Ascidians, or sea squirts, are lower chordates, and share basic gene repertoires and many characteristics, both developmental and physiological, with vertebrates. Therefore, decoding cis-regulatory systems in ascidians will contribute toward elucidating the genetic regulatory systems underlying the developmental and physiological processes of vertebrates. cis-Regulatory DNAs can also be used for tissue-specific genetic manipulation, a powerful tool for studying ascidian development and physiology. Because the ascidian genome is compact compared with vertebrate genomes, both intergenic regions and introns are relatively small in ascidians. Short upstream intergenic regions contain a complete set of cis-regulatory elements for spatially regulated expression of a majority of ascidian genes. These features of the ascidian genome are a great advantage in identifying cis-regulatory sequences and in analyzing their functions. Function of cis-regulatory DNAs has been analyzed for a number of tissue-specific and developmentally regulated genes of ascidians by introducing promoter-reporter fusion constructs into ascidian embryos. The availability of the whole genome sequences of the two Ciona species, Ciona intestinalis and Ciona savignyi, facilitates comparative genomics approaches to identify cis-regulatory DNAs. Recent studies demonstrate that computational methods can help identify cis-regulatory elements in the ascidian genome. This review presents a comprehensive list of ascidian genes whose cis-regulatory regions have been subjected to functional analysis, and highlights the recent advances in bioinformatics and comparative genomics approaches to cis-regulatory systems in ascidians.
INTRODUCTION
cis-Regulatory DNA controls gene expression and contains binding sites for transcription factors (Davidson, 2001; Howard and Davidson, 2004). A large fraction of metazoan genomes are thought to encode cis-regulatory DNAs (Duret and Bucher, 1997; Hardison, 2000; Onyango et al., 2000; Harafuji et al., 2002; Cameron et al., 2004). Deciphering cis-regulatory DNAs is important because they encode where, when, and how much each protein will be produced. Unlike protein-coding DNA sequences, however, the prediction of cis-regulatory elements solely by computational methods has been difficult, especially in higher eukaryotes, because the transcription factor binding sites are generally short, show degenerate sequences, and are often widely dispersed in large intergenic and intronic DNAs. In addition, characterization of the cis-regulatory regions of even a single gene usually requires labor-intensive experiments. For these reasons, only a limited number of cis-regulatory DNAs have been elucidated in metazoans.
Recent advances in molecular developmental biology and genomics of the urochordate ascidian offer a unique opportunity to investigate cis-regulatory DNAs in a chordate genome. An ascidian fertilized egg develops within a day into a tadpole-like larva consisting of ≈2600 cells (Satoh and Jeffery, 1995; Satoh, 2003). The ascidian larva shares a basic body plan with vertebrates (Satoh and Jeffery, 1995; Corbo et al., 2001; Satoh, 2001, 2003; Satoh et al., 2003). The draft genome of the ascidian Ciona intestinalis was determined; its 153~159 Mbp genome is estimated to contain 15,852 protein coding genes; on average, there is one gene every 7.5 kb of euchromatic DNA (Dehal et al., 2002). Because of a well-established cell lineage (Conklin, 1905; Nishida, 1987) and transparency, spatial gene expression patterns can be visualized in detail during ascidian development; extensive information is available on the expression profiles of thousands of genes in Halocynthia roretzi (Kawashima et al., 2000, 2002; Makabe et al., 2001) and C. intestinalis (Satou et al., 2001a, 2002; Nishikata et al., 2001; Kusakabe et al., 2002; Fujiwara et al., 2002; Ogasawara et al., 2002). Transient transgenesis can be used for efficient expression of exogenous genes in ascidian embryos and larvae (Fig. 1) (Hikosaka et al., 1992; Corbo et al., 1997a; Zeller, 2004). Simple electroporation methods permit the simultaneous transformation of hundreds of synchronously developing C. intestinalis embryos (Corbo et al., 1997a). Relatively short 5′ flanking sequences are generally sufficient to recapitulate the endogenous gene expression patterns of most tissue-specific genes in ascidians. The availability of the whole genome sequences of two Ciona species, C. intestinalis and C. savignyi, allows the application of comparative genomics approaches to identify cis-regulatory DNAs (Satoh et al., 2003; Johnson et al., 2004).
Fig. 1.
Expression of GFP driven by various tissue-specific promoters in C. intestinalis embryos and larvae. Promoter-GFP fusion constructs were introduced into fertilized eggs by electroporation. GFP fluorescence was observed specifically in muscle cells of mid tailbud embryos (A) and tadpole larvae (B) developed from eggs electroporated with promoter-GFP fusion construct of a myosin essential light chain gene (Kusakabe et al., 2004). (C) Nervous system-specific expression of GFP under the control of the promoter of a β-tubulin gene (Kusakabe et al., 2004). (D) Visualization of neuron morphology using a neuron-type specific promoter (Yoshida et al., 2004). Cell bodies and axons of GABAergic neurons were visualized by GFP fluorescence in a larva developed from an egg electroporated with a promoter-GFP fusion construct of the vesicular GABA transporter gene.
In addition to its value in understanding gene regulation, cis-regulatory DNA provides a powerful tool for tissue-specific genetic manipulation in the study of ascidian development and physiology. For example, cis-regulatory DNAs have been used to express wild-type or mutant forms of proteins that regulate developmental or physiological processes (Corbo et al., 1998; Fujiwara et al., 1998; Ono et al., 1999; Takahashi et al., 1999a; Imai et al., 2000; Satoh et al., 2000; Mitani et al., 2001; Okagaki et al., 2001; Di Gregorio et al., 2002; Keys et al., 2002; Kawai et al., 2005). cis-Regulatory DNAs were also used to demonstrate trans-splicing in ascidian embryos (Vandenberghe et al., 2001). Neurons and neural circuits were visualized in living embryos by expressing fluorescent proteins under the control of neuron-specific promoters (Fig. 1D) (Okada et al., 2001, 2002; Yoshida et al., 2004).
In spite of their simple genomes, ascidians share basic gene repertoires and many characteristics, both developmental and physiological, with vertebrates (Dehal et al., 2002; Satoh, 2003; Satoh et al., 2003; Meinertzhagen et al., 2004; Campbell et al., 2004). Therefore, decoding cis-regulatory systems in ascidians will contribute toward elucidating the genetic regulatory systems underlying the developmental and physiological processes of vertebrates. As of December 2004, functional analyses of cis-regulatory regions have been published for as many as 50 genes in ascidians (Table 1). This review presents a comprehensive list of ascidian genes whose cis-regulatory regions have been subjected to functional analysis, and highlights the recent advances in bioinformatics and comparative genomics approaches to cis-regulatory systems in ascidians.
Table 1.
Ascidian genes whose cis-regulatory regions were experimentally analyzed
EXOGENOUS DNA TRANSFER INTO ASCIDIANS
Foreign gene transfer into ascidian embryos was first reported in 1992 (Hikosaka et al., 1992). Hikosaka and his colleagues microinjected a plasmid DNA containing the bacterial β-galactosidase gene (lacZ) connected with the chicken β-actin promoter and the Rous sarcoma virus enhancer into fertilized eggs of C. savignyi. The microinjection was performed into eggs with an intact chorion (vitelline coat). The lacZ expression was observed in various cell-types of tailbud embryos and larvae, irrespective of injection of linear or circular form of the plasmid. The same paper also reported muscle-specific reporter expression by the promoter of a muscle actin gene (HrMA4a) of H. roretzi (Kusakabe et al., 1992). In this experiment, the bacterial gene encoding chloramphenicol acetyltransferase (CAT) was used as the reporter and the promoter-reporter fusion plasmid was microinjected into H. roretzi fertilized eggs without chorions. In subsequent studies, microinjection into H. roretzi eggs has been usually performed through the chemically softened chorion (Hikosaka et al., 1994; Kusakabe et al., 1995).
