Identical sets of methylated and nonmethylated genes in Ciona intestinalis sperm and muscle cells
- 6.3k Downloads
The discovery of gene body methylation, which refers to DNA methylation within gene coding region, suggests an as yet unknown role of DNA methylation at actively transcribed genes. In invertebrates, gene bodies are the primary targets of DNA methylation, and only a subset of expressed genes is modified.
Here we investigate the tissue variability of both the global levels and distribution of 5-methylcytosine (5mC) in the sea squirt Ciona intestinalis. We find that global 5mC content of early developmental embryos is high, but is strikingly reduced in body wall tissues. We chose sperm and adult muscle cells, with high and reduced levels of global 5mC respectively, for genome-wide analysis of 5mC targets. By means of CXXC-affinity purification followed by deep sequencing (CAP-seq), and genome-wide bisulfite sequencing (BS-seq), we designated body-methylated and unmethylated genes in each tissue. Surprisingly, body-methylated and unmethylated gene groups are identical in the sperm and muscle cells. Our analysis of microarray expression data shows that gene body methylation is associated with broad expression throughout development. Moreover, transgenic analysis reveals contrasting gene body methylation at an identical gene-promoter combination when integrated at different genomic sites.
We conclude that gene body methylation is not a direct regulator of tissue specific gene expression in C. intestinalis. Our findings reveal constant targeting of gene body methylation irrespective of cell type, and they emphasize a correlation between gene body methylation and ubiquitously expressed genes. Our transgenic experiments suggest that the promoter does not determine the methylation status of the associated gene body.
KeywordsMethylation Status Gene Body Gene Body Methylation Unmethylated Promoter Gene Body Region
genomewide bisulfite sequencing
CXXC affinity purification and deep sequencing
polymerase chain reaction.
DNA methylation is an epigenetic modification that is widely employed in eukaryotes, including plants, fungi and animals, and serves multiple critical functions [1, 2, 3]. In eukaryotes, this chemical modification is deposited at the 5 position of cytosine by members of the DNA methyltransferase (DNMT) family of enzymes. 5-methylcytosine (5mC) is distributed nonrandomly in the genome. Recent approaches have enabled genomewide studies of 5mC distribution and demonstrated that the pattern changes during early mouse embryogenesis  and adult mouse stem cell differentiation , representing alteration of local epigenetic states in differentiating cells. It has been recognized that some of the changes occur at gene promoters and lead to stable gene silencing [6, 7, 8]. Recently, an unexpected target of DNA methylation has also been revealed. In several animal and plant genomes, the transcribed regions of genes, or “gene bodies”, including exons and introns, have higher levels of DNA methylation than neighboring sequences [9, 10, 11, 12]. Typically, methylation occurs in the intragenic region and falls sharply at the 5′ and 3′ ends of the transcription unit. This characteristic modification pattern can be superimposed onto the regions occupied by elongating RNA polymerase II , raising the possibility that gene body methylation is associated with transcriptional elongation. In line with this hypothesis, gene body methylation often correlates with actively transcribed genes, but not with silent genes, in plants and animals, including human cells [9, 12, 13, 14, 15]. The role of gene body methylation, however, remains enigmatic .
Among invertebrates, gene bodies are the primary targets for DNA methylation. In our previous study, about 60% of all genes were deduced to be methylated in an invertebrate chordate, Ciona intestinalis. Lately, genomewide bisulfite sequencing (BS-seq) of multiple invertebrates, including C. intestinalis, Apis mellifera, Bombyx mori and Nematostella vectensis, confirmed that prominent gene body methylation is a common feature in the invertebrate genomes [13, 14, 17, 18, 19]. Unlike promoter methylation, gene body methylation is not related simply to the gene expression level [9, 10]. A subset of expressed genes is methylated, whereas unmethylated genes are also expressed at equivalent or even higher levels. Evidently, methylated genes are often evolutionarily conserved and encode essential cellular functions [11, 20]. These genes tend to be expressed in a wide variety of tissues, but the mechanism by which a subgroup of expressed genes becomes methylated is not clear. Recent genomewide studies of invertebrate DNA methylation have used a single tissue or whole adult bodies [13, 14, 17, 21, 22]. Thus, dynamic changes of gene body methylation in somatic cells during development or at the tissue differentiation event remain unknown. Motivated to investigate this phenomenon, we decided to study tissue variability of gene body methylation in an invertebrate animal.
