The tunicate Ciona intestinalis larva has a simple central nervous system (CNS), consisting of fewer than 400 cells, which is homologous to the vertebrate CNS. Recent studies have revealed neuronal types and networks in the larval CNS of C. intestinalis, yet their cell lineage and the molecular mechanism by which particular types of neurons are specified and differentiate remain poorly understood. Here, we report cell lineage origin and a cis-regulatory module for the anterior caudal inhibitory neurons (ACINs), a putative component of the central pattern generator regulating swimming locomotion. The vesicular GABA/glycine transporter gene Ci-VGAT, a specific marker for GABAergic/glycinergic neurons, is expressed in distinct sets of neurons, including ACINs of the tail nerve cord and others in the brain vesicle and motor ganglion. Comparative genomics analysis between C. intestinalis and Ciona savignyi and functional analysis in vivo identified the cis-regulatory module responsible for Ci-VGAT expression in ACINs. Our cell lineage analyses inferred that ACINs derive from A11.116 cells, which have been thought to solely give rise to glial ependymal cells of the lateral wall of the nerve cord. The present findings will provide a solid basis for future studies addressing the molecular mechanism underlying specification of ACINs, which play a critical role in controlling larval locomotion.
In a previous study, we identified and characterized the vesicular GABA/glycine transporter gene, Ci-VGAT, which is a specific marker for GABAergic/glycinergic neurons, including ACINs (Yoshida et al. 2004). In the present study, we identified a cis-regulatory module responsible for Ci-VGAT expression in ACINs and documented the cell lineage of this neuronal cell type. Comparative genomics analysis in combination with in vivo reporter assay identified the cis-regulatory module responsible for Ci-VGAT expression in ACINs. Considering these data and the known expression profiles of transcription factors and signaling molecules during early embryogenesis, we propose a molecular and cellular mechanism by which a subtype of GABAergic/glycinergic neurons is specified and differentiates in the CNS of the Ciona larva.
Materials and methods
Animals and embryos
Mature adults of C. intestinalis were collected from the fishing harbor of Murotsu, Hyogo, Japan, or obtained from the National BioResource Project (NBRP). They were maintained in indoor tanks of artificial seawater (Marine Art BR, Senju Seiyaku, Osaka, Japan) at 18°C. Eggs and sperm were obtained surgically from the gonoducts and fertilized in vitro. The embryos were allowed to develop at 18°C in artificial seawater.
The generation of the fusion gene plasmid, pCi-VGAT-EGFP, was described previously (Yoshida et al. 2004). pCi-VGAT(−2668/−1)-Kaede was made by inserting the whole upstream region of Ci-VGAT excised from pCi-VGAT-EGFP into the SalI/BamHI sites of the pSP-Kaede vector (Hozumi et al. 2010). To generate pCi-VGAT(−2668/−1ΔCNR)-Kaede, pCi-VGAT(for −1390/−1)-Kaede and pCi-VGAT(−890/−1)-Kaede, the whole pCi-VGAT(−2668/−1)-Kaede plasmid except for the region that we wanted to delete was amplified by polymerase chain reaction (PCR) using a high-fidelity DNA polymerase (PrimeSTAR HS DNA polymerase, Takara Bio, Japan) with a pair of primers (5′-TAAGATCTCACCGATTTGTTTTATTGGTT-3′ and 5′-AAAGATCTCATTCCAAACCTGTATTATCC-3′ for ΔCNR; 5′-GGCTCGAGAATACAGGTTTGGAATGAAC-3′ and 5′-CCCTCGAGTATTAATTGTAGCCGCGTT-3′ for −1390/−1; 5′-CACTCGAGACCGATTTGTTTTATTGGTT -3′ and 5′-CCCTCGAGTATTAATTGTAGCCGCGTT-3′ for −890/−1), digested with BglII (ΔCNR) or XhoI (−1390/−1, −890/−1) and then self-ligated. To generate the Kaede fusion construct pCi-fkh-min-Kaede, the minimal promoter of the forkhead/HNF-3β gene Ci-fkh (from −158 to +20 relative to the translation start site; Di Gregorio et al. 2001) was amplified by PCR from C. intestinalis genomic DNA with a pair of primers (5′-GACTGTCGACTCTTTGACCAATAATTTCGC-3′ and 5′-ACGGATCCGGTGGAGACGACAACATCAT-3′) and inserted into the SalI/BamHI sites of pSP-Kaede. To generate pCi-VGAT(CNR+fkh)-Kaede, the conserved upstream region, VGAT-CNR, was amplified by PCR with a pair of primers (5′-ACGAGCTCTGACCGGAAATCGATCG-3′ and 5′-CACTGGAGAACAAATCGGTGTTTAATTGC-3′) and inserted immediately upstream of the Ci-fkh minimal promoter.