Microinjection into eggs with intact chorions has been used for transgenesis of two widely used Ciona species, C. intestinalis (Kusakabe et al., 1996) and C. savignyi (Hikosaka et al., 1992, 1993; Deschet et al., 2003), and two molgulid ascidians, Molgula oculata and Molgula occulta (Kusakabe et al., 1996). However, microinjection into intact C. intestinalis eggs has been generally regarded as difficult (Zeller, 2004), and microinjection is usually performed into dechorionated eggs in this species (e.g. Sasakura et al., 2003a). Microinjection either before or after insemination can produce transgene expression in Ciona embryos. Micro-injection of DNA constructs into a particular blastomere has been achieved with dechorionated embryos of C. savignyi (Hikosaka et al., 1993) and with H. roretzi embryos with chemically softened chorions (Okada et al., 2002; Katsuyama et al., 2002).
Either circular or linearized forms of plasmid DNA constructs can be injected into ascidian eggs to obtain exogenous gene expression. In C. savignyi, Hikosaka et al. (1992) reported that both the efficiency and spatial patterns of lacZ expression were comparable between larvae developed from eggs injected with circular and linearized DNAs. In experiments using the Halocynthia tyrosinase gene promoter, however, when circular plasmids were injected, it was necessary to increase the quantity of injected DNA from a concentration of 5.0 ng/μl, which was sufficient for linearized plasmids, to 50 ng/μl to obtain reporter lacZ expression (Toyoda et al., 2004). Furthermore, a lacZ-fusion construct whose expression was not detected when it had been injected in a linearized form was capable of driving gene expression when it had been injected in a circular form (Toyoda et al., 2000).
Electroporation has become the most commonly used method to introduce exogenous DNAs into Ciona embryos (Corbo et al., 2001, Zeller, 2004). In 1997, Corbo and his colleagues first reported the electroporation of exogenous DNA into C. intestinalis embryos (Corbo et al., 1997a). The same method has also been applied to transgenesis of C. savignyi embryos (Nakatani et al., 1999). Electroporation is performed with dechorionated fertilized eggs. This method is faster, simpler, and more efficient than microinjection and permits the simultaneous transformation of hundreds of synchronously developing embryos. Like microinjection, electroporation has been used both to analyze cis-regulatory DNAs and to assess gene function (Corbo et al., 2001, Di Gregorio and Levine, 2002; Zeller, 2004). However, there are some limitations on application of electroporation: first, introduction of DNA into particular blastomeres is difficult; second, because dechorionated embryos do not develop normal tunics and become sticky, larvae that develop from electroporated eggs do not swim well. This problem can be a disadvantage for studies of neural function and swimming behavior.
Most transgenic studies in ascidians conducted so far are “transient” transgenesis. This is mainly due to difficulties in long term culturing and also due to the low frequency of germline transmission of transgene constructs. However, some laboratories have overcome these difficulties, and stable transgenic lines of ascidians have been established (Deschet et al., 2003; Sasakura et al., 2003a). Deschet and her colleagues generated stable transgenic lines of C. savignyi using a technique in which the endonulease I-SceI was coinjected into fertilized eggs with a transgene construct (Deschet et al., 2003). Sasakura and his colleagues used a Tc1/mariner superfamily transposon, Minos, to achieve germ-line transgenesis of both C. intestinalis (Sasakura et al., 2003a) and C. savignyi (Matsuoka et al., 2004). Minos has high activity in Ciona (Sasakura et al., 2003b); it was efficiently integrated into the genome of germ cells and stably transmitted to subsequent generations (Sasakura et al., 2003a). Enhancer trapping in Ciona by Minos was demonstrated and led to the identification of Musashi orthologue enhancers (Awazu et al., 2004).
CELL/TISSUE-SPECIFIC CIS-REGULATORY REGIONS
Larval Muscle
Ascidian larvae develop unicellular, striated muscle cells on each side of the tail. The number of muscle cells differs among species: 36 in C. intestinalis and 42 in H. roretzi (Nishida, 1987). Twenty-eight of the 42 muscle cells of the H. roretzi tadpole are derived from the B4.1 blastomeres (B-line), four from the A4.1 blastomeres (A-line), and 10 from the b4.2 blastomeres (b-line) of the bilaterally symmetrical 8-cell embryo (Nishida, 1987). The B-line muscle cells differentiate autonomously, whereas the A- and b-line muscle cells are specified conditionally (Nishida, 1992, 2002). Developmental mechanisms of the larval muscle have been a central subject in embryology of ascidians since E. G. Conklin proposed a maternal cytoplasmic determinant for muscle differentiation (Conklin, 1905; Nishida, 1992, 2002; Nishida and Sawada, 2001).
Muscle actin gene promoters are the first ascidian cis-regulatory DNAs whose function was analyzed (Hikosaka et al., 1992, 1993, 1994; Kusakabe et al., 1995; Satou and Satoh, 1996; Kusakabe, 1997). In H. roretzi, at least seven muscle actin genes are expressed in larval muscle (Kusakabe et al., 1995). Five of them (HrMA2, HrMA4a, HrMA4b, HrMA5, and HrMA6) form a tandem cluster in the genome (Kusakabe et al., 1992). The other two genes (HrMA1a and HrMA1b) form another cluster in which they are closely linked in a head to head arrangement and share a 340-bp bi-directional promoter (Kusakabe et al., 1995). Microinjection of lacZ fusion constructs of each of the seven HrMA genes demonstrated that short upstream regions (<200 bp) are sufficient for the muscle-specific expression of each gene (Hikosaka et al., 1994; Kusakabe et al., 1995; Satou and Satoh, 1996; Satoh et al., 1996). Another muscle-specific promoter identified and analyzed in Halocynthia is that of a myosin heavy chain gene, HrMHC1 (Araki and Satoh, 1996). The 242-bp upstream region (from −132 to +110) of HrMHC1 is sufficient to drive muscle-specific reporter gene expression. In C. intestinalis, promoter function has been analyzed for a number of muscle-specific genes, including genes encoding troponin I (Ci-TnI), troponin T (Ci-TnT), calcium-ATPase (Ci-SercaA), myosin essential light chain (Ci-MLC2), and myosin regulatory light chain (Ci-MRLC2) (Vandenberghe et al., 2001; Cleto et al., 2003; Davidson and Levine, 2003; Kusakabe et al., 2004). A muscle-specific enhancer was identified and characterized in the upstream region of the snail gene of C. intestinalis (Erives et al., 1998; Erives and Levine, 2000). Muscle-specific promoters were also identified in Molgula (Kusakabe et al., 1996).
E-box (CANNTG) and CArG box (CC(A/T)6GG) sequences are conserved in the HrMA promoters (Kusakabe et al., 1992; Kusakabe et al., 1995). An E-box is also present in the minimal promoter of HrMHC1 (Araki and Satoh, 1996). Deletion or mutation of these motifs, however, did not severely impair muscle-specific activity of the promoters (Hikosaka et al., 1994; Satou and Satoh, 1996; Satoh et al., 1996; Araki and Satoh, 1996). Instead, two cis-regulatory elements, B- and M-regions, were shown to be necessary and sufficient for muscle-specific transcription from the HrMA4a promoter (Satou and Satoh, 1996). The sequences of the B- and M-regions are 5′-TCGCACTTC-3′ and 5′-GTGATAACAACTG-3′, respectively. Both motifs were present in the promoter regions of three H. roretzi troponin I genes (Yuasa et al., 2002). A 12-bp sequence (5′-TTTT-TTCTTTCA-3′), apparently distinct from the B- and M-regions of HrMA4a, is critical for the promoter activity of HrMHC1 (Araki and Satoh, 1996).