The pattern of gene body DNA methylation found in invertebrates is strikingly different from that of globally methylated mammalian genomes . The genome of C. intestinalis provides an archetypal example of the gene body methylation pattern. BS-seq of selected regions of the sperm genome shows very high levels of DNA methylation in gene bodies (almost all CpG sites are methylated in 90% to 100% of the cells) , whereas promoter and intergenic regions are completely unmethylated. This “black or white” modification pattern is not biased toward transposons, which are often unmethylated in this animal [11, 24]. Thus, the major targets of DNA methylation in the C. intestinalis genome are euchromatic regions. In the current study, we first investigated global 5mC content in early developmental stages and adult tissues to find the stages or tissues in which DNA methylation varies. Next, we chose sperm and adult muscle, which have high and low levels of global 5mC, respectively, for genomewide mapping of 5mC. Using CXXC affinity purification and deep sequencing (CAP-seq), together with analysis of published BS-seq data, we designated body-methylated and unmethylated genes in each tissue. The intertissue comparison of gene body methylation demonstrates that an identical set of genes is targeted for methylation in the sperm and muscle cells. This result suggests that gene body methylation is not a major regulator of tissue-specific expression in C. intestinalis. By analyzing microarray expression data, we show that gene body methylation is associated with broad expression throughout development. Additionally, we found that a gene body became methylated or remained unmethylated in two independent transgenic lines, despite possessing the same driving promoter. This finding suggests that the promoter is not the critical determinant of gene body methylation status.
Global 5mC content of Ciona intestinalis genome is kept at high levels in early developmental stages and reduced in adult body wall tissues
In contrast to the constant levels of methylation in different stages of development, a large intertissue difference in global 5mC content was observed in some adult samples. Heart, intestine and reproductive tissues from gonad and mature sperm showed high 5mC content equivalent to early developmental stages (Figure 1A). Accordingly, at a single-gene level, the EF-1α gene body was highly methylated in both intestine and sperm (88% and 96%, respectively) (Figure 1B). On the other hand, levels of 5mC in body wall tissues, such as the cells from tunic (outermost layer of the body), outer mantle (thin epidermis and underlying connective tissue) and muscle, were approximately half those seen in embryos (Figure 1A). The reduced level of global 5mC in the body wall could be due to the locus-specific methylation changes, to an overall average decrease in DNA methylation levels, or to both. In favor of the second possibility, BS-seq of the EF-1α gene body in muscle and tunic cells was about half the level in embryos (54% and 62% of CpG sites, respectively) (Figure 1B). Loss of methylated CpG sites appeared to be random within this methylated domain, with a variable frequency at each CpG site. The global 5mC content was not correlated to the relative expression level of DNMT s in the adult tissues we examined (Additional file 1: Figure S1).
Stable targeting of gene body methylation in sperm and muscle cells
The resulting DNA methylation map clearly reflects a genomewide mosaic methylation pattern comprising methylated and unmethylated domains. Methylated domains regularly colocalize with genes due to comprehensive gene body methylation (Figure 2A). As a result of the mapping, approximately 63.7 Mb of the genome were identified as methylated, of which 83.0% overlapped with annotated genes. Figure 2B shows the positional relationship between genes and methylated domains. The methylated domain is located centrally over a gene with depletion at the gene ends and a further dip just upstream of the transcription start site (TSS). General features of gene body methylation in C. intestinalis sperm are consistent with those in other organisms which harbor gene body methylation [13, 14, 17]. The boundary of each methylated domain is related neither to the first methionine nor to the termination codon (Additional file 1: Figure S4).
Occasionally, two genes are covered by one methylated domain when genes converge in opposite orientations. As shown in Figure 2A (bottom: KH.C9.523.v1.A.ND1-1 and KH.C9.618.v1.A.ND1-1) and Additional file 1: Figure S5, tail-to-tail–oriented genes lack a 3′ drop in methylation, as the methylated domain continues from one gene body to the other. The transcriptional direction of these genes is evident because of the addition of 3′ poly(A) in expressed sequence tags (ESTs) (Ghost Database: Ciona intestinalis genomic and cDNA resources; http://ghost.zool.kyoto-u.ac.jp/cgi-bin/gb2/gbrowse/kh/). Their 5′ promoter and TSS are unmethylated. The genes concerned do not overlap, and no EST within the gap between them has been reported. It is unlikely that the intragenic region is too short to be detected as an unmethylated domain by CAP-seq, because the poly(A) sites of both genes are more than 500 bp apart, which is equivalent to numerous short unmethylated promoters that were successfully detected in this study (for example, in Figure 2A, bottom, between KH.C9.381 and KH.C9.523, and between KH.C9.618 and KH.C9.386). We found 157 similar tail-to-tail–oriented genes that were more than 500 bp apart and associated with a single, uninterrupted methylated domain.
