both NJ and ML methods was taken as the criteria.


A survey of the C. intestinalis genome identified 160 genes with homology to ion channels. Ion channel-like genes were grouped into several classes: voltage sensor-containing cation channels, transmitter-gated channels, anion channels, non-voltage-gated cation channels, intracellular calcium channels, and intercellular channels. Channel gene orthologs were identified as genes that showed bootstrap values >50% at the diagnostic branch points in the phylogenetic trees or, alternatively, a bidirectional best-hit relationship in the best-hit gene analysis (see also materials and methods).

Voltage Sensor-Containing Cation Channels

Main channel-forming subunits.

Four putative genes coding for α-subunit proteins of voltage-gated sodium, Nav, channels exist in the Ciona genome (Fig. 1A). One of these (Ci-Nav1) is a counterpart of TuNa1, a neuronal Nav channel gene previously characterized in another species of ascidian, H. roretzi (53). This gene is most closely related to the group of vertebrate Nav channels. The other three genes (Ci-Nav2, Ci-Nav3, Ci-Nav4) are only distantly related to the authentic chordate branch of the Nav channels. One of the three nonconventional Nav channels (Ci-Nav2) shows a high level of sequence similarity to the ascidian cation channel, TuNa2, which is predominantly expressed in Halocynthia larval motor neurons (46). Its cDNA expression profile (58) indicates that this gene is expressed in both the nervous system and endostyle in the adult ascidian. The products of both genes, Ci-Nav2 and TuNa2, share two unusual features in their primary structure: first, the lysine residue critical for sodium selectivity in the pore region of the domain III is replaced by glutamic acid in TuNa2; and the amino acid sequences of the III-IV linker, a region known to function as the inactivation ball (19), are not well conserved. These two features of the primary structure are shared by other Nav-like genes from three species of protostome: Drosophila, the cockroach, and squid. Recent biophysical studies have shown that, in the cockroach, the gene of this invertebrate-specific clade encodes a slowly inactivating calcium-permeable channel, suggesting that Ci-Nav2 could be permeable to both sodium and calcium (70). The other two Nav channels (Ci-Nav3, Ci-Nav4) are related to each other and located at a tree position that is intermediate between those for the protostome and vertebrate branches of the molecular phylogeny (Fig. 1A).

Ciona has single genes for the three main subclasses of voltage-gated calcium (Cav) channels: Ci-Cav1, Ci-Cav2, and Ci-Cav3 (Fig. 1B). The presence of representatives from all three channel types is consistent with a previous view that the diversification of these three predated the separation of deuterostome and protostome animals (38) (Supplemental Fig. S1). Because there are multiple paralogous genes belonging to each of the subfamilies of Cav1–3 in the Fugu genome and the human genome [Fig. 1B, Supplemental Fig. S1, and JGI Fugu rubripes v3.0 (JGI Fugu genome site):], duplications within each Cav subfamily (for example, into Cav1.1, Cav1.2, Cav1.3, Cav1.4) must have been established somewhere between the branch points of the urochordate and the shared ancestor of teleosts and mammals. In addition to these three prototype Cav channel genes, Ciona has several genes that are homologous to mammalian “Cav-like channels,” the functions of which remain to be determined. One is the four domain-type channel (35) recently called α1U, which is highly conserved in the Drosophila, C. elegans, and vertebrate genomes. Two other genes, Ci-TPC1 and Ci-TPC2, putatively encode the two-domain type channel, which has been reported in the human genome but in neither the Drosophila nor C. elegans genomes (Supplemental Fig. S2). Ciona has three genes closely related to the one-domain type of voltage-gated cation channel, CatSper, which is known to play a role in sperm motility (8) (Supplemental Fig. S2).

A minimum set of genes also seems to be present for voltage-gated potassium channel (Kv) proteins (Fig. 1C). Ciona has three Kv1, one Kv2, one Kv4, one gene distantly related to all Kv channels, and two KCNQ genes (Fig. 1D). Unexpected position of human Kv4 at the base outside the clade of Kv4-related genes (Fig. 1C) was not supported when only a group of Kv4-related genes was aligned independently (data not shown). No Kv3-like gene was found in the Ciona genome. Ciona has genes for the calcium-activated BK, Slack, and SK potassium channels (Supplemental Fig. S2, Table 1).

View this table:
Table 1.

Comparison of nos. of ion channel subunit genes among Drosophila, C. elegans, C. intestinalis, and H. sapiens

The CNG-ch, HCN-ch, and eag/erg/elk channel genes all belong to a large family of genes putatively encoding six-transmembrane cation channel proteins that contain an S4-like voltage-sensor region and a nucleotide-binding region at their COOH terminus (19). Ciona has a single gene for each of the erg, elk, and eag subfamilies (Ci-erg, Ci-elk, Ci-eag). The CNG channel genes number three in the Ciona genome compared with four in the Drosophila and six in the C. elegans genome (Table 1, Fig. 1E, Ci-CNG1 to -3). There are three HCN-like genes in the Ciona genome, whereas there is only a single gene in the Drosophila genome and none in C. elegans (Table 1, Fig. 1E). The Ciona genome also has four other genes (Ci-eag/HCN/CNG-diverged1–4 distantly related to known CNG/HCN/eag, erg, elk channels) (Fig. 1E).

Auxiliary subunits.