Analysis of a C. intestinalis snail (Ci-Sna) gene enhancer revealed that AC-core E-box sequences (CAA-CTG) and motifs recognized by T-box transcription factors (T-binding motif; GTGNNA) are important for gene expression in muscle lineage cells (Erives et al., 1998; Erives and Levine, 2000). Interestingly, the M-region of HrMA4a consists of an AC-core E-box and a T-binding motif, and the B-region also contains a T-binding motif (Erives and Levine, 2000). T-binding motifs are also found in the promoters of HrMA1a, HrMA1b, HrMHC1, and As-T2, the Halocynthia Tbx6 gene (Erives and Levine, 2000; Mitani et al., 2001). Mutation or deletion of T-binding motifs reduced activity of these promoters. Candidate transcription factors that recognize T-box motifs in these muscle-specific regulatory regions are CiVegTR and Tbx6 (As-T2) (Mitani et al., 1999; Erives and Levine, 2000; Mitani et al., 2001). Both the Halocynthia and Ciona Tbx6 genes are expressed in differentiating muscle cells (Yasuo et al., 1996; Mitani et al., 1999; Takatori et al., 2004). Ectopic and/or over expression of Halocynthia Tbx6 promoted the ectopic expression of HrMA4 and HrMHC1 in epidermal cells, and its inhibition of the Tbx6 function resulted in specific downregulation of HrMA4 and HrMHC1 transcription (Mitani et al., 1999). Thus, Tbx6 seems to be a major transcriptional activator in embryonic muscle cells.
The myogenic bHLH transcription factors, such as MyoD, bind to E-box sequences and are essential for the expression of many muscle-specific genes in vertebrates (Olson, 1990; Weintraub et al., 1991). The ascidian genome contains only one myogenic bHLH family gene: AMD1 in Halocynthia (Araki et al., 1994) and CiMDF in Ciona (Meedel et al., 1997). The myogenic bHLH gene is expressed in developing muscle cells of ascidian embryos. Sequence comparisons between C. savignyi and C. intestinalis depicted highly conserved noncoding sequences in the 5′ flanking region of the troponin I (TnI) gene (Johnson et al., 2004). Functional analyses of the C. savignyi TnI enhancer demonstrated that the conserved sequences contain cis-regulatory elements important for muscle-specific transcription (Johnson et al., 2004). One such conserved region contains a potential binding site for a myogenic bHLH factor; this sequence (5′-TGCAGCTG-3′) perfectly matches a muscle-specific motif, M9, found by a computational analysis of muscle-specific promoters of C. intestinalis (Fig. 2) (Kusakabe et al., 2004). Therefore, CiMDF may directly activate transcription of TnI. A number of promoters that lack E-boxes, however, can drive transcription in muscle cells of ascidian embryos (Hikosaka et al., 1994; Satou and Satoh, 1996; Satoh et al., 1996; Araki and Satoh, 1996; Kusakabe et al., 2004). Hence, the role of the myogenic bHLH factor in ascidian embryonic muscle cells remains ambiguous.
Fig. 2.
Computational discovery of DNA motifs associated with muscle-specific gene expression in C. intestinalis larvae (Kusakabe et al., 2004). (A) Motif logos of the six over-represented motifs (M1, M2, M3, M4, M5, and M9). The height of the stack of letters is proportional to the information content, and the relative frequency of each base is given by its relative height. A perfect match would have a score of two bits. (B) Positional distributions of the occurrence of six over-represented motifs found in upstream regions of the muscle-specific gene set. The positions of base 1 of the motifs were binned in 100-bp intervals. ‘Frequency’ represents the proportion of the number of motifs found in a 100-bp interval to the total number of motifs found in the 1000-bp upstream region. (C) Function of M2 sites in muscle-specific expression of a myosin essential light chain gene (Ci-MLC2). Schematic diagram showing the structure of wild type and mutant GFP constructs. A 643-bp genomic DNA, containing a 5′ flanking sequence, the first exon, the first intron, and a 5′ portion of the second exon of Ci-MLC2, is connected with an enhanced GFP coding sequence (EGFP). The exons are indicated by an open box (noncoding region) and solid boxes (coding region). Two M2 sites and the putative TATA-box are indicated by yellow ellipses and an orange box, respectively. Numbers at the 5′ end of the wild-type construct indicate nucleotide positions relative to the translational start site (+1). The percentages of positive embryos were indicated at the right side of the drawing representing each construct. Numbers in the parentheses indicate the number of positive and total embryos scored for each constructs.
Computational analyses of 5′ flanking regions of 22 muscle-specific genes of C. intestinalis identified several DNA motifs associated with muscle-specific gene expression (Fig. 2) (Kusakabe et al., 2004). One of these motifs, M2, was distributed preferentially in regions from −200 to −100 bp relative to the translational start sites (Fig. 2B). Mutations of M2 sites in the Ci-MLC2 and Ci-MRLC2 promoters greatly reduced gene expression in muscle cells (Fig. 2C). When M2 sites were located upstream of a basal promoter, the reporter GFP was specifically expressed in muscle cells (Fig. 2C). Motif M2 contains a consensus binding sequence for a CREB transcription factor. CREB and ATF proteins together constitute a subfamily of the basic leucine zipper (bZip) protein family. A recent study demonstrated that mammalian CREB binds directly to the retinoblastoma protein gene (RB) promoter and activates RB transcription in differentiating myoblasts by recruiting MyoD and the p300 coactivator (Magenta et al., 2003). This MyoD-mediated transactivation is independent of the E-boxes and of the direct binding of MyoD to the promoter sequences. Interestingly, there are no E-box sequences in the Ci-MLC2 and Ci-MRLC2 promoters. In the human RB promoter, the CREB binding site is located at positions from −192 to −187 relative to the transcription start site (Magenta et al., 2003). This location is similar to regions where M2 sites are frequently found in the Ciona muscle-specific promoters. These observations suggest that a Ciona CREB or CREB-related protein regulates muscle-specific transcription by a mechanism similar to that found in RB activation in mammalian skeletal muscle (Kusakabe et al., 2004).
Differentiation of primary muscle cells in ascidian embryos is triggered by maternally provided mRNA encoding a Zic-like zinc-finger protein, macho-1 (Nishida and Sawada, 2001). Ciona macho-1 recognizes a nucleotide sequence, 5′-GCCCCCCGCTG-3′, that resembles the mammalian Zic binding sites (Yagi et al., 2004a). Potential Zic binding sequences were found in 5′ flanking regions of transcription factor genes, Tbx6b and snail, both of which are specifically expressed in muscle lineage cells. So far, however, macho-1-binding sequences have not been reported in cis-regulatory regions of muscle-specific differentiation genes.
Notochord
The ascidian larva has a notochord comprising 40 cells, whose lineage has been described completely (Conklin, 1905; Nishida, 1987). The 32 anterior notochord cells are derived from the A4.1 blastomeres (A-line) of the bilaterally symmetrical 8-cell embryo and the eight posterior cells are derived from the B4.1 blastomeres (B-line). As the notochord is a key feature of chordates, the elucidation of the genetic circuitry underlying its formation in ascidian embryos is central to understanding molecular developmental mechanisms underlying chordate evolution (Satoh and Jeffery, 1995; Di Gregorio and Levine, 1998; Satou and Satoh, 1999).