Next, we generated a list of methylated genes in muscle cells to compare with that of sperm. Because adult muscle DNA methylation appeared variable in frequency at each CpG site (Figure 1B), we mapped methylation levels for each CpG site using genomewide BS-seq data of C. intestinalis adult muscle cells . This data set of 20.3 million reads mapped to unique genomic regions of the KH genome assembly and detected 1,200,307 methylated CpG sites (see Methods). The overall methylation statistics for the genome are equivalent to those in the original study. We found that 23.1% of the cytosine residues in the CpG context were methylated, whereas 21.6% were methylated in the study by Zemach et al. . Non-CpG methylation was significantly lower than CpG methylation: 0.2% of cytosines in the CHG context and 0.4% in the CHH context were methylated, which are comparable to the data reported by Zemach et al. (0.3% and 0.3%, respectively) . As shown in Figure 1B, the level of CpG methylation within methylated domains is lower in muscle cells (70.5% on average in the genomewide BS-seq data ) than in sperm cells (97.9% in a total of 100 kb of randomly selected genomic regions analyzed by BS-seq ). Next, we examined gene body methylation status in muscle for 14,480 genes, within which over 60% of CpG sites possessed ≥2 read coverage, corresponding to 94.8% of all gene models. Methylated cytosine residues in the genome were statistically defined (see Methods), and the degree of gene body methylation for each gene was represented by the ratio of methylated CpG sites to all CpG sites in the transcriptional unit. As in sperm, genes in muscle DNA were separated into highly methylated and entirely unmethylated categories (Figure 3B).
Number of methylated and unmethylated genes in sperm and muscle a
We found that the pattern of methylated and unmethylated genes is identical between muscle and sperm. This result implies that radical changes in gene body methylation status are not the major cause of decreased methylation levels in muscle cells. Instead, the reduced level of global 5mC in muscle is likely to be due to the lower average methylation level in methylated domains that are shared with sperm (see Additional file 1: Figure S3A).
Stably or maternally expressed genes are methylated
It has been proposed that DNA methylation in gene body regions regulates splicing [21, 30, 31]. We therefore investigated the relationship between gene methylation status and number of transcriptional variants. About 30% of all gene models were annotated with different transcriptional isoforms (C. intestinalis KH gene models), but we found no significant bias of these genes between the methylated and unmethylated groups (Additional file 2: Table S3).
Role of promoters in establishing gene body methylation pattern
To investigate why line E obtained intense DNA methylation in the GFP gene body, we examined the methylation status of DNA surrounding the insertion site. The results showed that the transgene was inserted into genomic position KHC7:1904808, which is within the first intron of polyglutamine-binding protein gene (KH gene model: KH.C7.662). This is a ubiquitously expressed and gene body methylated gene. Transgene methylation in line E may therefore be due to a positional effect of the surrounding insertion site. Similar cases were seen in transposons in C. intestinalis. Repetitive copies inserted in introns of methylated genes were passively methylated as a part of the methylated domain, whereas other copies located in an unmethylated area were methylation-free. Although these findings do not account for the domain organization of DNA methylation, they emphasize that promoter identity is unlikely to be the primary determinant of gene body methylation status.
Interestingly, although DNA methylation in line E is specifically targeted to the gene body, an unmethylated 5′ domain was detected (Figure 5B). This unmethylated 5′ area across the TSS of the transgene expands from about 800 bp upstream to 1 kb downstream of the TSS, including an exogenous eGFP sequence. The transgene is expressed from the intron of methylated host gene, owing to the unmethylated promoter. Therefore, this transgene recapitulates promoter hypomethylation in spite of dense methylation downstream. Given that endogenous promoters are incompatible with DNA methylation, they are likely to be important components in shaping the 5′ edge of the methylated gene body.