Auxiliary subunits modify the gating kinetics, localization, and expression levels of the α-subunits of voltage-gated channels and are therefore functionally important modifiers of electrical excitability. The Ciona genome contains β-, γ-, and α2/δ-subunit genes for Cav channels (Supplemental Fig. S2). For auxiliary subunits at voltage-gated potassium channels, Ciona has a single Kv β-subunit-like gene and two genes putatively encoding BK channel β-subunits (Table 1 and Supplemental Fig. S2), and at least one putative KChip gene encoding a calcium-binding auxiliary subunit for Kv4-class channel proteins (Ref. 2 and data not shown). These genes are present in neither the Drosophila nor the C. elegans genomes. By contrast, Ciona lacks many other auxiliary subunit genes. Vertebrate Nav β-subunits support the clustering of Nav channel proteins at the node of Ranvier and initial segments of vertebrate myelinated nerve fibers, through interactions with myelin-forming glial cells. The absence of a Nav β-subunit gene in the Ciona genome (Table 1) is therefore consistent with the lack of both myelin-related genes in the genome (11) and myelin-like structures in ascidian nerves (30). The Ciona genome also lacks a TipE gene, which is known to substitute for the role of Nav-β in Drosophila (21). The Ciona genome also lacks two other classes of auxiliary subunits for potassium channels: first, KCNQE genes including Mirp and MiniK genes for fine-tuning the gating of KCNQ-type potassium and erg channels; and, second, genes for Kv5,6,8,9 subunits, which are known to modulate the function of Kv2 channels (Table 1).

To summarize, Ciona has a simple set of prototype gene homologs similar to those for vertebrate voltage-gated channels but lacks the Kv3 gene and most of the auxiliary subunit genes.

Transmitter-Gated Ion Channels

GABA, glycine, glutamate, and acetylcholine (ACh) are all common chemical neurotransmitters in both vertebrate and invertebrate nervous systems. Despite the commonality of their neurotransmitter ligands, previous studies suggest that vertebrates and invertebrates do not necessarily share all forms of neurotransmitter receptors. For example, C. elegans and Drosophila have both histamine- and glutamate-gated Cl channels, but they lack G protein-coupled receptors, which are present in vertebrates (54). Moreover, vertebrates have serotonin-gated cation channels, glycine receptors, and ATP-gated channels that do not exist in the genomes of C. elegans and Drosophila.

Nicotinic ACh receptor family.

Two types of subunits of nicotinic ACh (nACh) receptors (nAChR) are known: α-subunits, with the ability to bind ACh, and non-α-subunits, which lack such binding (63). Ciona has four nAChR α-like genes and four non-α-subunit genes (Fig. 2A). All four α-like genes show two neighboring cysteines in the extracellular or C-loop, which is essential for ACh-binding activity (63). One Ciona α-like gene, Ci-nAChR-A1, clusters with the α-subunit genes of the nAChR receptor of vertebrate skeletal muscle in the molecular phylogenetic tree (Fig. 2A). This is consistent with the finding that ascidian larval muscle is known to express cation-permeable cholinergic receptors (51), resembling pharmacologically those at vertebrate skeletal muscle. Two Ciona α-like genes, Ci-nAChR-A7/8-1 and -2, cluster with the α7- and α8-subunit mammal genes, which are bungarotoxin-sensitive neuronal nAChR receptors, and with the nAChR receptor genes of C. elegans and Drosophila (Fig. 2A). This indicates that the bifurcation between neuronal and skeletal muscle-type α-nAChR subunit genes occurred early, before urochordates diverged from the other chordates. Ciona has a gene closely related to the α3-subunit, Ci-nAChR-A3, that is present in the genomes of neither Drosophila nor C. elegans. All four non-α-subunit genes in the Ciona genome lack the two neighboring cysteines in the C-loop region for ACh binding, consistent with the idea that they must coassemble with α-subunits to be functional. One gene is closely related to the β2- and β4-subunit genes of vertebrates, designated Ci-nAChR-B2/4, while three non-α-subunit genes are grouped just outside the non-α-subunit genes of vertebrate skeletal muscle nAChR (Fig. 2A, Ci-nAChR-B/G/D/E1 to -3). The pentameric nAChRs of vertebrate skeletal muscle contain four non-α-AChR subunits (β, γ, δ, and ε) (19). Without evidence of gene expression in the ascidian muscle lineage cells, it would be premature to conclude that the three Ciona non-α-genes are orthologs of vertebrate non-α-skeletal muscle subunits, but it is apparent that the divergence of these Ciona non-α-subunit genes (Ci-nAChR-B/G/D/E1–3) might have occurred independently of diversification of the non-α-subunits for vertebrate skeletal muscle.

Fig. 2.

Phylogenetic trees by the NJ method for transmitter-gated channels. A: ACh receptor (AChR) genes. B: glutamate receptor (GluR) genes. C: GABAA receptor (GABAAR)/glycine receptor (GlyR)-related genes.

In agreement with the presence of α-like nAChR subunit genes and the cholinergic nature of neuromuscular transmission (51), Ciona also has a single gene homologous to rapsyn, which is involved in clustering ACh receptors in postsynaptic domains of vertebrate skeletal muscle (Table 1).

The 5HT3 receptor is a serotonin-gated receptor channel with significant homology to nAChRs that hitherto has been found only in vertebrates. The Ciona genome has three genes related to the vertebrate 5HT3 receptor genes, and they constitute a single clade within the AChR/5HT3R tree (Fig. 2A).

Ionotropic glutamate receptor family.

The Ciona genome has single prototype genes for each of the AMPA receptor (Ci-GluR1/2/3/4) and N-methyl-d-aspartate (NMDA) receptor (Ci-GluR-NR1 and -NR2) types of ionotropic glutamate receptor (GluR), as supported both by the clear bidirectional best-hit relationship and by their high bootstrap values (Fig. 2B). There is one kainate receptor-like gene (Ci-GluR-KaiR-like). However, this similarity to kainite receptor gene was not significantly supported by the ML method. Ciona has two genes related to the δ-subunit subfamily, Ci-GluR-Delta1 and -2. Ci-GluR-Delta2 showed a bidirectional best-hit relationship, whereas Ci-GluR-Delta1 showed a unidirectional best-hit relationship. The presence of δ2-like subunit gene in Ciona was unexpected, because the δ2-subunit is expressed predominantly in cerebellar Purkinje neurons in mammals (39), whereas no structural counterpart of the cerebellum has been identified in the ascidian nervous system.

