Development of enhancer-trapping and -detection vectors mediated by the Tol2 transposon in zebrafish

Introduction

Enhancers are among the most important cis-regulatory elements that play a major role in cell-type-specific gene expression (Guenther et al., 2007; Heintzman et al., 2009), which widely reflects developmental patterning (Sagai et al., 2005) or human genetic disease (Grosveld et al., 1987). Enhancers recruit transcription factors and the transcriptional apparatus to activate their target genes (Smith & Shilatifard, 2014; Ren & Yue, 2015) and can be larger than one megabase (Mb) (Fraser & Engel, 2006). To date, many methods have been developed to discover enhancers, such as enhancer trapping (ET) (Kawakami et al., 2004; Balciunas et al., 2006; Scott et al., 2007; Asakawa et al., 2008), retrovirus-based technologies (Ellingsen et al., 2005; Laplante et al., 2006) and the chromatin immuno precipitation sequencing technique (Mcaninch & Thomas, 2014; Bu et al., 2017). Generally, a basic ET construct is composed of a reporter gene under the control of a minimal promoter (Bellen et al., 1989). Once integrated into the genome, enhancers in the chromosome cannot drive the expression of the reporter gene of the trapping construct unless they induce the activity of a minimal promoter (Bellen, 1999). The rapid development of ET technology has helped identify a large number of enhancers (Balciunas et al., 2004; Liu et al., 2015; Sang et al., 2018). A total of 11 enhancer-detection lines were characterized by Minos transposon-mediated enhancer detection screening in Ciona (Yoshida & Sasakura, 2012). Transgenic mice with lentiviral vectors carrying single-copy enhancer-detector probes encoding either the marker gene lacZ or Cre recombinase were generated and used as an enhancer-detection strategy in mammals (Kelsch, Stolfi & Lois, 2012). Among these methods, transposon-mediated ET represents the most effective insertion in vertebrate systems and has been developed as a gene-delivery tool for gene therapy and insertional mutagenesis (Kebriaei et al., 2017). For instance, it has been applied successfully in medaka (Grabher et al., 2003), zebrafish (Balciunas et al., 2004; Scott et al., 2007; Liu et al., 2015), mouse (Choi et al., 2018) and insect (Koelzer, KöLsch & Panfilio, 2014) models. Transposon-driven vectors generally harbor a reporter-protein-encoding region downstream of a minimal promoter, which is flanked by transposon inverted terminal repeats (Bier et al., 1989; Wilson et al., 1989; Dunne et al., 2014). It seems that the choices of minimal promoter and transposon are equally important for ET efficiency. The Sleeping Beauty, PiggyBac and Tol2 transposons are widely used in genetic research using animal models (Kawakami et al., 2004; Balciunas et al., 2006; Scott et al., 2007; Asakawa et al., 2008). Among them, Tol2 has the highest transgenic efficiency in zebrafish (Shen et al., 2018); thus, we chose Tol2 as the optimized candidate for ET. However, the type of minimal promoter that should be used in ET vectors to achieve a higher efficiency remains an urgent issue. The Krt4 and Gata minimal promoters from zebrafish have been applied in ET technology (Bessa et al., 2009; Chatterjee et al., 2010; Ogura et al., 2009; Trinh & Fraser, 2013). These two minimal promoters can drive green fluorescent protein (GFP) expression in transgenic zebrafish, which is fit for ET technology. We were eager to apply other minimal promoters in zebrafish; therefore, we selected an additional two minimal promoters, Myc and Oct4, which are derived from the mouse and are rarely used in zebrafish. To develop an efficient ET vector, we compared the basic expression (mCherry expression driven by the minimal promoter itself in the absence of an enhancer) and ET efficiency of four minimal promoters (Myc, Oct4, Krt4 and Gata) in zebrafish, which is an important model organism for the efficient detection of enhancer activity in vivo (Haffter et al., 1996) and a superior model compared with mouse models in large-scale screens. An insulator is a type of DNA element that can protect genes from nearby enhancers or silencers. According to previous studies, a 250 bp “core” containing the 5′HS4 element was effective in blocking enhancer activity (Recillas-Targa et al., 2002), and two copies of the core element functioned as a strong insulator when placed between the enhancer and the promoter (Eissenberg & Elgin, 1991; Chung, Bell & Felsenfeld, 1997). To detect the activity of enhancers efficiently and avoid the effect of the host genomic regulators surrounding them, 5′HS4 insulators were used to flank the mCherry reporter gene expression box, which is driven by an enhancer and a minimal promoter. We also evaluated the compatibility between minimal promoters and enhancers. To verify enhancer activity, here, we report an approach that combined ET, insulators and transposons. We demonstrated the feasibility of using this approach to trap active enhancers and attempted to detect their activity.