Electroporation of reporter DNA constructs and immunofluorescent detection of Kaede
Circular plasmid DNA constructs were electroporated into fertilized eggs of C. intestinalis as described by Corbo et al. (1997). After developing into larvae (24 h after fertilization at 18°C), they were fixed with 10% formalin in artificial seawater for 1 h at room temperature. Fluorescence was observed with fluorescent microscopes SZX12-RFL3 (Olympus, Japan) and Axioplan 2 (Carl Zeiss, Oberkochen, Germany). The localization of Kaede was visualized by whole-mount immunofluorescent staining with a rabbit anti-Kaede polyclonal antibody (PM012; Medical & Biological Laboratories Co., Ltd., Nagoya, Japan) as previously described (Horie et al. 2010). The primary antibody was diluted 1:1000 in 10% goat serum in phosphate-buffered saline (PBS) containing 0.1% Triton X-100 (TPBS). As the secondary antibody, an Alexa 488-conjugated anti-mouse IgG goat antibody (Molecular Probes, Inc., Eugene, OR, USA) was used. Light microscopic and fluorescent images were photographed by using a digital CCD camera VB-7010 (Keyence, Osaka, Japan).
Blastomere labeling and immunofluorescent staining of Ci-VGAT
Blastomeres were labeled with CM-DiI (Molecular Probes) dissolved in colza oil at a concentration of 10 mg/mL. When the embryos reached the 64- or 110-cell stage, an oil droplet containing CM-DiI was injected into A7.4 or A7.8 blastomeres on each side of the embryo at the 64-cell stage or into an A8.15 or A8.16 blastomere on one side of the embryo at the 110-cell stage, respectively. After developing into larvae (24 h after fertilization at 18°C), they were fixed with 10% formalin in artificial seawater for 1 h at room temperature. Fixed specimens were subjected to whole-mount immunostaining as previously described (Horie et al. 2010). The anti-Ci-VGAT antiserum (Horie et al. 2010) was diluted 1:1000 in 10% goat serum in TPBS and used as the primary antibody. As the secondary antibody, an Alexa 488-conjugated anti-mouse IgG goat antibody (Molecular Probes) was used. The stained specimens were mounted in Vectashield with 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI) under a confocal microscope (Leica SP5, Leica, Germany).
Visualization of prospective GABAergic/glycinergic neurons with Kaede under the control of the cis-regulatory region of Ci-VGAT
The 2668-bp 5′ flanking region immediately upstream of the start codon of Ci-VGAT was shown to partially recapitulate the endogenous expression patterns when EGFP was used as a reporter (pCi-VGAT-EGFP; Yoshida et al. 2004). In the CNS, Ci-VGAT is expressed in the brain vesicle, motor ganglion (also called ‘visceral ganglion’; we adopt here a terminology recently proposed by Nishino et al. ) and ACINs (Fig. 1; Yoshida et al. 2004; Horie et al. 2010), while EGFP signals were only detected in the brain vesicle and ACINs in the larvae electroporated with pCi-VGAT-EGFP (Yoshida et al. 2004). In this study, we used Kaede (Ando et al. 2002), in place of EGFP, as a reporter because we found that Kaede was expressed more efficiently than EGFP in C. intestinalis embryos and larvae when they were driven by the same cis-regulatory regions. For example, none of the developed larvae (n = 62) showed EGFP signals in the motor ganglion when pCi-VGAT-EGFP was electroporated (Yoshida et al. 2004), whereas Kaede signals in the motor ganglion were found in 30% of the larvae (n = 231) when pCi-VGAT(−2668/−1)-Kaede was electroporated. Therefore, Kaede expression driven by the 2668-bp 5′ flanking region recapitulated the endogenous patterns of both Ci-VGAT mRNA and Ci-VGAT protein localization (Figs 2C, 3A).
An upstream noncoding region of the VGATgene conserved betweenC. intestinalisandC. savignyi
Evolutionarily conserved noncoding sequences are good candidates for cis-regulatory regions (Boffelli et al. 2004; Kusakabe 2005; Yoshida et al. 2007). In order to predict the cis-regulatory regions that control neuron type-specific expression of Ci-VGAT, the nucleotide sequence of the 5′ flanking region of the Ci-VGAT was compared with that of the C. savignyi orthologue Cs-VGAT (Ensembl Gene ID: ENSCSAVG00000010463). By using VISTA tools (Frazer et al. 2004), we found a highly conserved non-coding region (CNR) in the upstream intergenic region from −1266 to −1068 of Ci-VGAT (Fig. 2B). This highly conserved 199-bp region was named VGAT-CNR for conserved noncoding region of Ci-VGAT. Blast analysis revealed that there are no other sequences significantly similar to VGAT-CNR in the C. intestinalis genome (data not shown).