A remarkable feature of vertebrate glutamate receptors is that the critical site for Ca2+ permeability of several of the GluR subtypes is subject to RNA editing, which has been most intensively characterized in the AMPA receptor genes. Conversion through RNA editing of Q to R in the M2 domain occurs in transcripts of vertebrate GluR subunits and eliminates the Ca2+ permeability of the receptor (60). The genome sequence of the putative AMPA receptor-like gene of Ciona shows that the site corresponding to the Q/R site of the mammalian AMPA receptors is also Q. Detailed comparison between the genome and cDNA sequences will clarify whether RNA editing also underlies conversion of the Ca2+ permeability of GluR channels in the urochordate.

In addition to the above GluR genes, Ciona has three other GluR-like genes with lower homology to vertebrate GluR channel genes, and these form a Ciona-specific clade (Fig. 2B; Ci-GluR-Div1 to -3). This separate ascidian gene group suggests that gene diversity among ascidian GluR-like genes was established independently of the diversification of GluR genes in vertebrate evolution, i.e., that it occurred after the urochordate/vertebrate split and that the counterparts were lost in the vertebrate lineage. Two other Ciona genes (Ci-GluR-Div4 and -5) seem distantly related to vertebrate NMDA receptor genes, although this relationship was not supported by the ML tree (Fig. 2B).

Finally, Ciona has two genes putatively encoding a GRIP-like protein and PSD-95, which are involved in the scaffolding of GluR at the postsynaptic membrane and which play roles in the processing of synaptic plasticity (data not shown).

Ionotropic GABA/glycine receptor family.

There are in total eight ionotropic GABA/glycine receptor (iGABA/GlyR) genes in the Ciona genome, including one GlyR-related gene and seven others, all GABAR-like genes; the seven include one α/γ/ε-related gene, one π-related gene, one β-related gene, and four ρ-related genes (Table 1, Fig. 2C). The presence of multiple subfamilies of genes for iGABA/GlyR genes in the Ciona genome agrees well with the gene diversities in Drosophila and C. elegans, suggesting that main diversifications of the iGABA/GlyR receptors predated the deuterostome/protostome bifurcation and also implying that GABA and glycine may both have acted as signaling molecules in basal metazoan ancestors. However, there are two remarkably distinct aspects of the Ciona genes that distinguish them from the iGABA/GlyR genes of Drosophila and C. elegans. First, one Ciona gene, Ci-GlyR, is grouped with vertebrate GlyR genes but with neither the genes for glutamate-gated Cl channels (GluCl) nor histamine-gated Cl channels, both of which are so far unidentified in vertebrates (Fig. 2C). Second, the Ciona genome contains four ρ-related subunits (Ci-GABAAR ρ1 to -4), subfamilies that are not found in the Drosophila and C. elegans genomes. All eight iGABA/GlyR genes putatively encode Cl-permeable channels, since their sequences encode conserved amino acids of the pore-forming region, including the cytoplasmic rings, intermediate rings, and extracellular rings at the M1 and M2 region, which are known to determine anion selectivity (Ref. 28 and data not shown).

Even though there is only a single GlyR-like gene, there are three genes present in the Ciona genome putatively encoding gephyrin (Table 1), which induces clustering of GlyR in vertebrate neurons. The fact that gephyrin genes outnumber the single GlyR gene in Ciona raises the possibility that ascidian gephyrins might have roles other than that of regulating the distribution of glycine receptors.

The Ciona genome does not contain the purinergic receptor channel P2X (Table 1). Correspondingly, we also could not find any gene homologous to the G protein-coupled type of purinergic receptor, P2Y, in the Ciona genome, consistent with the idea that neither ATP nor UTP are utilized as extracellular signaling molecules in ascidians.

To summarize, just as for voltage-gated channel genes, the Ciona genome contains conserved sets of prototype genes for most of the subfamilies of vertebrate transmitter-gated channels. Overall, the number of transmitter-gated channel genes is fewer in Ciona than it is in either C. elegans or Drosophila (Table 1). Our findings also suggest that the basic combinations of transmitter-gated channels found in vertebrates (the presence of GlyR and 5HT3R but the lack of GluCl and histamine-gated channels) were already established in ancestral forms before the divergence of urochordates.

Non-Voltage-Gated Cation and Anion Channels

Inwardly rectifying K+ channel and two-pore domain K+ channels.

Inwardly rectifying K+ channel (Kir) and two-pore domain K+ channel (K2P) genes encode important channels for determining the resting membrane potential and for the transport of potassium ions. Ciona has four Kir-like genes, including one gene (Ci-Kir2/5) related to the classical Kir subfamily (Kir2), two genes (Ci-GIRKA, Ci-GIRKB) related to Kir3 (GIRK), which is activated by a G protein, and one gene (Ci-Kir1/4/7) loosely related to subfamily of Kir1/4/7 (Table 1, Supplemental Fig. S3). Such diversity is found neither in C. elegans nor in Drosophila (Table 1), suggesting that these prototypes of the different Kir subfamilies emerged before the appearance of the urochordate line and probably after the divergence between protostomes and deuterostomes. On the other hand, the Ciona genome lacks a gene for the ATP-sensitive K+ channel subfamily, which is known in vertebrates to regulate excitability as a function of the metabolic state in muscle, brain, and endocrine cells (Table 1). Accordingly, there is no homolog in the Ciona genome of the SUR-type ATP-binding cassette (ABC) transporter gene, which coassembles with ATP-sensitive K+ channels in vertebrates.

K2P, the non-voltage-gated K+ channel consisting of two domains, also plays a critical role in determining the resting membrane potential. The Ciona genome has a total of five K2P-like genes, Ci-TWIK1 to -5, classified in two subgroups that are conserved among the genomes of Drosophila, C. elegans, and vertebrates. These five include four genes in the TWIK1-related clade and a single gene in the TASK3-containing clade (Supplemental Fig. S3). The number of Ciona two-domain K+ channel genes is far fewer than in the genomes of Drosophila (38) and C. elegans (36, 38) (Table 1).