Notochord lineage-specific promoters and enhancers have been identified and characterized for both developmental regulatory genes (Corbo et al., 1997a; Takahashi et al., 1999b; Di Gregorio et al., 2001) and a downstream differentiation gene (Di Gregorio and Levine, 1999). The first notochord-specific cis-regulatory DNA identified in chordates is an enhancer of the C. intestinalis Brachyury gene (Ci-Bra) (Corbo et al., 1997a). Brachyury encodes a sequence-specific activator that contains a T-box DNA binding domain and is critical for notochord differentiation in vertebrate embryos (Hermann et al., 1990; Smith, 1999). Brachyury is expressed exclusively in the H. roretzi (Yasuo and Satoh, 1993) and C. intestinalis (Corbo et al., 1997a) notochord precursor cells. The spatial and temporal patterns of Brachyury expression coincide with the developmental fate restriction of the notochord lineages. Regulatory mechanisms of notochord-specific activation of Ci-Bra have been extensively studied (Corbo et al., 1997a; Fujiwara et al., 1998; Corbo et al., 1998; Takahashi et al., 1999b; Yagi et al., 2004b). The minimal Ci-Bra enhancer for notochord specific-expression is 434 bp in length (Corbo et al., 1997a). The Ci-Bra enhancer contains recognition sequences of Suppressor of Hairless [Su(H)] (Corbo et al., 1998) and Snail (Ci-Sna) (Fujiwara et al., 1998). Su(H) activates the Ci-Bra enhancer in all the mesodermal lineages, including the notochord, muscle, and trunk mesenchyme (Corbo et al., 1998). The Ci-Sna gene is expressed in the tail muscle cells and trunk mesenchyme, and Ci-Sna directly represses Ci-Bra expression (Corbo et al., 1997b; Fujiwara et al., 1998). A forkhead transcription factor gene, FoxD, and a zinc finger transcription factor gene, ZicL, are also involved in the process of notochord specification in Ciona (Imai et al., 2002a, 2002b; Yagi et al., 2004b). FoxD activates the expression of ZicL in the A-line notochord cells (Imai et al., 2002a). ZicL expression is essential for the differentiation of A-line notochord cells but not of B-line notochord cells (Imai et al., 2002b). ZicL directly binds to the proximal upstream region of Ci-Bra that is required for the basal activation of the gene (Yagi et al., 2004b). Several experiments consistently suggested that ZicL is a direct activator of Ci-Bra (Yagi et al., 2004b). In B-line notochord cells, the Notch signaling pathway, which is downstream of FoxD, is likely to be involved in the activation of Brachyury (Imai et al., 2002a).
The function of cis-regulatory regions of the Brachyury gene has also been analyzed in H. roretzi (Takahashi et al., 1999b). Deletion analysis of the 5′ flanking region of the H. roretzi Brachyury gene (As-T) demonstrated that a module between −289 and −250 bp is responsible for notochord-specific expression of a lacZ reporter. The 5′ flanking region of As-T contains a potential T-binding motif (ACCTAGGT) around −160 bp. Takahashi et al. (1999b) suggest that the T-binding motif is responsible for autoactivation of the gene.
Brachyury is a key regulator of notochord formation in ascidian embryos (Yasuo and Satoh, 1998; Takahashi et al., 1999a; Satou et al., 2001b). Ci-Bra triggers the transcription of various downstream genes in notochord cells (Takahashi et al., 1999a; Hotta et al., 2000). The upstream regulatory region of a Ci-Bra target gene, Citropomyosin-like (Ci-trop), has been characterized (Di Gregorio and Levine, 1999). A minimal, 114 bp enhancer is sufficient to direct the expression of the heterologous promoter in the notochord. This enhancer contains Ci-Bra binding sites and deletion of the sites inactivates notochord-specific activity of a Ci-trop/lacZ transgene. Thus, Brachyury seems to directly activate notochord-specific downstream genes in ascidian embryos.
Another cis-regulatory DNA that might be involved in notochord development is an element in the upstream region of Ci-IκB, a gene encoding IκB homologue of C. intestinalis (Kawai et al., 2005). When the GFP fusion protein of Ci-rel1, a NF-κB/Rel homologue of C. intestinalis, was overexpressed in the notochord, the expression of Ci-IκB was remarkably enhanced in notochord cells at the beginning of the tailbud stage. The 1.0-kb upstream region specifically drives lacZ reporter expression in the notochord, while mutation or deletion of a κB consensus sequence in this upstream region results in decreased lacZ expression in the notochord and ectopic lacZ expression in mesenchyme and epidermis. Ci-rel1 can directly bind to this κB consensus sequence in vitro. Thus, transcriptional regulation mediated by the NF-κB/Rel signaling pathway may be involved in notochord formation in ascidian embryos.
Heart
Trunk ventral cells (TVCs) of the ascidian larva are precursors for the adult heart and derived from the B7.5 blastomere pairs of the 110-cell stage embryo (Hirano and Nishida, 1997; Satou et al., 2004). The ascidian heart first appears after metamorphosis as a tube with a single layered myoepithelium that is continuous to a single layered pericardial wall (Ichikawa and Hoshino, 1967; Davidson and Levine, 2003). Conserved migration patterns and expression of Nkx indicate that early heart specification is conserved among chordates (Davidson and Levin, 2003). Several genes are expressed in both the larval muscle and TVCs (Satou et al., 2001a; Kusakabe et al., 2002; Davidson and Levine, 2003). The ascidian Mesp gene is specifically and transiently expressed in B7.5 cells and is essential for the specification of heart precursor cells, which express Nkx, HAND, and HAND-like (NoTrlc) genes, suggesting that a mechanism for heart specification beginning with Mesp through Nkx and HAND is conserved among chordates (Satou et al., 2004).
In Halocynthia, the promoter of muscle actin genes can drive lacZ expression in the TVCs (Kusakabe et al., 1995). Davidson and Levine (2003) identified tissue-specific enhancers for five genes [Ci-TnI, Ci-TnT, Ci-NPP, 29h10, and Ci-Hndx (Ci-NoTrlc)] expressed in the C. intestinalis TVCs. Bioinformatics analysis identified shared sequence motifs within the 5′ flanking regions of Ci-NPP and Ci-Hndx, which show overlapping expression in the TVCs and endoderm (Davidson and Levine, 2003). The shared motifs include putative binding sites for Nkx, Tinman, and GATA factors. A genomic DNA fragment containing the putative Ci-Hndx enhancer predicted by the shared motif distribution directed an authentic pattern of lacZ expression in the TVCs, trunk lateral cells, which are mesenchymal cells derived from A7.6 blastomeres, and endoderm.
Endoderm
As in the case of vertebrate embryos, the endoderm of ascidian embryos is specified autonomously, and specified endoderm then induces notochord specification (Nishida, 1993, 2002; Satoh, 2001). The autonomy is dependent on maternal determinants that are prelocalized in the cytoplasm of eggs and early embryos (Nishida, 1993). In Ciona embryos, the nuclear accumulation of β-catenin is most likely the first step of endoderm specification (Imai et al., 2000). β-catenin is a downstream component of the Wnt signaling pathway and, to exert its function, enters the nucleus and activates downstream genes through TCF/LEF1, which is a HMG box transcription factor (Willert and Nusse, 1998; Sharpe et al., 2001). The nuclear accumulation of β-catenin in Ciona presumptive endodermal cells regulates many downstream genes (Imai et al., 2000; Imai, 2003).
Cititf1 is a potential β-catenin target and an early and specific marker of endoderm development in C. intestinalis (Ristoratore et al., 1999; Satou et al., 2001c). cis-Regulatory regions responsible for the endoderm specific expression of Cititf1 have been isolated and characterized (Ristoratore et al., 1999; Fanelli et al., 2003). The 315-bp genomic fragment spanning from −355 to −41 relative to the transcription start site is sufficient to activate a heterologous promoter in a pattern essentially identical to that of the endogenous Cititf1 gene (Fanelli et al., 2003). This enhancer contains at least three distinct regulatory regions; two of which are responsible for activation of transcription in the endoderm and in the mesenchyme, respectively, while the third is a negative control region that represses transcription in mesenchyme.