Whereas gene body methylation is the most widely conserved DNA methylation pattern in eukaryotes, its discovery is rather recent. To explore its mechanism and biological significance, genomewide DNA methylation has been investigated in various animals by BS-seq, but so far comparison between different tissues or cell types within a specific species has not been reported. In this first methylome comparison between C. intestinalis sperm and muscle, we found that the targets of gene body methylation are identical in these highly dissimilar cells. During spermatogenesis, sperm-specific gene expression alters as cells mature . Although most genes are silenced when canonical histones are exchanged with protamines, a small number of genes whose expression is essential for sperm cells continue to be expressed. Considering the dynamic alteration of gene expression, the DNA methylation pattern we observed in the sperm genome, which is identical to that in muscle cells, does not reflect the gene expression status of a specific spermatogenesis stage. Given that the paternal genome of C. intestinalis does not undergo genomewide demethylation and reestablishment of methylation pattern after fertilization, the methylation pattern in sperm may be a default epigenetic state.
A constant DNA methylation pattern was also reported in the sea urchin, where methylated and unmethylated genome fractions in sperm, embryos and adult intestines did not appear to exchange, as measured by reassociation kinetics between these genomes . In the current study, we employed higher-resolution methods to investigate genomewide DNA methylation targets in single-gene resolution and found that methylated and unmethylated domains are constant throughout development. The idea that DNA methylation status dynamically changes to regulate gene expression during differentiation appears to be inapplicable to the deuterostome genomes, although the possibility that DNA methylation temporarily changes in specific genomic regions is not completely excluded. An additional possibility is that the distribution of 5-hydroxymethylcytosine (5hmC), generated by oxidation of 5mC by the ten eleven translocation (TET) family of enzymes, is dynamically regulated instead. However, it remains an open question at present when TET proteins and 5hmC appeared in evolutionary history.
About one-half of genes are methylated and the other half are unmethylated in C. intestinalis. We detected a clear enrichment of methylated genes in maternally and stably expressed genes (Figure 4B). Maternally expressed genes, comprising 35% of all genes, are largely housekeeping genes which encode proteins required for essential cellular functions . Those mRNAs and proteins are stored in eggs in sufficient quantities to afford rapid consumption during early development. Later in development these genes maintain basal transcription ubiquitously, in a fashion similar to that of stably expressed genes, which are also methylated. The data therefore support the conclusion that the majority of methylated gene bodies are associated with ubiquitously expressed genes.
The molecular mechanism by which cytosine methylation is added preferentially to ubiquitously expressed genes is unclear. One of the conceivable scenarios was that a cis element embedded in the ubiquitous promoter solely controls gene body methylation; however, our transgenic analysis did not support this idea. It is conceivable that a specific combination of cis and trans elements in a promoter and gene coding region may be required for gene body methylation.
On the basis of our observations, the DNA methylation pattern may be determined by a combination of multiple mechanisms. At first, it seems that DNA methylation is predominantly targeted to the body of ubiquitous genes and leaves the 5′ promoter, intergenic region and tissue-specific genes unmodified. Instead, in the transgenic line E, the Ci-Trl promoter sequence integrated into the intron of endogenously methylated gene was primarily protected from DNA methylation. In addition, methylated intergenic regions of the subset of genes in a convergent orientation (Figure 2A) also suggested that DNA methylation is not always restricted to the gene body. The methylated intergenic regions may lack a property required to actively create the unmethylated promoter domain. One possibility is that trimethylation of lysine 4 of histone 3 (H3K4) deposited at an active promoter prevents methylation, as a mutually exclusive distribution of DNA methylation and H3K4 methylation has been reported in genomewide studies in mammals  and this histone modification is unable to bind de novo DNMT s .
Herein we report that methylated and unmethylated gene groups are identical in C. intestinalis muscle and sperm cells, although their global levels of DNA methylation in the genome are different. The constant targeting of gene body methylation, regardless of cell type, indicates that DNA methylation is not a major regulator of tissue-specific gene expression. Instead, gene body methylation is linked to ubiquitously expressed genes, although this does not seem to be determined by their promoters alone. Overall, these studies indicate the presence of specific epigenetic states in ubiquitously expressed gene bodies.
Reference genome sequence and gene models
C. intestinalis sperm DNA extracted from North Pacific specimens were kindly provided by Shota Chiba and William Smith, University of California, Santa Barbara, USA. Adult specimens were obtained from the Maizuru Fisheries Research Station of Kyoto University and the Misaki Marine Biological Station of University of Tokyo through the National BioResource Project of the MEXT, Japan. Tissues, eggs and sperm were obtained surgically. After fertilization, embryos were raised in filtered seawater at 18°C. Adult specimen and sperm of C. intestinalis transgenic lines were provided by Ciona intestinalis Transgenic Line Resources (http://marinebio.nbrp.jp/ciona/).