Amiloride-sensitive cation channels.

Amiloride-sensitive cation channels are two-transmembrane sodium-permeable channels. In mammals, amiloride-sensitive cation channels consist of two main subfamilies: the acid-sensing ion channel (ASIC) group, channels expressed in neurons, and the epithelial Na+ channel (ENaC) subfamily, abundantly expressed in epithelia such as in the lung, kidney, and intestine. The Ciona genome contains two genes that belong to the ASIC branch (Ci-ASIC/degenerin6,7) and five additional genes (Ci-ASIC/degenerin1–5)that show only weak homology to amiloride-sensitive cation channels (Fig. 3A). One of the five, Ci-ASIC/degenerin3, showed a bidirectional best-hit relationship with the human amiloride-sensitive cation channel gene, ACCN5, supporting the conclusion that it encodes an amiloride-sensitive cation channel. The C. elegans and Drosophila genomes also show species-specific branches of putative amiloride-sensitive cation channel genes, but the five Ciona genes with weak homology to amiloride-sensitive cation channel genes lack any close relationship to the genes of the C. elegans and Drosophila clades, indicating that the diversities for this channel group of deuterostomes and protostomes are derived independently. Although NJ-based tree suggested that these five genes were weakly related to the vertebrate ENaC genes (Fig. 3A), it is unlikely that Ci-ASIC/degenerin1–5 correspond to the orthologs of ENaC, since neither alignment with MUSCLE nor tree formation with the ML method supported this relationship. The best-hit gene analysis also did not support the orthology of these genes to the vertebrate ENaC subfamily (Supplemental Table S1).

Fig. 3.

Phylogenetic trees by the NJ method for non-voltage-gated cation channels and anion channels. A: amiloride-sensitive channel genes. B: TRPC channel genes. C: polycystic kidney disease (PKD)-related channel genes. D: transient receptor potential (TRP) and TRPN/A channel genes. E: CLCA channel genes.

TRP-related channels.

TRP-related channels are six-transmembrane cation channels that respond to various sensory modalities, such as temperature sensation, pain sensation, and osmotic stress (8). Exceeding the number found even in the human genome, Ciona has a total of 27 TRP-related genes (Table 1), with representatives covering all four known subfamilies, TRPM, TRPC, TRPV/N/A, and TRPP [or polycystic kidney disease (PKD)-related channel] (Fig. 3, B–D; see Supplemental Fig. S3). There are two TRPM-like genes, one of which, Ci-TRPM2/4/8, codes for NUDT9, a Nudix hydrolase domain corresponding to that found in the mammalian TRPM2 channel (Supplemental Fig. S3). The Ciona genome has a total of eight TRPC-like genes (see Fig. 3B). One of these is grouped with vertebrate members, whereas the other seven genes are located outside the vertebrate branch. The single gene grouped with vertebrate TRPC genes (Ci-TRPC4/5) codes for a TRPC-specific TRP box sequence, EWKEAR, at the COOH terminus. This motif is well conserved in the other seven TRPC-like genes, although alanine is replaced by H or Y or Q. Of the seven genes located outside the mammalian TRPC channel genes, five contain multiple ankyrin repeats at the NH2 terminus, as also do mammalian TRPC channels (Fig. 3B).

The Ciona genome has an unusually large diversity of PKD-related channel (TRPP) genes, nine in all (Table 1, Fig. 3C). The vertebrate TRPP subfamily consists of three subclasses, PKD1, PKD2, and the mucolipin-related gene, all of which were first identified as genes responsible for human genetic disorders. The Ciona genome seems to contain counterparts both for PKD1 and PKD2 (Ci-PKD1 and Ci-PKD2, respectively) that are known to coassemble to form Ca2+-permeable channels in the kidney (17) (Fig. 3C). Although orthology of Ci-PKD1 to mammalian PKD1 was not significantly indicated by the ML-based tree following alignment with CLUSTAL software (Supplemental Fig. 6C), this was supported by both the bidirectional best-hit analysis and alignment with MUSCLE software followed by NJ-based tree formation. Ciona has only a single mucolipin-like gene (Ci-mucolipin), which fits in between its Drosophila and human homologs, not contradictory to the intermediate position of ascidians. In addition to these prototype genes, the Ciona genome contains six other TRPP-related genes, broadly scattered in the phylogenetic tree (Fig. 3C).

The Ciona genome has two TRPV-like genes, but none of these is grouped with vertebrate members of this family but rather with the “osm-9 channel” gene of C. elegans which is required for osmotic/mechanical sensation (Fig. 3D) (45). One of the two Ciona osm9-related genes, Ci-Osm9-related2, is likely the counterpart of a previously identified osmosensitive channel from another ascidian, H. roretzi (32). Ciona has only a single gene member of the TRPN subfamily, Ci-TRPN, related to the teleost gene NOMPC, which putatively encodes a mechanosensory channel in hair cells and neuromast organs of the lateral line in zebrafish (62) (Fig. 3D). The Ciona genome has four TRPA-like genes with extensive ankyrin-repeats at the COOH terminus, whereas in mammals there is only a single TRPA member (ANKTM1, Fig. 3D), which has recently been claimed as a mechanosensory channel in cochlea (10). It would be interesting to know whether the Ciona TRPV, TRPN, or TRPA-like genes express in peripheral neurons of the larval trunk (65) or in ciliated secondary neurons of the adult coronal organ (7) or other candidate mechanoreceptor neurons.

Cl channels.

