The identification of the enhancers associated with each developmentally regulated gene is a first step to clarify the regulatory mechanisms underlying embryogenesis. The electroporation technique using chicken embryo is a powerful tool to identify such enhancers. The technique enables us to survey a large genomic region and to analyze the enhancers in great detail. Comparison of the genomic sequences of the chicken and other vertebrate species identifies conserved non-coding sequence blocks to which the functionally identified enhancers often correspond. In this review, I describe in detail the methods to analyze the enhancers using the chicken embryo electroporation and genome comparison.
Precise regulation of gene expression is essential for embryonic development. The regulatory regions assorted in individual genes in a genome are usually identified as enhancers. Identification of the enhancers is thus a first step to decipher regulation of the genes realized through the network of transcription factors and signaling cascades. Methods taking advantage of unique features of chicken embryos have been developed to identify and characterize the enhancers of genes of developmental interest (Uchikawa et al. 2003; Uchikawa et al. 2004; Matsumata et al. 2005).
The advantages of chicken embryo for developmental studies are derived from their ease of direct observation and tissue manipulation in ovo, and have been used in many important studies (Stern 2004). Until a decade ago, the gene manipulation techniques for the chicken embryo lagged behind the mouse system where transgenic and gene-targeting mice were in routine use. In the analysis of gene regulation using chicken cells, transfection of the tissue-cultured cells was the only method. However, there was a revolutionary change of the situation following the successful application of the electroporation technique to developing chicken embryos (Muramatsu et al. 1996; Funahashi et al. 1999). It became possible to introduce DNA into a chicken embryo with a high efficiency and even to manipulate activities of the genes in chicken embryos by this technique. This opened up a new avenue to the study of gene regulation in developing embryos. Even powered by a combination with the embryo electroporation technique is the tissue manipulation that is uniquely possible using chicken embryos (Uchikawa et al. 2003; Takemoto et al. 2006).
Genomic comparison of the chicken and other vertebrate species is often used to predict potential regulatory sequences (Uchikawa et al. 2004; Izumi et al. 2007). Advances of genome projects now enable comparison of the entire genomic sequences among different vertebrate species, and the regulatory regions of genes are often recognized as non-coding DNA sequence blocks that are highly conserved across the vertebrate species (Hardison 2000; Woolfe et al. 2005). Projects to screen the enhancers in human/fish genomes are in progress using such a maneuver (Visel et al. 2007; Woolfe et al. 2007). Chicken, as well as frog (also discussed in this issue), is at a phylogenetic distance from mammals and fishes just appropriate for the determination of conserved sequence blocks of functional relevance (Khokha & Loots 2005; Ogino et al. 2008). Belonging to the same amniotes, chicken is expected to share many features of developmental regulation with mammals. Analysis of mechanisms of gene regulation in the chicken should therefore indicate universal mechanisms conserved among vertebrates, and those unique to amniotes.
In this review, the methods to analyze the regulatory regions based on chicken embryo electroporation and genome comparison are described in detail.
Strategy to survey regulatory regions of a gene using embryo electroporation
An essential first step to analyzing gene regulation is to characterize the stages and tissues of an embryo where the gene of interest is expressed. Then analysis of the regulation of the gene is carried out using chicken embryo electroporation selecting appropriate developmental stages. The early disc-shaped embryo may be a first choice for the application of electroporation, where exogenous DNAs are efficiently introduced to a large area of the embryo (Fig. 1) (the methods are detailed below).
Where should we look for the regulatory regions in a wide genomic region including the gene of interest? Regulatory regions are often assumed to be located in the 5′ flanking region of a gene, but in fact many regulatory regions are dispersed far away from the transcriptional start sites including the 3′ regions (Kleinjan & van Heyningen 2005). Although a hint for the localization of regulatory regions is given by genome comparison among chicken and other species (see below), a wide surrounding region of the gene should be surveyed without bias. For example, to study the regulation of the Sox2 gene, we examined a 50-kb region encompassing the gene in the first trial and identified a number of enhancers that regulate Sox2 expression in various stages of the neural development (Fig. 2) (Uchikawa et al. 2003). We also identified additional enhancers of the Sox2 gene outside of the 50-kb region (R. Okamoto et al. unpubl. data, 2008). Thus, some enhancers may regulate a gene from a large distance.