The Ciona forkhead/HNF-3β gene [Ci-fkh; also referred to as Ci-FoxA-a (Yagi et al., 2003)] is expressed in the notochord, endoderm, and rudimentary floor plate of the CNS during embryonic development (Corbo et al., 1997b; Di Gregorio et al., 2001). A regulatory sequence (AS) located ~1.7 kb upstream of the transcribed region is shown to be essential for expression in these tissues (Di Gregorio et al., 2001). The analysis of various truncated and deleted Ci-fkh promoter sequences showed that transcriptional repression is an essential component of Ci-fkh regulation. Removal of repressor sites from the AS resulted in the persistence of the early Ci-fkh expression pattern in the lateral ependymal cells of the spinal cord. A series of internal deletions allowed the uncoupling of a region responsible for notochord/endoderm expression from the cis-regulatory elements controlling expression in the CNS. Partial uncoupling of the notochord and endoderm expression patterns was obtained by mutating a T-box recognition sequence in the distal enhancer. This mutation specifically disrupted expression in the endoderm, whereas staining in the notochord was essentially unaffected. DNA binding assays showed that a GST fusion protein containing the T-domain of the Ciona Brachyury protein (Ci-Bra) binds this T-box element, although a different member of the T-box family is probably responsible for regulating Ci-fkh expression in the endoderm. DNA binding assays revealed a number of high affinity Ci-Fkh recognition sequences in the 5′ regulatory region. Mutations in one of these sites, adjacent to the AS, diminish the expression mediated by an otherwise normal Ci-fkh/lacZ transgene, thereby providing evidence for autoregulation (Di Gregorio et al., 2001).
Epidermis
Restriction to epidermal cell fate occurs as early as at the 16-cell stage in ascidian embryogenesis (Conklin, 1905; Nishida, 1987). Differentiation of epidermis is autonomous and controlled by maternal determinants (Nishida, 1994, 2002). HrEpiA, HrEpiB, HrEpiC, and HrEpiD are epidermis-specific genes of the H. roretzi embryo, which show different temporal expression patterns during embryogenesis. Putative cis-regulatory elements shared by HrEpiB and HrEpiD were investigated by microinjecting lacZ-fusion constructs into H. roretzi fertilized eggs (Ueki and Satoh, 1995). The 5′ upstream regions from −345 to +200 of HrEpiB and −166 to +108 of HrEpiD are sufficient for epidermis-specific expression of the reporter gene. The 5′ flanking regions of HrEpiB and HrEpiD share several sequence motifs, which might be responsible for regulation of gene expression during epidermal cell differentiation. The 5′ upstream sequences of HrEpiC that are associated with specific expression patterns of this gene were also analyzed (Ishida and Satoh, 1999). Restriction site mapping and sequencing of genomic clones showed that the H. roretzi genome contains two copies of HrEpiC genes, HrEpiC1 and HrEpiC2, aligned tandemly in about 8 kb of the genome. A 103-bp 5′ flanking region was sufficient for the minimal epidermis-specific expression of HrEpiC1. The region between −281 bp and −198 bp of the 5′ flanking region was associated with the amplification of the minimal expression of the reporter gene in the epidermis and also with activating HrEpiC1 on schedule at the 64-cell stage.
Nervous system
The ascidian larva has a central nervous system (CNS) derived from the dorsal neural tube and consisting of about 330 cells, of which about 100 cells are neurons (Meinertzhagen and Okamura, 2001). The expression patterns of developmental regulatory genes along the anteroposterior axis are conserved in the developing CNS between ascidians and vertebrates, suggesting that developmental mechanisms of the CNS are conserved among chordates (Wada et al., 1998; Hudson and Lemaire, 2001; Wada and Satoh, 2001; Imai et al., 2002c). The anterior brain vesicle (sensory vesicle) of the CNS contains two sensory organs, an eyespot (ocellus) and a gravity sense organ (otolith), which are responsible for the swimming behavior of the larva (Tsuda et al., 2003). The ascidian larva also has a peripheral nervous system consisting of the adhesive organ and epidermal neurons in the trunk and tail (Takamura, 1998).
A promoter of a synaptotagmin gene (syt) was identified and characterized in H. roretzi (Katsuyama et al., 2002), and it has been used to label motor neurons and to express wild-type and mutant forms of neuronal proteins (Ono et al., 1999; Okada et al., 2001, 2002). The 3.4-kb upstream region of the H. roretzi syt gene can drive reporter gene expression in neuronal and epidermal cells, but not elsewhere. Deletion analysis of the syt promoter suggested that syt expression in neurons and in the embryonic epidermis depends on distinct cis-regulatory regions. In C. intestinalis, cis-regulatory regions of four genes, Ci-Gαi1, Ci-arr, Ci-vAChTP, and Ci-vGAT, each of which is expressed in distinct sets of neurons in the central nervous system, were isolated and characterized (Yoshida et al., 2004). Ci-Gai1 is widely expressed in various types of neurons (Yoshida et al., 2002), and Ci-arr, Ci-vAChTP, and Ci-vGAT are expressed in ocellus photoreceptor cells, cholinergic neurons, and GABAergic neurons, respectively (Nakagawa et al., 2002; Takamura et al., 2002; Yoshida et al., 2004). The function of an upstream regulatory region of a pan-neuronal gene, Ci-tubulin-β2, was also examined in C. intestinalis (Kusakabe et al., 2004). The reporter gene driven by the 5′ flanking region of Ci-tubulin-β2, Ci-Gαi1, Ci-arr, and Ci-vAChTP recapitulated the endogenous gene expression patterns, while the Ci-vGAT promoter can drive GFP expression in particular subsets of neurons expressing the endogenous gene. Deletion analysis revealed that the Ci-Gαi1 promoter consists of multiple regulatory modules controlling expression in different types of cells.
The ocellus and otolith in the brain vesicle each contain a single pigment cell, which is important for sensory functions (Sakurai et al., 2004). The promoters of two pigment-cell specific genes, HrTyr (tyrosinase) and HrTyrp (tyrosinase-related protein), have been identified and characterized in H. roretzi (Toyoda et al., 2000, 2004; Wada et al., 2002). A 1.8-kb 5′ flanking region of HrTyr is sufficient for specific reporter gene expression in pigment cell precursors (Toyoda et al., 2000). Deletion analyses of the 5′ region revealed that the upstream region from −152 to the translation start site can drive gene expression in the pigment cells. An essential positive cis-regulatory element lies close to position −152, positive elements are also located in the region between −1.8 and −1.69 kb, and an additional regulatory sequence, which may act to restrict expression to the pigment cell lineage, is present between positions −250 and −170 (Toyoda et al., 2000). The 73-bp 5′ flanking region from −73 to −1 of HrTyrp is sufficient for gene expression in the pigment cells, although regions more upstream seem to be required for stronger expression (Toyoda et al., 2004). The H. roretzi Otx gene Hroth activates transcription of HrTyrp by binding either or both of sequences at −113/–108 and −90/–73 (Wada et al., 2002). The human Tyrp2 gene also has an Otx binding site and is activated by OTX2 (Takeda et al., 2003). Therefore, Otx may regulate expression of Tyrp genes in both vertebrates and ascidians. Because putative binding sites for Pax-3 are scattered in the upstream regions of HrTyrp and HrTyr (Toyoda et al., 2000, 2004) and overexpression of HrPax3/7 induces ectopic tyro-sinase enzyme activity (Wada et al., 1997), HrPax3/7 might activate transcription of HrTyrp and HrTyr. In vertebrates, Pax-3 may affect expression of tyrosinase family gene via the microphtalmia protein (Mitf), a basic helix-loop-helix protein transcription factor that is essential for melanocyte development (Goding, 2000). Interestingly, however, the minimal promoters of HrTyr and HrTyrp do not contain an M-box sequence, which is a potential binding site for Mitf and is conserved in the vertebrate tyrosinase family genes.