The Ciona genome has a number of genes for Cl channels, including the CLIC, CLCN, and CLCA groups (Fig. 3E; see Supplemental Fig. S3). The Ciona CLCN channel genes correspond to the mammalian subfamilies of CLCN3/4/5 and CLCN1/2 and CLCN6/7 (Supplemental Fig. S3). The Ciona genome has seven CLCA genes that exhibit considerable duplications (Fig. 3E), which makes a sharp contrast with the absence of such homologs in Drosophila or C. elegans (Table 1). This suggests that the CLCA family is a specific feature of chordate or deuterostome genomes (Fig. 3E). The Ciona genome has only a single CLIC-like gene (data not shown).

There is no cystic fibrosis transmembrane conductance regulator (CFTR)-like gene found among the numerous ABC transporter genes (Table 1). The Ciona genome also lacks a gene of the CLC-K subfamily, which is specific to the kidney in vertebrates (31). Its absence from Ciona is of course compatible with the absence of any organ like the kidney in ascidians. On the other hand, the Fugu genome also lacks a CLC-K-like gene as well as an ortholog of the ENaC gene (data not shown), even though this fish clearly has a kidney, whereas the kidney of Tilapia, a freshwater teleost, does express a CLC-K gene (44). These suggest that gene diversities for CLC-K and CFTR ion channels that underlie epithelial transports may have become established during the vertebrate divergence and adaptations.

In summary, the gene diversities of TRP, CLCN, and amiloride-sensitive channel genes in Ciona are contributed mainly by multiplications of the genes, apparently in the lineage-specific manner. This diversification in ascidian genes contrasts with the genes of voltage- and transmitter-gated channels, for which the Ciona genome has only a minimum set of prototype genes with a small number of lineage-specific duplications.

Other Channels

Inositol trisphosphate receptor and ryanodine receptors.

The inositol trisphosphate (IP3) receptor and ryanodine receptor channels play important roles in the intracellular release of Ca2+ from internal stores (5), which for example underlies Ca2+ waves in the fertilized Ciona egg (69). There is only a single gene for each of the IP3 and ryanodine receptor families in the Ciona genome (Table 1, Supplemental Fig. S3). On the other hand, Fugu has three IP3 receptor genes, each grouped with the corresponding mammalian subtype gene (Supplemental Fig. S3), indicating that, like the genes for voltage-gated channels, the diversity of IP3 receptors and ryanodine receptors was established during the course of chordate evolution after the divergence of the urochordate lineage.

Gap junctions and water channels.

Connexins serve to mediate the diffusion of small molecules between cells and thereby play many critical roles in intercellular communication during neural development and function (68). As previously reported (56), the Ciona genome has 17 putative connexin genes. Although these Ciona genes all belong to the large family of vertebrate connexins, most are independent of any subclass of vertebrate connexins in the phylogenetic tree (data not shown). As pointed out previously (56), Ciona lacks homologs of the invertebrate innexin gene family and, instead, has two genes with some homology to the human pannexin genes, which revealed another clear deuterostome/protostome distinction (56). Water channels play roles in osmoregulation and the regulation of cell volume in a variety of species from plants to mammals. The Ciona genome contains a total of six aquaporin-like genes, four of these related to human aquaporin-8, namely Ci-AQP8-1 to -3 and Ci-Drip, and one called Ci-AQP3 that is related to human aquaporin-3/7/9/10 (see Supplemental Fig. S3). A sixth gene is located at some distance from the other five genes.

Clustering of Homologous Genes in Tandem

In our survey of ion channel genes, we noticed that multiple homologous genes are often clustered in restricted regions of the genome. These include genes for one pair of GABAA-R genes (Ci-GABAAR-ρ1:Ci-GABAAR-ρ2; see Supplemental Table S1 of the total list of Ciona channels for gene nos.), four eag/HCG/CNG-related K+ channel genes (Ci-eag/HCN/CNG-diverged4:Ci-eag/HCN/CNG-diverged1:Ci-eag/HCN/CNG-diverged3:Ci-eag/HCN/CNG-diverged2), two pairs of amiloride-sensitive channel genes (Ci-ASIC/degenerin6:Ci-ASIC/degenerin7 and Ci-ASIC/degenerin5:Ci-ASIC/degenerin4), a pair of TRP-C-like channel genes (Ci-TRPC-related6:Ci-TRPC-related5), two pairs of connexin genes (Ci-connexin-related-5:Ci-connexin-related-6 and Ci-connexin-related-11:Ci-connexin-related-4), one pair of nAChR genes (Ci-nAChR-B/G/D/E2:Ci-nAChR-B/G/D/E3), and one pair of CLCA Cl channel genes (Ci-CLCA1:Ci-CLCA3). In the case of the genes for the eag/HCN/CNG-related K+ channels, four K+-channel like genes are aligned next to each other in the same direction in tandem with an intergenic region of <2 kb. This arrangement might reflect the origin of four genes from two rounds of duplications from a single ancestral gene. Because these channel species are known to produce channel functions by heterologous subunit assembly, the clustering of homologous genes in tandem in the Ciona genome may reflect an operon-like regulation of a set of genes. The significance of such clustering of ion channel genes must, however, await extensive examination of expression patterns of the corresponding transcripts and studies of promoter regions of the individual genes.

Ion Channel Genes Are Expressed During Embryogenesis

Recently, many examples have shown that ion channel activities play critical roles in the morphogenesis (6) and early development of embryos (41). We searched the C. intestinalis EST database of Kyoto University (; Ref. 58) for ion channel genes, which are expressed in early embryos. Eggs and early embryos express genes putatively encoding the following: TWIK, connexin, Kir2/5, PKD-related channel, eag/HCN/CNG diverged channel, Cav3-related channel, Cav channel α2/δ-, β-subunits, two-domain Ca2+ channel (TPC2), KCa β-subunit, and voltage-dependent Cl channels (Table 2). Early expressions of auxiliary subunits of the gene for Cav channel and Kir2/5 are compatible with the developmental profiles of ion currents previously revealed electrophysiologically (64), in which voltage-gated Ca2+ channels and inward rectifier K+ channels emerge at early stages.