How is the survey of a genomic sequence carried out for an enhancer activity? The length of plasmids affects the efficiency by electroporation, with the upper limit of circular plasmid length being 10 kb. For this reason, 5-kb long genomic fragments are used for insertion into the plasmid backbone including the reporter gene to be tested for enhancer activity. Genomic sequences to be tested for enhancer activities may be extracted from either genomic DNA or phage/bacterial artificial chromosome (BAC) clones (Fig. 1). An important point is to minimize introduction of mutations in the DNA sequences to be tested. The BAC clones are indeed convenient to prepare a series of DNA fragments scanning a genome region, as efficient polymerase chain reaction (PCR) amplification of sequences up to 5 kb with high fidelity DNA polymerase is possible. Of course, the fragments using digestions of restriction enzymes or sheared with mechanical force are also suitable to be subcloned without mutations, but it may be more laborious (Matsumata et al. 2005). It is recommended to prepare a set of overlapping DNA fragments of approximately 5 kb to cover the entire genomic region of interest.
The next step is chicken embryo electroporation. It is important to use the type of electrode suitable for electroporating the tissues of embryos of specific stages (Fig. 3). Parameters of electric pulses, in particular voltages between the electrodes, must be optimized through a series of pilot experiments, as well as the concentration of DNA to be delivered between the cathode and the tissue. The optimum concentration of DNA varies depending on the reporter vectors, as the reporter-encoded proteins are detected with different sensitivities. If one wishes to label and confirm the successfully electroporated domains of embryos, non-specific LacZ or mRFP1 (Campbell et al. 2002) vectors may be co-electroporated with enhanced green fluorescent protein (EGFP) vectors to test the enhancer activity. Efforts to minimize DNA dilution (e.g. instantaneous application of the electric pulse) after delivering the DNA solution to the site of electroporation are also recommended.
Fluorescence of reporter proteins are detected 2–3 h after the electroporation in cases of vector gene activation by a strong enhancer, and enhancer activity may be followed at least for 2 days if the activity of the enhancer persists. To detect activities of enhancers, the choice of time and tissues in developing embryos is an important factor, as the vector DNAs in a closed circular form are not integrated into the genome after electroporation and will be diluted with development of embryos. The electroporation of early disk-shaped embryos in New's culture is a way to introduce the vectors in a broad range of embryonic tissues, and to observe the whole embryo during the neurulation period. Under these conditions, tissue manipulation can be combined with electroporation, and the time-dependent change of reporter expression can be easily monitored (Fig. 4) (Uchikawa et al. 2003).
Use of genome comparison
The genome sequences of many animal species have been determined and deposited to databases, such as Ensembl (http://www.ensembl.org/), and the data are accessible through websites. The genome comparison among different vertebrate species indicates the presence of abundant conserved non-coding sequence blocks, where the extent of conservation depends on the phylogenetic distance between the species (Bejerano et al. 2004; Woolfe et al. 2005). According to our experience, the conserved sequence blocks identified by the comparison of chicken and mammalian genomes include the enhancers. Thus, comparison of the chicken and mammalian genomes helps prediction of the potential regulatory regions (Uchikawa et al. 2003; Uchikawa et al. 2004; Izumi et al. 2007).
How widely the regulatory regions are distributed around a gene and how much they correspond to the conserved sequence blocks is not all predictable and variable depending on the gene. In the case of the Sox2 gene we studied, the majority of conserved sequence blocks were located within the 30-kb span centered by the gene, and a significant fraction of the conserved sequence blocks corresponded to the enhancers. Thus, the regulatory regions would be first searched in the surrounding gene and gradually broadened to the outside. Once a particular DNA fragment is found to have an enhancer activity, it is helpful to examine the inclusion of conserved sequence blocks; such blocks are good candidates of the regulatory regions responsible for the enhancer activity.