Materials and Methods

Results

Optimizing the minimal promoter of ET vectors

To determine the optimal minimal promoter for the ET vector, four Tol2-mediated ET vectors were constructed using the Myc, Oct4, Krt4 and Gata minimal promoters to drive the mCherry ORF (Fig. 1A), respectively. These vectors were injected with or without Tol2 transposase mRNA into zebrafish embryos, to check ET efficiency. The rate of positive embryos (carrying the mCherry signal) at 12 or 72 hpf was the highest in the Krt4 group, followed by the Gata, Myc and Oct4 groups (Figs. 1B and 1C). However, the basic expression level was also the highest in the Krt4 group compared with the remaining three groups, which suggests that the Krt4 minimal promoter itself drives mCherry, resulting in a high background expression that disturbs ET (Trinh & Fraser, 2013). The rate of mCherry-positive embryos at 12 or 72 hpf was second highest in the Gata group, while the basic expression noise was negligible in this group, which suggests that the Gata minimal promoter can be used in ET. Furthermore, our results showed reoccurring patterns in the cerebellum of embryos injected with the Krt4 group (Fig. 1D), which indicated that Krt4 may remain active as a strong promoter (Bellen, 1999), as opposed to what was observed for the Gata group (Fig. 1E). In the Myc or Oct4 group, the rate of positive embryos was very low compared with the Krt4 and Gata groups.

Minimizing position effects in pEDVs using insulators

To verify the newly identified enhancer, we constructed detection vectors by introducing two 5′HS4 insulators based on the ET vectors pEDV-Gata and pEDV-Krt4 (Fig. 2A). The results showed that the mCherry expression rate (∼5%, n = 301) in embryos injected with pEDV-Gata was decreased significantly compared with that detected in embryos injected with pTol2-Gata-mCherry (23%, n = 308) (Fig. 2B). In addition, embryos injected with pEDV-Gata did not show any expression patterns compared with the multiple expression patterns detected in embryos injected with pTol2-Gata-mCherry (Figs. 1E, 2E and 2F). This indicates that the 5′HS4 insulators blocked the genome position effects and can be used for the identification of enhancer activity (Figs. 2B and 2F). However, the rate of mCherry positivity among embryos injected with pEDV-Krt4 was over 55% (n = 301), which was slightly lower than that observed for the embryos injected with pTol2-Krt4-mCherry (67%, n = 302) (Fig. 2B). The expression of mCherry was confirmed in the heart, eye and telencephalon tissues of embryos injected with pEDV-Krt4 (Fig. 2D). Taken together, these results suggest that Gata is superior to Krt4 for ET or enhancer detection.

Compatibility of minimal promoters and enhancers

To determine whether Gata is compatible with different enhancers, enhancers identified previously (Charité et al., 2001; De La Calle-Mustienes et al., 2005; Bessa et al., 2009; Iklé, Artinger & Clouthier, 2012; MacDonald et al., 2013), CNS1, Z48, Hand2 and Hs769, were subcloned upstream of Gata, and the resulting enhancer-detection vectors were named pEDV-Gata-CNS1, pEDV-Gata-Z48, pEDV-Gata-Hand2 and pEDV-Gata-Hs769, respectively (Fig. 3A). Using a bioinformatics approach, we found that CNS1 was conserved in vertebrates (Fig. 4A), Z48 was conserved in fishes (Fig. 4B), Hand2 was only identified in zebrafish (Fig. 4C) and Hs769 was conserved in mammals and chicken (Fig. 4D). Moreover, the functions of these enhancers, with the exception of Hs769, were verified previously (Bessa et al., 2009; Iklé, Artinger & Clouthier, 2012; MacDonald et al., 2013). Subsequently, each donor plasmid was co-injected with the Tol2 mRNA into one-cell-stage embryos. We found that embryos that were injected with both pEDV-Gata-Z48 and pEDV-Gata-CNS1 generated distinct mCherry expression profiles at 72 hpf, with an mCherry positivity rate in embryos of about 45% (n = 308). In contrast, embryos that were injected with pEDV-Gata-Hand2 and pEDV-Gata-Hs769 did not express the mCherry gene (Fig. 3B), which suggests that neither Hand2 nor Hs769 can activate the Gata minimal promoter. However, the enhancer activity of Hand2 has been confirmed (Iklé, Artinger & Clouthier, 2012). To verify this, the Gata minimal promoter was replaced with the Beta-globin minimal promoter, and the mCherry gene was replaced by the gene encoding the GFP. We found that Z48, Hand2 and CNS1 regulated β-globin to drive GFP gene expression (Fig. 3C), and that the expression patterns were similar to those reported in previous studies (Figs. 3F–3H). However, Hs769 did not regulate GFP expression (Fig. 3C). These results suggest the presence of a compatibility problem between enhancers and minimal promoters; thus, additional minimal promoters should be considered for enhancer-activity testing.

