pTransgenesis: a cross-species, modular transgenesis resource

The ability to engineer genetically modified organisms is essential for establishing the function of genes during development, disease, homeostasis, repair and regeneration (Gama Sosa et al., 2010; Ristevski, 2005). However, a crucial step in engineering genetically modified organisms is the design and generation of the transgene DNA constructions required for a given experiment. For the past thirty years, DNA constructions have been created primarily through restriction enzyme digestion and ligation. However, cloning with restriction enzymes becomes progressively more cumbersome as the complexity of the engineered constructs increases. For this reason, a site-specific recombination-based DNA cloning method was developed that circumvents the use of restriction enzymes (Hartley et al., 2000). The advent of recombination-based cloning brought a series of diverse and pioneering studies showing the utility of this technology in creating DNA constructions for transgenesis (Fisher et al., 2006; Hope et al., 2004; Ikeya et al., 2005; Kappas et al., 2008; Kwan et al., 2007; Nyabi et al., 2009; Semple et al., 2010; Skarnes et al., 2011). However, none had yet been designed specifically for use in Xenopus, a widely used model organism (Amaya, 2005), and the ability to use them across multiple models was limited. When we began to develop a transgenesis plasmid resource for Xenopus, we sought to design a system that would encapsulate multiple advances demonstrated in previous Multisite Gateway-based cloning projects, but we wished to expand on them to make them more universally useful to the developmental biology community at large. In particular, we wished to decouple the screenable elements from the transgenesis-promoting sequences, thus facilitating the transfer of this plasmid resource across different model systems, such as Xenopus, mammals, fish and flies.

By combining these attributes, we created a new modular, cross-species plasmid resource, which we have named pTransgenesis. The pTransgenesis resource is the first modular cloning system that allows the interchange of DNA elements for transgenesis between Xenopus, zebrafish, Drosophila and mammalian cell culture models. The pTransgenesis design and associated resources will greatly facilitate the efficient generation of transgenic organisms and the transfer of transgenic reagents across various developmental model organisms.

We adapted Invitrogen’s Gateway Multisite Cloning Kit (CA, USA) to create the pTransgenesis vectors (note, this is not the Gateway ‘Pro’). BP reactions were performed using Invitrogen’s BP Recombinase and PCR products were cloned with Invitrogen’s pCR8 GW TOPO kit. LR recombinations were performed using Invitrogen’s LR Clonase II+ (Ishibashi et al., 2012), bacterial transformations with DH5a competent cells (Invitrogen); typically 50-100% of colonies yield correct recombinations.

Restriction enzyme-mediated integration (REMI) transgenesis was performed as described (Breckenridge et al., 2001; Kroll and Amaya, 1996). I-SceI transgenesis was performed by injecting 2 nl of a 10 pg/nl reaction mixture as described previously (Ishibashi et al., 2012). The Cre mRNA and Tol2 mRNA was made using the SP6 mMessage Machine mRNA Kit (Ambion) from a pCS2-CreNLS or pC2-TP template (Kawakami, 2007). For zebrafish injections, ∼1 nl of a 20 pg/nl DNA solution with or without 25 pg Tol2 mRNA was injected at the 1-cell stage. HeLa cell lines were made by plasmid transfection (Lipofectamine 2000, Invitrogen), with a selection of 3 μg/ml of puromycin in DMEM performed at 48 hours post-transfection. Transgenic Drosophila were made using P-element mediated transgenesis (Bestgene, CA, USA), and were crossed with engrailed-Gal4 (Bloomington Stock Center) (Brand and Perrimon, 1993; Millard and Martin, 2008), srp-Gal4 (Bruckner et al., 2004) and eval-Gal4 (Bloomington Stock Center) (Landgraf et al., 1999) driver lines.

Sectioning and immunostaining was performed as described previously (Chalmers et al., 2003) using a primary mouse anti-green fluorescent protein (GFP) antibody (1:500, Roche). For in situ hybridization, digoxygenin (DIG)-labelled probes were generated using the X. tropicalis expressed sequence tag (EST) clones from the TTpA043k14 (Vimentin) and TGas140h10 (Nectin2) clones using 10× DIG labelling mix (Roche) (Gilchrist et al., 2004). DIG probe hybridization and staining was performed according to methods described by Harland (Harland, 1991).

Whole-mount imaging was performed using a Leica MZ FLIII fluorescent stereomicroscope and Northern Eclipse software, HeLa cells were imaged with a Olympus IX70 inverted fluorescent microscope with Northern Eclipse software and confocal imaging was performed with an Olympus Fluoview FV1000 imaging system and accompanying software.

The pTransgenesis plasmid resource utilizes four separate, modular plasmid libraries (also referred to as reservoirs or positions, Fig. 1A). We named these plasmid positions with the following terminology: p1, the selection marker, designed to contain all necessary DNA sequences to allow selection of transgenic organisms; p2, the promoter DNA sequence, which is sufficient to drive expression of a coding gene; p3, the coding gene with a poly-adenlyation signal; and p4, containing sequences that facilitate genome integration and/or chromosomal attachment sites. Incubation of a ‘p1 selection marker’, ‘p2 promoter’, ‘p3 coding gene’ and a ‘p4 transgenesis’ vector in the presence of recombinase enzyme in vitro produces a final DNA construction that contains these individual elements in a predictable order (Fig. 1B).