View this table:
Table 2.

Ion channels expressed in early embryogenesis


Ciona Has Restricted Sets of Voltage- and Transmitter-Gated Channels That Are Chordate Prototype Genes

The Ciona genome shows only a rather minimal set of voltage-gated channels that are presumed to represent ancestral prototypes of those also inherited by vertebrates. The number of such genes is roughly as large as those in Drosophila and C. elegans, except that the Ciona genome has more Kv1-like and Nav-like genes. These comparisons suggest that diversification within each subfamily of voltage-gated channels occurred only after the divergence between urochordates and other chordates. In fact, even though fish brains are themselves highly diverse (4), fish genomes already show a pattern of gene diversity for voltage-gated channels that is as great as that in mammalian genomes. For example, the Fugu genome contains prototype genes corresponding to all five of the mammalian genes, Cav1.1 to Cav1.5 [see Supplemental Fig. S2 and the JGI Fugu genome site (], and electric fish genome contains multiple Nav channel genes that are grouped with subfamily members of mammalian Nav channel genes (40). Two rounds of gene duplication events are proposed to have occurred in early chordate evolution (22, 52). Much of the gene diversity in voltage-gated ion channel genes could have arisen by means of these genome duplications, and as a result might have allowed a greater level of diversification among ion channel function in ancestral vertebrates. Further genome comparisons, both between different teleosts and between teleosts and basal acraniates, will be necessary to resolve the details of this diversification with respect to the Ciona genome, however. In the opposite direction in metazoan evolutionary history, before the split between protostomes and deuterostomes, action potentials generated by a sodium conductance mechanism have been reported in sponges, from a level of metazoan organization lacking neurons (37). Thus electrical excitability and behavior apparently preceded the evolution of structural nervous systems and were made possible through the early emergence of voltage-gated ion channels. The molecular homologies of these channels found here with those in the basal metazoans are not known, however, and resolution of this question will again require further comparative genomic analyses.

The Ciona genome also shows a minimal set of transmitter-gated channels that are closely related to vertebrate channels. Even though immunoreactivity of the larval CNS has not been demonstrated to many neurotransmitters (66), there is biochemical or physiological evidence for the utilization of various neurotransmitters in ascidians, including ACh (51), glutamate (14), and 5-hydroxytryptamine (serotonin; 5HT) (49) as well as GABA (Ref. 29 and T. Koropatnick and I. A. Meinertzhagen, unpublished observations). Corresponding to this physiological evidence, our analysis here on gene sequences suggests that Ciona has genes for receptors to ACh, glutamate, and 5HT as well as GABAA receptors (Fig. 2). Some species-dependent variations in the genes for transmitter-gated channels are also seen in Ciona; these include GABAR-ρ, GlyR, 5HT3R, muscle-type nAChR, and the gene for a glutamate receptor of the δ-type, none of which is present in Drosophila or C. elegans, but all of which are present in both the Ciona and human genomes. On the other hand, genes for histamine-gated and glutamate-dependent Cl channels are absent from both the Ciona and human genomes. The absence of histamine-gated channels is in accordance with the absence from the Ciona genome of histidine decarboxylase, the gene encoding the synthetic enzyme for histamine (11).

Such differences in the spectrum of transmitter-gated receptor channels are in accordance with both the diversity of neurotransmitter substances in protostomes and deuterostomes and the ancient provenance of those substances. In particular, the finding that the Ciona genome contains genes that putatively code for receptor channels gated by serotonin and glycine indicates that the establishment of systems utilizing serotonin and glycine for fast synaptic transmission predated the origin of urochordates. Comparative pregenomic evidence has previously been interpreted to indicate that certain amine and amino acid ligands may already have arisen as signaling molecules in unicellular forms (3), even though their presence in forms earlier than true metazoans is, by itself, no evidence for neurotransmitter action. In this connection, it is noteworthy that C. elegans GluCl β-subunit can also be gated by glycine (34), suggesting that glycine-sensitive receptors evolved independently in C. elegans and chordates. It seems reasonable to anticipate that neurotransmitter signaling complexity should increase in proportion to the requirements imposed by the number of different classes of neurons in nervous systems of either extant species or their ancestors. The fact that the number of transmitter-gated channel genes is fewer in the genome of Ciona than it is in the genome of Drosophila possibly reflects, for example, the fact that ascidians have a simpler nervous system than that of Drosophila, in which the order of 100 classes of neurons occur in the visual system alone (43). On the other hand, the same reasoning cannot hold for C. elegans, in which, like Ciona, the number of neurons and possibly also neuron classes is also small (67), while their transmitter-gated channel genes are considerably divergent (Table 1).

The predictable aspect of our findings, the conservation of transmitter-gated channels between ascidian and vertebrate genomes, is seen not only in the overall homology of genes but also in the coding sequences for amino acid residues that are critical to biophysical function. For example, it is well known that the Ca2+ permeability and high Mg2+ sensitivity of NMDA receptors are essential for neuronal plasticity and thus required for such functions as learning and neural development (13). The NMDA receptor has a heterooligomeric structure comprising two types of subunits (NR1 and NR2). Asparagine (N) in the M2 domain, the pore-forming region, is the critical site for the Ca2+ permeability and Mg2+ sensitivity of the NMDA receptor. The predicted amino acid sequence of both ascidian NR-like genes has asparagine in the critical site of the M2 domain, just like all known vertebrate NMDA receptor subunits. The ascidian channel protein of the NR1/NR2 complex is therefore most likely to be both Mg2+ sensitive and Ca2+ permeable, just like the corresponding vertebrate channels. In Drosophila and C. elegans, one of the NR subunits has a nonasparagine residue in this critical region. The composition of ascidian putative NMDA channels is not yet known, and expression studies on Ciona NMDA receptors will be required to demonstrate that these too are heteromeric and thus to examine whether the high Ca2+ permeability and Mg2+ sensitivity of an NR1/NR2 channel complex could be a chordate-specific design.