To what extent are the enhancers conserved among species? The parameters to compare the genomic sequences is empirically determined and varied depending on the cases (Prabhakar et al. 2006; Stephen et al. 2007). For genomic sequence comparison between fish and mammals, 60% sequence identity over the length of 40 bp is scored as significant (Woolfe et al. 2005), and for comparison between humans and rodents, 100% sequence identity over the length of 200 bp is scored (Bejerano et al. 2004), while the most appropriate parameters may vary depending on the genetic locus. Our experience with the Sox2 genomic region indicates that scoring with the filter condition of 60% sequence identity over 100 bp long is appropriate for comparing chicken and mammals or frog genomes, and the condition of 60% sequence identity over 50 bp long is appropriate for comparison between chicken and fish genomes (Uchikawa et al. 2004).
The comparative analysis of genomes of the chicken and a wide range of vertebrate species (mammal, frog and fish, as well as marsupial) provides a global view about how conserved sequences evolved (Fig. 2). In the Sox2 locus, while the conserved sequence blocks of functional relevance were identified by comparison between chicken and human/mouse genomes, these sequence blocks are buried within longer stretches of sequence similarity when human and mouse genomes were compared. These sequence blocks are also identified by comparison of chicken and opossum/xenopus genomes. Comparison of chicken, opossum, mouse and human genomic sequences indicate that mouse sequences are uniquely diversified in the possession of the generally conserved sequence blocks. Conserved sequence blocks nos. 3, 21 and 25 are lacking in the mouse genome, but are found in another three. The xenopus genome also keeps the majority of the conserved sequence blocks. In particular, enhancer L active in lens fibers is uniquely conserved between chicken and xenopus genomes, which is consistent with the occurrence of Sox2 expression in lens fibers in chicken and xenopus eyes, and its absence in the mouse lens fibers (Kamachi et al. 1998; Schlosser & Ahrens 2004). It is to be noted that the conserved sequence block no. 11 is active as an enhancer in ES cells (Tomioka et al. 2002) and is conserved among amniotes. In fish genomes, conservation of the sequence blocks is more limited and appears to depend on the fish species. The sequence blocks are more clearly conserved in zebrafish than in tetraodon. In addition, enhancer N–1 is not conserved in fish genomes, while other early neural enhancers (N-2, N-3, N-4 and N-5) are conserved throughout the vertebrate species. In contrast to pan-neural expression of Sox2 in higher vertebrates, Sox2 expression is restricted to the anterior neural plate in fish (Uchikawa et al. 2004; Okuda et al. 2006). These results indicate that conserved sequence blocks reflect the regulation of the gene in individual species.
Construction of vectors
To construct the reporter vectors to be used for enhancer analysis, combination of the promoter sequence and the protein-coding reporter gene must be carefully determined. Promoter to initiate the transcription of the reporter coding sequence should faithfully reflect the activity of the enhancer. To fulfill this condition, the promoter must be non-specific and not too strong. We find that the thymidine kinase (tk) promoter of Herpes simplex virus is just appropriate. In electroporated chicken embryos, the tk promoter has no power to express the reporter gene (EGFP, LacZ) to a detectable level unless an enhancer sequence is inserted. The tk promoter is thus used to construct the ptkEGFP and analogous vectors to assess enhancer activities of genomic fragments. An alternative choice may be the original promoter of the gene that the enhancers activate in the genome, unless the activity of the promoter is not very strong.
The fluorescent proteins are excellent to report the enhancer activity in live embryos. EGFP is one such example having sufficient brightness and relatively fast maturation time. The Venus, a yellow fluorescent protein (YFP) variant, is another example (Nagai et al. 2002). Red-colored fluorescent proteins are also useful. For example, mCherry, an mRFP1 variant, possesses fastest maturation among the fluorescent proteins (Shaner et al. 2004). When combinations of multiple reporters with different fluorescent colors are used in an experiment, the combination of (EGFP and mCherry) or (cyan fluorescent protein [CFP], Venus and mCherry) is recommended. Enzyme proteins (e.g. beta-galactosidase and luciferase) are also used for reporters to quantitate enhancer activities. In addition, the real-time bioluminescence imaging system using the luciferase provides a powerful means to monitor dynamic changes of reporter expression in living embryos (Masamizu et al. 2006).