Global DNA methylation analysis
Genomic DNA was prepared from embryos and adult tissues by the conventional phenol extraction method . The DNA concentration was measured by using the Qubit fluorometer (Invitrogen, Carlsbad, CA, USA). Global DNA methylation level was measured using Methylamp Global DNA Methylation Quantification Kit (Epigentek, Farmingdale, NY, USA) according to the manufacturer’s instructions. Methylated DNA was used as a positive control. The analysis was repeated in triplicate.
Sodium bisulfite conversion of genomic DNA was conducted using BisulFlash DNA Modification Kit (Epigentek). Bisulfite PCR primers were designed using the MethPrimer tool and database (http://www.urogene.org/methprimer/index.html) (Additional file 2: Table S4). PCR products were cloned using the StrataClone PCR Cloning Kit (Agilent Technologies, Santa Clara, CA, USA), and randomly selected multiple clones were subsequently subjected to Sanger sequencing, aligned and analyzed for their methylation status.
CXXC affinity purification and deep sequencing
Sperm DNA of North Pacific specimen was sonicated using a Bioruptor ultrasonicator device (Diagenode, Denville, NJ, USA) for 10 seconds on the high setting to produce fragmented DNA ranging from 200 to 800 bp with a peak of 500 bp. CAP-seq was performed as previously described with a minor change . An initial optimization of the salt-wash condition was conducted ranging from 300 mM to 800 mM and adopted 400 mM. DNA (35 μg) was bound to the CXXC matrix in a 100 mM NaCl-containing column buffer, washed and then eluted using buffer containing 1 M NaCl. Eluted fractions were pooled, concentrated and precipitated. An Illumina library of affinity-purified and input DNA were prepared as previously described . High-throughput sequencing was conducted using the Illumina Genome Analyzer IIx and HiSeq 2000 (Illumina, Inc, San Diego, CA, USA) with a read length of 50 bp.
Analysis of high-throughput sequence data
Single end reads obtained through high-throughput sequencing were quality-filtered and mapped to the genome assembly using Bowtie with the –best option (http://bowtie-bio.sourceforge.net/manual.shtml). Mapped affinity-purified and input sequences in the form of BedGraph Track Format files (UCSC Genome Bioinformatics; http://genome.ucsc.edu/goldenPath/help/bedgraph.html) were processed and analyzed using Perl scripts . Regions with sufficient read coverage were identified using H, L and G parameters, which are read height, length in base pairs and gap permitted in the length parameter, respectively. The parameters for CAP-seq were calibrated to identify hypomethylated domains in 40 kb of the cos41 genomic region previously investigated by BS-seq . The parameters, H 4 L 90 G 700, were validated in irrelevant 60-kb regions in total, where BS-seq has been conducted before. The same parameters were used for the analysis of the input reads. Methylated regions in sperm were determined as regions with the input coverage and deprived of the affinity-purified coverage.
The gene body methylation status of each gene was assessed by calculating a ratio of gene region overlapped with methylated domains using the intersectBed utility in BEDTools v2.11.2 .
We retrieved BS-seq data of adult muscle  from the Sequence Read Archive database (http://www.ncbi.nlm.nih.gov/sra). Short reads were mapped to the KH genome assembly using Bismark v0.5.2  with a default setting (http://www.bioinformatics.babraham.ac.uk/projects/bismark/). We discarded reads mapped to multiple genomic positions. Next, the coverage (X) and the number of converted short reads (m) for each cytosine residue was extracted from the mapping results by running the Bismark methylation_extractor. In the following step, we assessed the methylation status for every cytosine residue with X ≥ 2. We estimated an error rate that included the error of bisulfite conversion, sequencing and mapping. The error rate was calculated from the mapping result of a mitochondrial genome, which has no or very low DNA methylation . The error rate was estimated to be 0.00091. We tested whether m is explained only by the estimated error rate, given X by binomial testing for each cytosine residue, by following the method described by Lister et al. . Methylated cytosine residues were defined as those with calculated P-values <0.01.