Ciona Lacks Genes for Fine-Tuning Subunits and for the Fast Conduction of Excitation

Our survey indicates that, with the exception of BK β-, Kv β-, and Cav β-, γ-, and α2/δ-subunits, most auxiliary subunits for voltage-gated channels are absent from the Ciona genome. Auxiliary subunits play critical roles in the fine-tuning of membrane excitability in recent vertebrates (27), and we therefore assume that such functions are lacking in Ciona. KCNQE subunits, which are critical subunits for modulating slowly activating K+ channels in cardiac cells and neurons, are also absent in Ciona. Ciona moreover lacks the gene for Nav β-subunits, which induce clustering of Nav channels at the initial segment and node of Ranvier of myelinated neurons, consistent with the fact that myelination does not occur in the ascidian nervous system (30) and that there is no myelin-related gene in the Ciona genome (11). Thus, either these features were all lost in ascidians, or, much more likely, the molecular architecture for fast neuronal conduction must have arisen after the urochordate branch diverged from the other chordate line. In support of the latter interpretation, Ciona also lacks α1S-type (or Cav1.1) Ca2+ channel and skeletal muscle Nav channel genes, which are both required for fast twitch in vertebrate muscles. Because the Fugu genome seems to contain a set of genes for rapid electrical conduction, including genes of the skeletal muscle type Cav channel and Nav channel as well as the Nav β-subunit gene, we conclude that these genes all arose around the emergence of, and have been conserved in, vertebrates. These findings also suggest that the innovation of both Nav β-subunit and myelin-related genes may have arisen concurrently during the course of chordate evolution, in which these genes were required to establish vertebrates' locomotory systems. Examination of amphioxus, which can dart with extreme rapidity and may therefore also have fast conduction pathways, but which lacks myelin, may be illuminating.

Contrast in gene Diversities Between Neuronal Excitability Channels and Channels Related to Homeostatic Control in the Ciona Genome

In contrast to the channel genes that are related to neuronal excitability and synaptic transmission, channels related to homeostasis of electrolyte and fluid volume show a more remarkable pattern of diversity, one that also reveals branches unique to ascidians. The numbers of TRP-related channels, connexins, and amiloride-sensitive cation channels in the Ciona genome are comparable to those recorded in the human genome. Why then are homeostasis-directed channel genes so diverse in ascidians relative to neuronal channel genes?

The wide gene diversity of TRP-related channels, connexins, Cl channels, and amiloride-sensitive cation channels might reflect the evolutionary history of adaptations of urochordates to changes in their physical environment. Ciona generally inhabits the coastal substratum and thus is frequently exposed to changes in physical conditions such as salinity, osmotic pressure, and temperature. Such changes are more serious for the survival of sessile adults than they would be for the transient lifestyle of the free-swimming larvae. Thus it is possible that this variety of channels related to homeostasis may facilitate adaptation of Ciona adults to the intense pressure of environmental changes. This must await future studies on the detailed characterization of functional properties of the relevant channels and their expression patterns in adult tissues. In addition, this issue will also be assessed by future studies of homolog genes from other marine invertebrates, especially other urochordate organisms, including appendicularians and nonchordate deuterostomes such as sea urchin, as well as in amphioxus.

Sequential Acquisition of Ion Channel Genes During Chordate Evolution

Surveys of the Ciona genome reveal the distribution of ion channel genes in three groups, those unique to vertebrates, those unique to chordates, and those shared with protostome invertebrates. From these distributional profiles, predicted major events in the evolution of ion channels are summarized in Fig. 4. The first group, those genes unique to vertebrates, contains genes related to advanced epithelial ion transport and to pain sensation. We could not find any gene for amiloride-sensitive cation channels belonging to the ENaC clade, a gene within a branch of the CLC-K channel genes, or a CFTR-like gene, even though other members of the amiloride-sensitive cation channel gene family, the CLCN channel gene family and the family of ABC-transporter genes, do exist in the Ciona genome. These genes absent in the Ciona genome putatively encode channels that are involved in the robust transport of sodium and chloride ions and especially play critical roles in vertebrate renal, pulmonary, and alimentary function. Ascidians are ammonotelic animals that lack a kidney or other related excretory organ; excretion and ion transport depend instead on other organs such as the branchial sac and esophagus (16). A pulmonary alveolar epithelium is also absent in ascidians. On the other hand, a prototype gene putatively encoding an ATP-dependent Kir channel (Ci-Kir2/5), which is essential for renal function in mammals, is present in the Ciona genome but is found in neither the Drosophila nor C. elegans genomes (Table 1, Fig. 4). It is also noteworthy that the Fugu genome contains a CFTR-like gene but lacks both CLC-K and ENaC homologs, even though Fugu does have nephrons, whereas the kidney in freshwater Tilapia does express a CLC-K gene (44). These findings suggest that these channels involved in advanced ion transport, such as for renal function or respiration, were acquired or sometimes discarded during the process of chordate evolution, possibly associated with loss of a marine habit or adaptation to life on land.

Fig. 4.

Evolutionary events in ion channel diversification. Summary showing the major events in the appearance and disappearance of genes during the radiation from the presumed deuterostome/protostome ancestor, as predicted by profiles of ion channel genes from the genomes of the human, Drosophila, Chaenorhabditis elegans, Ciona, and Fugu.