The function of cis-regulatory regions has also been analyzed for several transcription factor genes that are expressed in the central nervous system, including ascidian homologs of snail (Corbo et al., 1997b; Erives et al., 1998; Boffelli et al., 2004), forkhead/HNF-3β (Di Gregorio et al., 2001; Boffelli et al., 2004), Msx (Russo et al., 2004), Otx (Oda-Ishii and Saiga, 2003; Bertrand et al., 2003), Hox3 (Locascio et al., 1999), Pitx (Christiaen et al., 2005), and Distalless (Harafuji et al., 2002). Most of these transcription factor genes show complex expression patterns in multiple germ layers.
The Ciona snail gene Ci-Sna is expressed in the tail muscle, trunk mesenchyme, and CNS, including the brain vesicle and lateral ependymal cells of the nerve cord (Corbo et al., 1997b). A 3.0-kb 5′ flanking region of Ci-Sna is sufficient to direct the expression of a heterologous promoter in the CNS, muscle, and mesenchyme (Erives et al., 1998). Expression in the CNS persisted when the B4.1 enhancer, which is necessary for expression in the primary muscle and mesenchyme, had been removed from the 3.0-kb 5′ flanking sequence. Similarly, mutations of AC-core E-boxes in the B4.1 enhancer resulted in the loss of reporter expression in the primary muscle cells while still activating the gene expression in the brain vesicle and nerve cord (Erives et al., 1998). Recently, a CNS-specific enhancer of Ci-Sna was identified by an intraspecies sequence comparison and succeeding experimental analysis (Boffelli et al., 2004). Thus, different cis-regulatory elements control the Ci-Sna expression in the CNS and mesoderm.
The Ciona forkhead/HNF-3β gene Ci-fkh (Ci-FoxA-a) is expressed in the posteriormost region of the brain vesicle and in the ventral ependymal cells of the spinal cord (Di Gregorio et al., 2001). The ventral ependymal cells have been proposed to represent a rudimentary floor plate (Corbo et al., 1997b; Olsen and Jeffery, 1997; Shimauchi et al., 1997). Expression of the Ci-fkh/lacZ fusion construct containing 3 kb of the 5′ flanking region in the nerve cord is restricted to the ventral ependymal cells, while truncated constructs containing less than 1.7 kb of the 5′ flanking region exhibit expression in the lateral ependymal cells as well as loss of expression in the ventral cells (Di Gregorio et al., 2001). These results suggest that distal regions of the Ci-fkh promoter contain one or more repressor elements responsible for restricting expression to the floor plate. Removal of these elements causes ectopic expression in the two lateral rows of ependymal cells. The same region also contains sequences that are important for the activation of Ci-fkh. The sequence interval between −1.77 and −1.65 kb contains a cis-regulatory sequence responsible for the normal expression in the ventral ependymal cells. Deletion of a 20-bp upstream sequence from −1.65 to −1.63 kb causes the expanded lacZ expression in the lateral ependymal cells. This 20-bp region contains divergent binding sites for Ci-Sna, which is transiently expressed in the lateral ependymal cells (Corbo et al., 1997b). The Ci-Sna repressor binds to this region, excludes expression from the lateral ependymal cells, and restricts Ci-fkh expression to the ventral ependymal cells (Di Gregorio et al., 2001). Deletion analyses of the Ci-Sna promoter also showed that the 5′ flanking regions from −0.96 to −0.9 kb and from −0.61 to −0.55 kb are important for expression in the CNS.
The Ciona msx gene Ci-msxb has a complex expression pattern during embryogenesis (Aniello et al., 1999). At the beginning of gastrulation, the transcript appears in the precursors of mesenchyme cells, muscle, nerve cord, endodermal strand, CNS, and primordial pharynx (stomodæum). In the larva, Ci-msxb is expressed in the primordial pharynx and in the neck, connecting the brain vesicle and the visceral ganglion. In the juvenile, the expression is restricted to the pharynx and neural gland. A 3.8-kb upstream region of Ci-msxb contains regulatory elements that can recapitulate the endogenous gene expression pattern (Russo et al., 2004). The region extending from position −239 to −97 of the Ci-msxb promoter contains cis-regulatory elements responsible for restricted expression of the lacZ reporter in the nervous system. A 30-bp sequence, N3, in this region shows specific binding of nuclear proteins and can direct transcription from basal promoters in the nervous system.
In ascidians, otx is the earliest gene known to be expressed in the prospective neural tissue. It plays an important role in early neural development (Wada et al., 1996, 2002, 2004; Wada and Saiga, 1999; Hudson and Lemaire, 2001; Satou et al., 2001c). Bertrand et al. (2003) identified a cis-regulatory element necessary for activation of Ciona otx in the precursor cells of the anterior neural plate (a6.5 blastomeres). This element (a-element) is located between −1541 and −1418. When the a-element was placed upstream of the basal promoter of Ci-Bra, Ci-Gsx, or HrMA4a, it directed transcription in neural precursors at the 110-cell stage. In the absence of a basal promoter, the a-element had no activity, showing that it acts as an enhancer rather than as a promoter. Extensive analyses of function of the a-element established that neural fate is induced by FGF9/16/20, acting via a combination of maternal GATA (GATAa) and Ets (Ets1/2) transcription factors, which synergistically bind and activate the a-element (Bertrand et al., 2003). The function of cis-regulatory regions has been also examined in the H. roretzi otx gene, Hroth (Oda-Ishii and Saiga, 2003). Three upstream sequences of 10, 16, and 70 bp, respectively, are shown to be responsible for Hroth expression in the brain vesicle.
Expression of the paralogous group 3 Hox gene (CiHox3) of C. intestinalis is confined to the anteriormost region of the visceral ganglion of the larval CNS (Locascio et al., 1999). In vivo analysis of CiHox3 promoter function revealed that a cis-regulatory element(s) necessary for expression in the CNS is present in the 80 bp of the 5′ flanking region from −1943 to −1864 of CiHox3. This enhancer activates transcription not only in the visceral ganglion, but also in the brain vesicle. Furthermore, deletion of upstream regions of CiHox3 resulted in both the loss of CNS expression and the gain of ectopic expression in mesenchyme. Therefore, there seem to be both positive and negative regulatory elements in the CiHox3 upstream region.
Pituitary homeobox (pitx) genes are critical molecular determinants of various processes of craniofacial development, including pituitary organogenesis, in vertebrates. The C. intestinalis pitx gene Ci-pitx produces two distinct mRNA variants, Ci-pitxa/b and Ci-pitxc, which are expressed in mutually exclusive embryonic domains (Christiaen et al., 2005). The Ci-pitxa/b isoform is expressed in the anterior neural boundary (ANB) at the tailbud stage and in the tail muscle and the stomodæum (oral siphon rudiment/primordial pharynx). On the other hand, Ci-pitxc is expressed asymmetrically in the epidermis, left visceral ganglion, and left posterior trunk endoderm as well as in ocellus photoreceptor cells. Separate promoters and regulatory elements regulate the transcription of these two variants. Interspecific comparison of pitx locus between C. intestinalis and C. savignyi identified 10 conserved noncoding regions (Christiaen et al., 2005). Among these regions, two regions, D1 and P2, were shown to contain cis-regulatory modules, each of which drives gene expression in a distinct subset of cells in the ABN/stomodæal expression domain.