VGAT-CNR is a cis-regulatory module responsible for ACIN-specific gene expression
To analyze VGAT-CNR function, we deleted VGAT-CNR from the pCi-VGAT(−2668/−1)-Kaede construct. The deletion of VGAT-CNR resulted in the complete loss of Kaede expression in ACINs, whereas the expression in the brain vesicle and motor ganglion was unaffected (−2668/−1ΔCNR in Figs 2C, 3B). When the 5′ flanking region was deleted down to immediately upstream of VGAT-CNR, Kaede was expressed in a proper expression pattern of Ci-VGAT (−1390/−1 in Figs 2C, 3C). In contrast, when the upstream region was deleted down to 890 bp from the start codon, the resulting construct lacking VGAT-CNR revealed no reporter expression in ACINs (−890/−1 in Figs 2C, 3D). These results strongly suggest that VGAT-CNR is necessary for Ci-VGAT expression in ACINs. Our results also suggest that VGAT-CNR is dispensable for expression in the brain vescile and motor ganglion.
We further examined whether VGAT-CNR alone is sufficient for gene expression in ACINs or other cis-regulatory sequences located downstream from VGAT-CNR are also required. To test this, VGAT-CNR was connected to the minimal promoter of the C. intestinalis forkhead gene (Ci-fkh; Di Gregorio et al. 2001). The Ci-fkh minimal promoter alone did not drive Kaede expression in the CNS of the larva (fkh min in Figs 2C, 3F), whereas Kaede signals were frequently observed in ACINs (32%) when VGAT-CNR was placed upstream of the Ci-fkh minimal promoter (CNR+fkh in Figs 2C, 3E). In these larvae, however, the reporter expression was much less frequent in the brain vesicle (4%) and absent in the motor ganglion (CNR+fkh in Figs 2C, 3E). Thus, VGAT-CNR is sufficient to drive the reporter expression in ACINs. The reporter expression was also detected in mesenchyme cells in the larvae electroporated with the CNR-fkh construct (74%). This may be partly due to an intrinsic property of the Ci-fkh minimal promoter, because it can by itself activate transcription in this cell type (11%) (fkh min in Figs 2C, 3F). Our result also suggests that VGAT-CNR enhances mesenchymal expression from the Ci-fkh minimal promoter (cf. CNR+fkh and fkh min in Fig. 2C).
The ACINs are derived from lateral nerve cord precursors
In C. intestinalis, expression profiles of transcription factors have been revealed at a single-cell resolution up to the mid-tailbud stage (Imai et al. 2004, 2006, 2009; Ikuta & Saiga 2007). Therefore, the elucidation of cell lineage in combination with cis-regulatory region analyses would provide important clues to identify transcription factors that regulate specification and differentiation of a particular neuronal type as well as to decipher the gene regulatory networks starting from a fertilized egg, as recently demonstrated for the cholinergic neurons in the motor ganglion of the C. intestinalis larva (Stolfi & Levine 2011). To determine the cell lineage of ACINs, we labeled two or one blastomeres with DiI at the 64- or 110-cell stage, respectively. After the DiI-labeled embryos developed to the swimming larval stage, we visualized GABAergic/glycinergic neurons and nuclei with anti-Ci-VGAT and DAPI, respectively. The cell lineage of ACINs was inferred based on the confocal observation of triple-labeled specimens.
The bilateral positioning of ACINs posterior to the motor ganglion indicates that these neurons originate from the A-line neural lineages. We therefore started our cell lineage analyses by labeling A7.4 or A7.8 blastomeres, which are precursors of the A-line neural lineages, with DiI at the 64-cell stage (Fig. 4A). The A7.4 blastomeres are known to generate the ventral nerve cord and a part of the brain vesicle, while the A7.8 blastomeres give rise to the lateral nerve cord, motor neurons, and secondary muscle cells (Nishida 1987; Nicol & Meinertzhagen 1988; Cole & Meinertzhagen 2004). Anti-Ci-VGAT positive ACINs were found among A7.8 descendants but not among those of A7.4 (Fig. 4B,C; see also videos S1 and S2). The results clearly indicate that ACINs are derived from the A7.8 lineage.