The Ciona genome also seems to lack genes related to pain sensation. The neural mechanisms underlying vertebrate pain sensation depend on the functions of several specific classes of ion channels that are exclusively expressed in peripheral sensory neurons. These include capsaicin receptors, Nav1.8, Nav1.9, and ionotropic purinergic (P2X-type) receptors, counterpart genes for which are all absent from the Ciona genome. Searches of the Fugu genome show that this teleost contains genes putatively encoding both P2X and capsaicin receptors (data not shown). Ciona also lacks neurotrophins and their receptors, which are important for the modulation of pain sensation. On the other hand, the Ciona genome does have genes putatively encoding tachykinin receptors and ASIC-related cation channels, as well as a substance P-like gene, that also mediate the signaling of pain sensation in vertebrates (33). Without functional studies, it is still premature to conclude that ascidians lack pain-sensing pathways, but the evolutionary acquisition of pain-signaling pathways in the chordate lineage is likely to have depended not only on the acquisition of novel ion channels but also on the persistence of these ancient membrane or ligand proteins.

For genes of the second group, those unique to chordates (or deuterostomes), our survey also highlights the many ion channel genes that are not present in the Drosophila and C. elegans genomes but conserved between Ciona and vertebrate genomes (Fig. 4). These genes include those putatively encoding GIRK, 5HT3R, GlyR, BK channel β-subunit, Kv β-subunit, ASIC channel, CLCA channel, GluR δ-like channel, connexin, two-domain type Cav channel, and CatSper-like cation channel. Even though their status may require revision in the light of future genome sequencing projects from other species, especially from nonchordate deuterostomes, currently these genes can all be defined as those unique to chordates.

In the third group of genes, those shared with Drosophila and C. elegans, one of our surprising findings is that some ascidian channel genes belong to lineages that have been specific so far to protostome channels. The Nav-like channel (Ci-Nav2) gene is, for example, related to the Nav-like channels of insects and the squid Loligo (46, 70). The distribution of these genes does not involve the genes of any vertebrate genome. Ci-Nav2 and its counterparts in other invertebrates share similar properties: the diagnostic DEKA sequence, conferring the sodium selectivity filter, is represented by DEEA, and the fast inactivation ball region in the III-IV linker is not conserved (70). In another case, three ascidian putative GluR genes are located in the outgroup of the vertebrate GluR channel gene tree. C. elegans and Drosophila GluR genes have also several GluR channel-like genes in the outgroup for vertebrate GluR genes. These suggest that vertebrates inherited only restricted sets of prototype Nav and GluR genes. An additional case is provided by the TRPV subfamily, in which two Ciona genes (Ci-Osm9-related1, Ci-Osm9-related2) are rather more related to the C. elegans osm-9 gene (8) and its relatives of Drosophila and C. elegans than to vertebrate TRPV genes (Fig. 3D). In summary, Ciona exhibits a repertoire of both vertebrate and protostomatous channel genes. This ambivalence in their affinities raises the possibility that the vertebrate ancestor lost several ion channel genes originally shared with protostomes and intensively multiplied a restricted set of the inherited genes during the course of its evolution.

We have gained insights into the evolutionary history of ion channel genes in chordate animals through a systematic survey of ion channel genes from the Ciona genome. A more detailed analysis of the provenance and diversification of channel genes, along with modifications to these imposed by vertebrate physiological functions such as advanced epithelial transport, kidney function, and pain sensation, will require comparative genomic analysis of additional basal chordates. For the latter, current projects, especially on appendicularian (61), amphioxus (15), medaka (47), and zebrafish teleost genomes (The Wellcome Trust Sanger Institute Danio rerio Sequencing Project:, will all provide essential landmarks to follow the evolutionary path of chordate ion channels. Wide-range surveys of ion transporter genes not evaluated in the present study will also provide important anciliary information. With the detailed description of ascidian embryology and physiology, and recently developed molecular tools, information in the present study will be useful for future studies to address the question of how multiple channel molecules are integrated into animal functions and development.


This research was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan to Y. Okamura and Y. Mori (Grant-in-Aid for Creative Scientific Research; 2001–2006, no. 13GS0016), Y. Okada (no. 14207002), and N. Satoh (no. 12202001) and from the Natural Sciences and Engineering Research Council of Canada to I. A. Meinertzhagen (no. OPG0000065).


We thank Dr. A. Hazama for helpful comments on the gene diversity of aquaporins and Drs. T. Tomiki and H. Takahashi for discussions. We thank M. Takagi at the National Institutes for Physiological Sciences for technical support.

Present address of Y. Hara, T. Yoshida, and Y. Mori: Laboratory of Molecular Biology, Dept. of Synthetic Chemistry and Biological Chemistry, Graduate School of Engineering, Kyoto Univ., Kyoto, Japan.

Present address of H. Watari: Graduate Program in Neurobiology and Behavior, Univ. of Washington, Seattle, WA.

Present address of M. Nishida: Dept. of Pharmacology and Toxicology, Graduate School of Pharmaceutical Sciences, Kyushu Univ., Higashiku, Fukuoka, Japan.


  • 1 The Supplemental Material for this article (Supplemental Table S1 and Supplemental Figs. S1–S6) is available online at

  • Article published online before print. See web site for date of publication (

    Address for reprint requests and other correspondence: Y. Okamura, Okazaki Institute for Integrative Bioscience, National Institutes of Natural Sciences, Higashiyama 5-1, Myodaiji-cho, Okazaki, Aichi 444-8787, Japan (e-mail: yokamura{at}



  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 27.
  24. 28.
  25. 29.
  26. 30.
  27. 31.
  28. 32.
  29. 33.
  30. 34.
  31. 35.
  32. 36.
  33. 37.
  34. 38.
  35. 39.
  36. 40.
  37. 41.
  38. 42.
  39. 43.
  40. 44.
  41. 45.
  42. 46.
  43. 47.
  44. 48.
  45. 49.
  46. 50.
  47. 51.
  48. 52.
  49. 53.
  50. 54.
  51. 55.
  52. 56.
  53. 57.
  54. 58.
  55. 59.
  56. 60.
  57. 61.
  58. 62.
  59. 63.
  60. 64.
  61. 65.
  62. 66.
  63. 67.
  64. 68.
  65. 69.
  66. 70.
View Abstract