HETEROLOGOUS BASAL PROMOTERS
Activity of cis-regulatory elements has been demonstrated by connecting them with heterologous basal promoters derived from various ascidian genes. The most widely used basal promoter is the Ci-fkh basal promoter (Erives et al., 1998; Di Gregorio et al., 2001). The Ci-fkh basal promoter is a TATA-less promoter that can be activated by different promoters in various tissues (Erives et al., 1998; Di Gregorio and Levine, 1999; Harafuji et al., 2002; Davidson and Levine, 2003; Boffelli et al., 2004; Johnson et al., 2004; Christiaen et al., 2005). This basal promoter was used for a genome wide-enhancer screen (Harafuji et al., 2002) and for functional evaluation of conserved noncoding genomic sequences (Boffelli et al., 2004).
The B4.1 enhancer of Ci-Sna activates transcription from the basal promoter of Ci-Bra only when fused in the syn orientation (Erives et al., 1998). In contrast, the same enhancer activates transcription in both orientations when it is placed upstream of the Ci-fkh basal promoter. Erives et al. (1998) pointed out that the Ci-Bra promoter contains a perfect TATAAA sequence while the Ci-fkh promoter appears to be TATA-less. The Ci-Bra basal promoter is also activated by enhancers of Ci-fkh (Di Gregorio et al., 2001) and Ci-otx (Bertrand et al., 2003).
Other basal promoters that have been used for functional assays of heterologous cis-regulatory sequences include those of Ci-Sna (Di Gregorio et al., 2001), CiHox3 (Fanelli et al., 2003; Russo et al., 2004), HrMA4a (Bertrand et al., 2003), and Ci-Gsx (Bertrand et al., 2003). Another example of promoter activation by heterologous cis-regulatory sequences is enhancer trapping in the Ciona genomes using the Minos transposon. In germline transgenesis using a Minos vector containing an upstream region of CiTPO, which is an endostyle-specific gene encoding thyroid peroxidase, Sasakura and his colleagues obtained enhancer trap lines of C. intestinalis (Sasakura et al., 2003; Awazu et al., 2004) and C. savignyi (Matsuoka et al., 2004).
INTERSPECIFIC ACTIVITY OF CIS-REGULATORY DNA
Studies on the activity of cis-regulatory DNAs in different species, both closely- and distantly-related, can provide clues to understand the diversity and evolution of organisms. Interchangeability of cis-regulatory regions also has a practical application because it allows us to use a regulatory DNA from one species as a molecular tool to express genes and proteins in another species. It has been shown that cis-regulatory regions are usually interchangeable between C. intestinalis and C. savignyi (Nakatani et al., 1999; Deschet et al., 2003; Mastuoka et al., 2004; Johnson et al., 2004).
Ascidians are divided into two major orders, Enterogona and Pleurogona. Ciona belongs to the former order, which has a single gonad, and Halocynthia to the latter, which has a pair of gonads. These two orders diverged early during the evolutionary history of ascidians (Wada et al., 1992; Wada, 1998; Swalla et al., 2000). Nonetheless, Halo-cynthia cis-regulatory regions tested so far often show authentic expression patterns in Ciona, and vice versa. For example, a Halocynthia muscle actin promoter is specifically activated in tail muscle cells of C. savignyi (Hikosaka et al., 1993) and C. intestinalis (Corbo et al., 1997a). The −3.5 kb upstream region of Ci-Bra drives a lacZ reporter in notochord cells of H. roretzi (Takahashi et al., 1999b). When the Ci-Bra promoter was deleted down to −250 bp, however, the reporter gene was not expressed in the notochord of H. roretzi, but it was expressed in the notochord of C. intestinalis. Thus some alterations seem to have occurred in organization of promoters during evolution of Ciona and Halocynthia (Takahashi et al., 1999b). Promoters of muscle actin genes from M. oculata and M. occulta, which are diverged members of the order Pleurogona (Hadfield et al., 1995; Swalla et al., 2000), can also drive reporter gene expression in muscle cells of C. intestinalis (Kusakabe et al., 1996).
Evolutionary changes of promoter activity have been examined between two closely-related species, M. oculata and M. occulta, with different modes of development (Kusakabe et al., 1996). Molgula oculata shows indirect development with a tadpole larva (urodele development), while M. occulta has lost the tailed larva, a mode known as anural development. The anural larva of M. occulta lacks a neural sensory organ and a tail with a differentiated notochord and muscle cells. Molgula occulta produces notochord and muscle precursor cells during embryonic development but they remain undifferentiated in the posterior region of the larva (Swalla and Jeffery, 1990). The urodele species M. oculata has a larval muscle actin gene MocuMA1, which is single-copy and intronless (Kusakabe et al., 1996). The anural species M. occulta has two paralogous muscle actin genes, MoccMA1a and MoccMA1b, which are also intron-less and likely to be orthologous to MocuMA1. The coding regions of MoccMA1a and MoccMA1b genes contain critical deletions and/or insertions that would make their translated proteins nonfunctional actins, suggesting they are pseudogenes. A fusion gene construct of the 702-bp upstream region of MocuMA1 fused with lacZ is expressed in the tail muscle cells of urodele embryos. Interestingly, this construct is also expressed in vestigial muscle cells of the M. occulta anural larva, suggesting that transcription factors responsible for muscle-specific expression of muscle actin genes have been retained in M. occulta embryos. The function of muscle actin pseudogene promoters in the anural species M. occulta also has been investigated by microinjecting promoter-lacZ fusion constructs into C. intestinalis eggs. The results indicate that both MoccMA1a and MoccMA1b promoters retain muscle-specific activity although it is reduced in MoccMA1b. These analyses on muscle actin genes and their promoters suggest that the regression of muscle cell differentiation in the anural embryo of M. occulta is mediated by loss-of-function mutations of muscle actin genes rather than by changes in muscle-specific transcription factors.
The activity of ascidian cis-regulatory DNAs has been examined in vertebrates. Locascio et al. (1999) generated transgenic mouse embryos bearing a lacZ reporter connected with CiHox3 regulatory regions. When a 2.3-kb genomic fragment including the region involved in neural-specific CiHox3 expression in Ciona embryos was used, the transgenic embryos consistently gave reporter expression in a pattern reminiscent of mouse Hox regulatory elements. However, when they tested a smaller promoter fragment that was responsible for the specific reporter gene expression in the Ciona nervous system, the lacZ construct did not generate any specific pattern of expression in transgenic mouse embryos. Conversely, when lacZ constructs of the mouse kreisler-dependent enhancer regions, involved in the expression of Hoxa3 and Hoxb3 in the hindbrain, were elec-troporated into Ciona embryos, no nervous system-specific expression was observed (Locascio et al., 1999). These results suggest that the regulation of Hox3 gene expression in the mouse and ascidian CNS does not appear to be mediated by the same highly conserved elements. Mice and ascidians might use different regulatory elements and components for Hox3 expression, and/or similar components might be involved but they have diverged and are unable to function across a large evolutionary distance (Locascio et al., 1999).
BIOINFORMATICS AND COMPARATIVE GENOMICS
Given the increasing number of sequenced metazoan genomes, the prediction of cis-regulatory elements by computational methods has become an attractive approach to identify cis-regulatory DNAs on a genomic scale (Ohler and Niemann, 2001; Markstein and Levine, 2002; Halfon and Michelson, 2002). In addition to its phylogenetic proximity to vertebrates, the compactness of the genome and the manipulability of gene expression and function make ascidians attractive model organisms to study cis-regulatory sequences by computational approaches. Furthermore, the availability of the whole-genome sequences from two Ciona species and intraspecies polymorphisms allows comparative genomics approaches to identify cis-regulatory DNAs. Recently published studies using different approaches demonstrate that bioinformatics and comparative genomics can help identify cis-regulatory DNAs in the ascidian genome (Boffelli et al., 2004; Christiaen et al., 2005; Johnson et al., 2004; Kusakabe et al., 2004).