Electroporation of chicken embryo
For the success to detect enhancer activity associated with a DNA fragment, an efficient introduction of exogenous DNAs into chicken tissues is an essential factor. In our laboratory, four methods of electroporation are used to detect enhancers in different stages of chicken embryo (Fig. 3). The electroporation of the early flat embryos at stage 4 is a suitable way of introducing the DNA into the large area of the embryos that gives rise to ectodermal and mesodermal cells during neurulation stages. Embryos at stage 7, still flat, are used to electroporate DNAs in a more restricted area of embryos, for example, to examine the enhancer activities in cranial placodes. In later embryos, the neural tube is electroporated with a high efficiency (Itasaki et al. 1999; Timmer et al. 2001). Head ectoderm of stage 10 embryos is useful to examine the enhancer activities of lens cells. Thus, the appropriate stages and tissues of embryos must be chosen for electroporation, depending on the enhancers to be tested.
Detailed analysis of enhancers
Once a genomic fragment is shown to have an enhancer activity, the genetic elements included in the fragment that are involved in the activity and specificity of the enhancer can be determined by fine analysis of the fragment (Fig. 5). In many cases, an enhancer consists of several elements, with an essential element, usually called the ‘core’ element, having the specificity similar to the full length-fragment, and other non-core elements contributing to the strength of enhancer activity (Goto et al. 1990; Takemoto et al. 2006; Inoue et al. 2007). The core element as a monomeric sequence usually does not exhibit enhancer activity, but multimerization of the element generates enhancer activity with similar specificity as the original enhancer. The identification of the essential element is helpful to predict the transcriptional factor binding sites and to facilitate understanding of the gene regulation (Kamachi et al. 2001; Takemoto et al. 2006).
Introduction of a series of mutations and deletions to the enhancer sequence and examination of their effects allows determination of the regulatory elements (Fig. 5A). Co-electroporation of two vectors expressing different fluorescent proteins that depend on mutated and unmutated (wild type) enhancers is useful in determination of the mutational effect (Fig. 5B). Minimal DNA concentrations of 2 mg/mL for ptkEGFP/Venus vectors, and 1 mg/mL for ptkmRFP1/mCherry vectors are used for this kind of analysis (Inoue et al. 2007). The computer-aided search for potential binding sites of transcriptional factors (Vavouri & Elgar 2005) is often helpful when a short regulatory element (< 100 bp) is analyzed. These analyses will eventually determine genetic elements involved in regulation of a gene that correspond to nuclear factor binding sites.
Challenge to technical hurdles
The stages and tissues of an embryo where electroporation techniques are applied should be preferably extended. An example is in embryos earlier than stage 4 if we wish to examine enhancer activities during dynamic movement of gastrulation. Although the earliest stage embryo soon after a freshly laid egg could be successfully electroporated to label the cells with strong expression vector of, for example, CAGGS-EGFP or CMV-EGFP (Cui et al. 2006), the consistency and efficiency of introduction of exogenous DNAs by electroporation at these stages are not satisfactory for the analysis of known and accessible enhancers. The use of transfection reagents may facilitate electroporation of the reporter vectors in early embryos (Albazerchi et al. 2007).
Use of the transposon system causing integration of exogenous gene into the genome of somatic cells may allow long-term observation of enhancer activities in developing embryos as the reporter vectors are integrated into the genome (Sato et al. 2007).
Chicken genome is useful not only to predict the potential regulatory regions by genome comparison but also for its compactness compared with human/mouse genomes. It is desirable that successful introduction of the large size vector constructs, such as BAC vectors, into the embryos will facilitate surveying the enhancers. As electroporation of a large-size DNA (~150 kb) into the cultured cells has been reported (Knutson & Yee 1987), the improvement of an embryo electroporation technique may overcome the vector size problem.
This work was supported by Grants-in-Aid for Scientific Research to M.U. (18770201), to H. Kondoh (17107005), and to Y. Kamachi (18017019 and 18570197) from the Ministry of Education, Culture, Sports, Science and Technology of Japan. The author thanks members of the Kondoh laboratory for stimulating discussions and Roger Y Tsien for kindly providing mRFP1 plasmid.