Next, we labeled A8.15 or A8.16 blastomeres (Fig. 4D) that are the two daughter cells of A7.8 at the 110-cell stage. The Ciona larva has two bilateral pairs of ACINs, and hence two ACINs are present in each side of the nerve cord. The Ci-VGAT signals in the two ACINs on the same side were overlapped with DiI fluorescence when A8.15 had been labeled, whereas they did not overlap when A8.16 had been labeled (Fig. 4E,F; see also videos S3 and S4). Thus, we concluded that ACINs are all derived from the A8.15 lineage.
The cell lineage of ACINs was further inferred by counting nuclei of the DiI-labeled descendants of the A8.15 cell and comparing the distribution patterns of the nuclei with the previously reported mitotic profiles of the C. intestinalis CNS (Cole & Meinertzhagen 2004; Imai et al. 2009; Stolfi & Levine 2011). In the DiI-labeled region anterior to the ACINs, there are 21 or 22 nuclei (n = 3); this region includes five motor neurons (A10.57, A11.117, A11.118, A12.239 and A13.474) and other descendants of A11.119 and A11.120. Posterior to ACINs, 14–16 nuclei (n = 3) are found in the DiI-labeled region, which solely includes lateral nerve cord cells; these cells are presumed to be descendants of A11.115. Two nuclei are located between the two ACINs (n = 3), and these four cells are inferred to be descendants of A11.116. The proposed relationship between cell lineage and cell distribution is summarized in Figure 5A.
The nervous system contains numerous types of neurons, each of which exhibits distinct morphological and physiological properties. The different types of neurons are characterized by sets of genes specifically expressed in particular types of neurons, such as those specifying neurotransmitter phenotypes of neurons. On the other hand, different types of neurons share proteins that define basic neuronal features, such as proteins involved in membrane excitation, axonal architecture, and synaptic vesicle regulation. Regulation of neuron-specific gene expression is a subject of broad interest and has been extensively studied by many researchers in different model organisms. Nonetheless, the mechanisms by which genes are specifically expressed in particular types of neurons remain largely unknown because of the complexity of both the regulatory system and the nervous system organization, especially in vertebrates. The simple neuronal and genomic organization of ascidian larvae should provide us with an unparalleled opportunity to investigate gene regulatory networks controlling particular neuronal types in chordates.
A neural circuit controlling swimming behaviors of the ascidian larvae has been proposed to consist of several bilateral pairs of cholinergic motor neurons in the motor ganglion and two bilateral pairs of ACINs (Horie et al. 2009, 2010; Nishino et al. 2010). We previously identified and characterized Ci-VGAT, which is expressed in subsets of neurons, including ACINs, in the larval CNS of C. intestinalis (Yoshida et al. 2004). In order to decipher the gene regulatory network controlling specification of ACINs, in this study, we dissected the 5′ regulatory sequence of Ci-VGAT, identified the cis-regulatory module for the expression in these interneurons and revealed their cell lineage origin. Together with the established regulatory codes at single-cell resolution during embryogenesis (Imai et al. 2009), the present findings should provide a basis for the complete elucidation of the developmental program from fertilization to terminal differentiation of a particular type of neurons in a chordate CNS.
Highly conserved noncoding regions as a cis-regulatory module
The present in vivo analysis of the cis-regulatory regions of Ci-VGAT revealed several functional properties of VGAT-CNR. First, VGAT-CNR specifically activates transcription in ACINs and is dispensable for it in the brain vesicle and motor ganglion. Second, VGAT-CNR does not require other cis-acting elements in order to activate transcription from a basal promoter. Third, VGAT-CNR can activate transcription in ACINs from a heterologous basal promoter. Fourth, VGAT-CNR as a whole is not a repressor, although the possibility remains that it contains one or more repressor elements that confine its activity to ACINs. These properties of the VGAT-CNR indicate that it acts as a typical cis-regulatory module similar to those known in Drosophila and sea urchins (Howard & Davidson 2004).
This study is the first report to identify a cis-regulatory module for a particular type of neurons in C. intestinalis. cis-Regulatory modules are information processing devices hardwired into the genomic DNA sequence, and act through protein-DNA and protein–protein interactions on them (Howard & Davidson 2004). Thus, the identification of molecular interactions occurring on VGAT-CNR should be a promising next step to uncover the gene regulatory network controlling the development of ACINs.