Genes regulated by the same transcription factors are expected to share cis-regulatory elements in their flanking and/or intronic noncoding regions. DNA motifs over-represented in 5′ flanking regions of potentially co-regulated genes were identified by computational analyses of 5′ flanking regions of 50 tissue-specific genes from genome databases of C. intestinalis and C. savignyi (Kusakabe et al., 2004) (Fig. 2). Three groups of potentially co-regulated genes (photoreceptor, pan-neuronal, or muscle-specific gene groups) were selected according to their spatial expression patterns, which were determined by whole-mount in situ hybridization. Several DNA motifs were distributed predominantly in upstream regions of the co-regulated gene groups (Fig. 2A). Some of the motifs show substantial similarities to binding sites for known transcription factors, while others show no distinct similarities to known transcription factor binding sites. Furthermore, some of the muscle-specific motifs are enriched in regions that are located specific distances from the translational start site (Fig. 2B). For example, one motif, called M2, is preferentially distributed to regions from −200 to −100 bp relative to the translational start site. Another motif, M1, is more frequently found in the distal half than in the proximal half of the 1000-bp upstream regions. An M9 site, which contains a GC-core E-box, is often found downstream of the most proximal M2 site. As already mentioned in a previous section of this article, in vivo functional analysis of M2 sites in muscle-specific promoters demonstrated that M2 sites are critical for the muscle-specific expression of some genes (Fig. 2C), suggesting the validity of computational prediction of cis-regulatory elements.
Sequence comparisons between the genomes of organisms separated by varying degrees of evolutionary distance currently serve as an essential means to identify genes as well as cis-regulatory elements (Ansari-Lari et al., 1998; Nobrega et al., 2003; Thomas et al., 2003; Boffelli et al., 2003, 2004). Both intraspecific and interspecific comparisons have been used to identify noncoding regulatory elements in Ciona. The genome of C. intestinalis shows a high rate of allelic polymorphism, with an average 1.2% of the nucleotides differing between chromosome pairs of a single individual (Dehal et al., 2002). This high degree of allelic variation, more than 15-fold that noted in humans, is probably a consequence of the large effective population size of C. intestinalis. Boffelli et al. (2004) exploited sequence variation within C. intestinalis to computationally identify regions subjected to fast and slow rates of evolution. They determined the extent of sequence polymorphism in several C. intestinalis subpopulations collected at multiple locations worldwide. Regions with low mutation rates efficiently demarcated functionally constrained sequences: these include a set of noncoding cis-regulatory elements as well as the location of coding sequences (Boffelli et al., 2004). By in vivo reporter assays using C. intestinalis embryos, slow-evolving regions in the 5′ flanking regions of Ci-fkh and Ci-Sna were shown to contain tissue-specific cis-regulatory elements, while fast-evolving regions failed to drive gene expression.
The genetic distance between the two Ciona species, C. intestinalis and C. savignyi, is so large that unconstrained sequences do not display more similarity than expected by chance (Johnson et al., 2004) (Fig. 3). Nevertheless, the two species are morphologically quite similar to each other, and they are virtually identical in embryogenesis. In fact, interspecific hybrids can be obtained, and are easily reared to the larval stage (Byrd and Lambert, 2000). This suggests that the essential mechanisms of early development are conserved between the two species. Johnson et al. (2004) compared nucleotide sequences of eight C. intestinalis and C. savignyi loci, including troponin I, synaptotagmin, α-tubu-lin, Noto9, forkhead (FoxA-a), snail, tropomyosin-like, and Brachyury. Coding exons showed a high degree of sequence similarity, whereas untranslated regions, introns, and intergenic regions rarely contained conservation beyond what was expected by chance (Fig. 3). Against this background of extremely low noncoding sequence similarity, the intergenic regions of these loci contain short significantly conserved sequences located 5′ to the predicted start of transcription. LacZ reporter constructs that contained these conserved noncoding sequences recapitulated endogenous expression patterns of the genes in both species. As a case study for the sequence-guided dissection of gene regulation and expression, Johnson et al. (2004) further conducted a detailed functional and computational analysis for the troponin I gene (TnI). They identified a 363-bp minimally sufficient regulatory region (MSRR) of TnI, which contains four highly conserved sites and is sufficient for strong muscle-specific expression in embryos. Deletion and mutation analyses of TnI MSRR demonstrated that the four short conserved regions contain cis-regulatory elements important for the muscle-specific gene expression. One of the highly conserved regions contains putative binding sites for myogenic bHLH transcription factors, and mutations of these sites greatly diminish muscle-specific activity. Interestingly, as described in a previous section, this conserved region also contains a sequence that perfectly matches the muscle-specific motif M9 identified by an independent computational study of Ciona muscle-specific promoters (Kusakabe et al., 2004). Sequence comparisons between orthologous loci of C. intestinalis and C. savignyi have also been used to identify cis-regulatory regions of Otx (Bertrand et al., 2003), Brachyury (Yagi et al., 2004a), Musashi (Awazu et al., 2004), and Pitx (Christiaen et al., 2005). These studies consistently illustrate the value of the interspecific comparison of the Ciona genomes for analysis of genetic regulatory systems.
Fig. 3.
Comparison of 5′ flanking regions of two brain-specific genes, BCO/RPE65 (Nakashima et al., 2003) (A) and opsin1 (Kusakabe et al., 2001) (B), between C. intestinalis and C. savignyi by dot-matrix plots. In contrast to the well-conserved protein-coding exons (black dotted circles), introns, untranslated regions, and intergenic regions show almost no similarity. Conserved sequences found in these noncoding regions (red dotted circles) are good candidates for cis-regulatory elements.
FUTURE PERSPECTIVES: GENOME-WIDE APPROACHES TO GENETIC REGULATORY NETWORKS
As discussed above, in silico approaches have become of practical use for the analysis of cis-regulatory DNAs in ascidians. It has been possible to perform genome-wide searches for tissue-specific enhancers in Ciona by simply attaching random genomic DNA fragments to a basal promoter and then electroporating these into developing embryos (Harafuji et al., 2002). Genome-wide gene expression profiles can be obtained from high-throughput in situ hybridization (Makabe et al., 2001; Satou et al., 2001a, 2002; Nishikata et al., 2001; Kusakabe et al., 2002; Fujiwara et al., 2002; Ogasawara et al., 2002) and DNA microarray (Ishibashi et al., 2003; Azumi et al., 2003, 2004) analyses. The sequences and distribution patterns of DNA motifs shared by a group of co-expressed genes can be incorporated into a model promoter structure, and the model can then be used for genome-wide prediction of co-regulated promoters (Berman et al., 2002; Markstein et al., 2002; Halfon et al., 2002; Bulyk et al., 2004; Wenick and Hobert, 2004). This strategy could lead to identification of the co-regulated gene batteries, which could reveal the molecular bases of cell and tissue identity. A genome-wide analysis identified 389 transcription factor genes in the C. intestinalis genome (Imai et al., 2004). Among these genes, cDNA clones are available for 352 genes, and their expression profiles have been analyzed. Use of these transcription factor resources should facilitate identification of transcription factors that recognize cis-regulatory elements. The genome-wide application of these resources and technologies will break new ground in the field of gene network studies in chordates.
Acknowledgments
I thank Drs. Brad Davidson and William R. Jeffery for critical reading of the manuscript. I also thank Prof. Motoyuki Tsuda for discussion and encouragement. Our research described here was supported in part by Grants-in-Aid for Scientific Research from the MEXT and JSPS.