Microarray data analysis
Microarray data were derived from the study by Azumi et al. . The corresponding KH gene model to the array probe sequence was searched by local BLAST with an E-value cutoff of 1E-20. Duplicated gene models were removed. Gene models which hit multiple probes belonging to different expression groups were retained.
Transgene insertion sites
Transgene insertion sites in lines E and F were identified as described by Sasakura et al. .
The CAP-seq data have been deposited in the National Center for Biotechnology Information Sequence Read Archive under the accession number DRA000388-390.
We thank Dr Shota Chiba and Dr. William Smith (University of California, Santa Barbara) and the National BioResource Project (NBRP) of the MEXT, Japan, for providing animals. We also thank Dr Rob Illingworth and Dina De Sousa for help with CAP-seq and Dr Hiroki Takahashi and Dr Shuichi Wada for technical advice. We extend thanks to Dr Naoto Ueno for comments on the manuscript. We thank Kazuko Hirayama for her great support. This work was supported by funding from the Japanese Science and Technology Agency, Precursory Research for Embryonic Science and Technology (PRESTO) program (to MMS), and grant 074948 (to AB) from the Wellcome Trust. This study was carried out partly under the National Institute for Basic Biology cooperative research program (11–335).
- 14.Feng S, Cokus SJ, Zhang X, Chen PY, Bostick M, Goll MG, Hetzel J, Jain J, Strauss SH, Halpern ME, Ukomadu C, Sadler KC, Pradhan S, Pellegrini M, Jacobsen SE: Conservation and divergence of methylation patterning in plants and animals. Proc Natl Acad Sci USA. 2010, 107: 8689-8694. 10.1073/pnas.1002720107.PubMedCentralCrossRefPubMedGoogle Scholar
- 17.Xiang H, Zhu J, Chen Q, Dai F, Li X, Li M, Zhang H, Zhang G, Li D, Dong Y, Zhao L, Lin Y, Cheng D, Yu J, Sun J, Zhou X, Ma K, He Y, Zhao Y, Guo S, Ye M, Guo G, Li Y, Li R, Zhang X, Ma L, Kristiansen K, Guo Q, Jiang J, Beck S: Single base-resolution methylome of the silkworm reveals a sparse epigenomic map. Nat Biotechnol. 2010, 28: 516-520. 10.1038/nbt.1626.CrossRefPubMedGoogle Scholar
- 18.Nanty L, Carbajosa G, Heap GA, Ratnieks F, van Heel DA, Down TA, Rakyan VK: Comparative methylomics reveals gene-body H3K36me3 in Drosophila predicts DNA methylation and CpG landscapes in other invertebrates. Genome Res. 2011, 21: 1841-1850. 10.1101/gr.121640.111.PubMedCentralCrossRefPubMedGoogle Scholar
- 26.Illingworth RS, Gruenewald-Schneider U, Webb S, Kerr ARW, James KD, Turner DJ, Smith C, Harrison DJ, Andrews R, Bird AP: Orphan CpG islands identify numerous conserved promoters in the mammalian genome. PLoS Genet. 2010, 6: e1001134-10.1371/journal.pgen.1001134.PubMedCentralCrossRefPubMedGoogle Scholar
- 28.Azumi K, Sabau SV, Fujie M, Usami T, Koyanagi R, Kawashima T, Fujiwara S, Ogasawara M, Satake M, Nonaka M, Wang HG, Satou Y, Satoh N: Gene expression profile during the life cycle of the urochordate Ciona intestinalis. Dev Biol. 2007, 308: 572-582. 10.1016/j.ydbio.2007.05.022.CrossRefPubMedGoogle Scholar
- 42.Satou Y, Mineta K, Ogasawara M, Sasakura Y, Shoguchi E, Ueno K, Yamada L, Matsumoto J, Wasserscheid J, Dewar K, Wiley GB, Macmil SL, Roe BA, Zeller RW, Hastings KEM, Lemaire P, Lindquist E, Endo T, Hotta K, Inaba K: Improved genome assembly and evidence-based global gene model set for the chordate Ciona intestinalis: new insight into intron and operon populations. Genome Biol. 2008, 9: R152-10.1186/gb-2008-9-10-r152.PubMedCentralCrossRefPubMedGoogle Scholar
- 43.Sambrook J, Fritsch EF, Maniatis T: Molecular Cloning: A Laboratory Manual: Volume 2. 1989, Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press, 18.47-18.59. 2Google Scholar
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.