In accordance with our previous report (Yoshida et al. 2007), the present results also illustrate the limit of detection of cis-regulatory sequences by comparative genomics between C. intestinalis and C. savignyi. Our results suggest that there must be cis-regulatory elements for expression in the brain vesicle and motor ganglion outside of VGAT-CNR. To discover poorly conserved cis-regulatory sequences, different bioinformatics and experimental approaches must be required; such approaches include comparisons of more closely related genomes, namely intraspecific comparisons (Boffelli et al. 2004), and the discovery of highly represented motifs in regulatory regions of a potentially co-regulated gene battery (Kusakabe et al. 2004; Kusakabe 2005). The use of next-generation sequencing, such as ChIP-Seq (Robertson et al. 2007; Jothi et al. 2008) and digital genome footprinting (Hesselberth et al. 2009), is another possible approach. It should be noted that these computational or high-throughput approaches should be combined with experimental verification in vivo, and that some nonconserved cis-regulatory elements might be discovered only by ‘wet’ experiments.
Possible application of VGAT-CNR to the study of neural function
The identification of VGAT-CNR enables us to genetically manipulate the functions of ACINs and the neural circuits containing them. For example, optogenetics using VGAT-CNR in combination with light-gated ion channels (Nagel et al. 2003; Zhang et al. 2006) and a light-activated chloride pump (Han & Boyden 2007; Zhang et al. 2007) would be able to dissect cellular function in the control of swimming behavior. Neural circuit functions can also be addressed by the directed expression of mutant neuronal proteins, such as dominant-negative and hyperactive forms (Okada et al. 2002).
Cell lineage and developmental mechanisms of ACINs
Despite the simplicity of the ascidian larva, the specification and differentiation mechanisms of most of its neurons have not been elucidated yet. This is partly due to the relatively late differentiation of most neurons during development and to the lack of neuronal cell lineage documented up to the mature swimming larval stage. The only exception is the case of cholinergic neurons in the motor ganglion; these cells begin to express specific markers relatively early at the tailbud stage, and their cell lineage and specification mechanisms have been elucidated in some detail (Stolfi & Levine 2011). In the present study, we successfully elucidate the cell lineage of a particular type of neuron whose specification occurs rather late in larval development.
Our results suggest that ACINs are derived from the A11.116 cells (Fig. 5), which have been thought to give rise to only glial ependymal cells of the lateral wall of the tail nerve cord (Nicol & Meinertzhagen 1988; Cole & Meinertzhagen 2004). Because an A11.116 cell will undergo at least two rounds of mitosis (Cole & Meinertzhagen 2004; Ikuta & Saiga 2007), an ACIN and an ependymal cell might be generated by asymmetric division of an A11.116 progeny (Fig. 5B). Another possibility is that A11.116 divides asymmetrically to generate one ACIN progenitor and one ependymal progenitor, each of which divides symmetrically to generate two daughters of a same fate. It has been revealed that A11.116 cells express transcription factor genes, including Hox1, Hox5, Pax6 and SoxB1 (Ikuta & Saiga 2007; Imai et al. 2009). Potential binding sites for these transcription factors are present in VGAT-CNR, and Hox and Pax6 sites are conserved between C. intestinalis and C. savignyi (Fig. 5C). In VGAT-CNR, however, there are also three stretches of highly conserved sequences that do not match the consensus sequences for Hox, Pax6 and SoxB1 (Fig. 5C). Injection of the antisense morpholino oligonucleotides against Hox1 or Hox5 into C. intestinalis eggs did not affect expression of Ci-VGAT in the anterior nerve cord of the larvae (Ikuta et al. 2010), suggesting that ACINs normally developed in these larvae. The combination of these and additional transcription factors and their differential regulation between sister cells by extracellular signals may be involved in the specification process of ACINs.
This study was supported in part by Grants-in-Aid for Scientific Research from MEXT (17018018) and JSPS (17310114, 18370089, 22310120, 22657023), and by research grants from Hyogo Science and Technology Association (23I094) and the Institute of Medical Science, University of Tokyo (2010-(1)-102). The work in the group of H.Y. is supported by the Centre National de la Recherche Scientifique, the Université Pierre et Marie Curie and the Agence Nationale de la Recherche (ANR-09-BLAN-0013-01). K.N. is supported by the Global COE program of MEXT. Y.S. was supported by the Innovative Research Support Programs (Pilot Models) from the University of Tsukuba. T.H. is supported by Precursory Research for Embryonic Science and Technology from the Japan Science and Technology Agency. We thank Drs Kohei Hatta, Kenji Watanabe, Yoshiki Higuchi, and Masashi Nakagawa for their valuable discussions; Yuki Miyamoto, Yoko Ikeda and Yasuko Terashima for their technical assistance; the National BioResource Project (NBRP) of MEXT and all members of the Maizuru Fisheries Research Station of Kyoto University and Misaki Marine Biological Station of the University of Tokyo for providing us with C. intestinalis adults; and the Murotsu Fisheries Cooperative and fishermen at Murotsu Port for generously allowing us to use the fishery facility.