Discussion

Enhancer trapping is an effective technique that can be used to characterize enhancers that regulate the spatiotemporal expression patterns of genes in cells. ET mediated by transposons, which was established in the fruit fly Drosophila melanogaster, is an efficient technique for enhancer identification (Liu et al., 2015). ET vectors have two key features: (1) the vector can be inserted into the host genome efficiently and (2) the minimal promoter can maintain a low basic background/noise ratio (Parinov et al., 2004). To construct an efficient ET vector, we used the Tol2 transposon to mediate the minimal promoter expression cassette. Our results suggested that the Tol2 transposon significantly increased the integration rate of vectors. To obtain an ideal minimal promoter with a low background expression, we compared four minimal promoters, namely Myc, Oct4, Krt4 and Gata. We found that the expression noise of Gata was almost negligible, and that its ET efficiency was significantly higher than that of Myc and Oct4. The mCherry positivity rate and the expression background were the highest for Krt4 compared with the remaining promoters, even in the absence of the Tol2 mRNA. This indicated that Krt4 itself can drive gene expression without the activation of enhancers, which may disturb the identification of enhancers in the genome (Bellen, 1999). Gata (isolated from zebrafish) was more sensitive to nearby enhancers in the zebrafish genome than were Myc and Oct4 (isolated from mice). Thus, we assumed that the zebrafish Gata minimal promoter was ideal for ET in zebrafish. Therefore, we used the Gata minimal promoter to assemble the enhancer-detection vector. Because the minimal promoter can be affected by enhancers or silencers in genomes (Chung, Whiteley & Felsenfeld, 1993), we used the 5′HS4 insulator (from chicken) in the detection vector, to block the activity of nearby cis-regulatory elements. Our results showed that 5′HS4 shielded the minimal promoter from genome position effects. Moreover, the 5′HS4 insulator worked well in zebrafish, even though it was isolated from the chicken genome and is not conserved in zebrafish. Our results supported the hypothesis that the 5′HS4 insulator may play a crucial role in minimizing position effects in different species. This was also demonstrated by Bessa et al. (2009). In this study, we found that Krt4 drove the expression of a reporter gene by itself, even when shielded with insulators, indicating that the Krt4 minimal promoter might be a strong promoter that is not suitable for enhancer detection.

Here, we constructed two zebrafish enhancer-detection vectors containing 5′cHS4 insulators, the Gata or Beta-globin minimal promoter and the GFP or mCherry reporter gene. To test the capacity of the promoter to respond to enhancers, four enhancers (CNS1, Z48, Hand2 and Hs769) were cloned upstream of Gata or Beta-globin. We found that Gata was sensitive to CNS1 and Z48, but not to Hand2 and Hs769, while Beta-globin was sensitive to CNS1, Z48 and Hand2, but not to Hs769. CNS1 was conserved in vertebrates, including fishes, amphibians, birds and mammals, and the similarity of the sequence was verified by others in previous studies (MacDonald et al., 2010, 2013). CNS1 was expressed in the telencephalon (Figs. 3E and 3H), which agreed with our previous studies (Shen et al., 2018). Z48 and Hand2 were cloned from zebrafish, Z48 was conserved in fish and Hand2 was only identified in zebrafish. Z48 drove a strong expression in the midbrain (De La Calle-Mustienes et al., 2005; Bessa et al., 2009) and Hand2 promoted expression in the pharyngeal arch in zebrafish (Charité et al., 2001; Iklé, Artinger & Clouthier, 2012). Similar expression patterns were detected here (Figs. 3D, 3F and 3G). Interestingly, Hs769 promoted expression in the neural tube in mice and was conserved in mammals and chicken. However, Hs769 had no effect on the Gata or Beta-globin minimal promoter. These results suggest that the sensitivity of the different promoters to enhancers varied. They also indicate the existence of a compatibility problem between enhancers and minimal promoters and highlight the possibility that the interactions between enhancers and promoters are somewhat specific (Kelsch, Stolfi & Lois, 2012; Rickels & Shilatifard, 2018). A previous study showed that a lentiviral enhancer probe containing a different minimal promoter (hsp68 and thylmp) produced recombination patterns as a result of the random integration of the probe into different genomic loci; those results suggested the presence of interactions between the promoter in the probe and the enhancer in the genomic DNA. Obviously, it is important to obtain appropriate minimal promoters that can reflect enhancer activity.

In summary, we attempted to provide a useful method for the rapid, large-scale genomic screening of enhancers and for the efficient detection of enhancer activity in zebrafish, which will accelerate our understanding of functional genomics.

Conclusions

We constructed a valuable enhancer-trapping vector, pTol2-Gata-mCherry, to efficiently explore additional enhancers in zebrafish. Moreover, we developed enhancer-detection vector successfully to identify enhancer activity. Also, we found the compatibility of minimal promoters and enhancers. These data here represent a systematic approach to the discovery and verification of enhancers.

Supplemental Information