The important outcome of this design is that p3 coding sequences, like GFP, which are used repeatedly in experiments across a wide variety of species, are easily coupled to modular p1, p2 and p4 constructions. Hence, we created a p3 coding sequence library that can be shared and utilized in any compatible species. For this, we cloned into the p3 reservoir 16 different fluorescent proteins, sequences for Gal4-UAS binary transgenic approaches, variations of the Cre-recombinase, and coding sequences that allow genetic manipulation in vivo, such as the dominant-negative Fgf receptor construct (Fig. 1C).

To demonstrate the utility of the pTransgenesis cloning framework, we used Xenopus, a model which did not possess a recombination-based cloning system. Notably, the Xenopus model allows the generation of non-mosaic, fully transgenic organisms in the F0 generation via REMI transgenesis, thus allowing us to rapidly validate pTransgenesis elements using this method (Kroll and Amaya, 1996).

We incorporated three Xenopus-compatible selection markers into the p1 selection marker position: the γ-crystallin promoter driving GFP in the lens (Offield et al., 2000), the elastase promoter driving GFP in the pancreas (Beck and Slack, 1999), and the αCE2x6 promoter driving eCFP (enhanced cyan fluorescent protein) in the lens and pronephros (Matsuo and Yasuda, 1992). Into the p2 position, we inserted five previously characterized tissue-specific promoters, one heat shock-inducible promoter, and four ubiquitous promoters (Fig. 2, Table 1). In addition, we created p4 vectors amenable to I-SceI- and Tol2-mediated transgenesis, some containing SAR-CH4 sequences (human interferon-β scaffold attachment region-chicken β-globin DNase I hypersensitive site 4), reported to protect integrated transgenes from positional effects (Fig. 2A, Fig. 1C) (Allen and Weeks, 2005; Ramezani et al., 2003; Sekkali et al., 2008).

Following recombination with selectable markers in p1, Katushka red fluorescent protein (RFP) in p3 and transgenesis elements in p4, as shown in Fig. 2A, we tested the functionality of several p2 promoter constructs by REMI transgenesis in F0 X. laevis tadpoles (Fig. 2B-K; Fig. 3D,E, middle panels). In all cases, RFP expression was appropriate to the promoter sequences driving it. We next confirmed the functionality of the p4 I-SceI site by establishing transgenic lines via I-SceI-mediated transgenesis (supplementary material Fig. S1).

The pTransgenesis vector framework allows the rapid testing of uncharacterized DNA sequences for promoter activity owing to a commercially available PCR product cloning kit. This method allows cloning of PCR products directly into to the p2 position of the pTransgenesis framework, allowing immediate recombination upstream of a p3 coding gene and downstream of a desired p1 selection marker (Fig. 3A-C). Using this strategy, we tested functionally the promoters of two genes; vimentin, which is highly expressed in glia during Xenopus neural development (Fig. 3D, upper panels) (Yoshida, 2001) and nectin-2, which is expressed in the superficial layer of the neuroepithelium (Fig. 3E, upper panel) (Morita et al., 2010). The new promoters were found to be capable of driving VenusGFP expression in a pattern similar to the endogenous expression patterns of vimentin and nectin-2 in both the F0 and F1 generation, validating the use of these promoters for the study of neural development (Fig. 3D,E, middle and lower panels).

Also, we generated a large repertoire of fluorescent proteins in the p3 coding sequence position, including 15 fluorescent proteins with peak emissions ranging from 475 nm to 630 nm (supplementary material Fig. S2A) (Belousov et al., 2006; Mizuno et al., 2003). Using the p2 CMV promoter, we validated these 15 fluorescent proteins in p3 by linking these sequences with a p1 γ-crystallin RFP or p1 γ-crystallin GFP construction and created F0 transgenic tadpoles using REMI (supplementary material Fig. S2B,C). Taken together, these experiments validated the use of pTransgenesis framework to create a wide variety of transgenic lines in Xenopus.

We adapted the pTransgenesis system to be compatible with two powerful binary transgenic strategies, the Gal4-UAS system, utilized in a variety of model species, including Drosophila, zebrafish and Xenopus (Brand and Perrimon, 1993; Fischer et al., 1988; Hartley et al., 2002; Scheer and Campos-Ortega, 1999) and the Cre-loxP recombination system.

Compatibility of Gal4-UAS binary transgene expression and pTransgenesis was achieved by generating one construct to express p3 Gal4 under the control of a p2 promoter (selectable by RFP in the lens RFP; supplementary material Fig. S3A) and another expressing a p3 transgene of interest downstream of p2 UAS repeats (selectable by GFP in the lens; supplementary material Fig. S3B). Double transgenic lines are easily identified owing to the co-expression of GFP and RFP. We generated a construction expressing Gal4 under the control of the NBT promoter (i.e. expressed in the nervous system; supplementary material Fig. S3C), and VenusGFP or a dominant-negative form of the FGFR1 (XFD) tagged with VenusGFP downstream of p2 UAS (supplementary material Fig. S3D,E). VenusGFP expression was only observed in the nervous system in the double transgenics showing that the binary Gal4-UAS system works (supplementary material Fig. S3F). However, double-transgenic embryos expressing XFD-VenusGFP showed a consistent tail elongation phenotype, suggesting a role for FGF signalling in the nervous system during tail elongation (supplementary material Fig. S3G,H). Importantly, the tail and body length phenotype was also observed in the F1 generation by crossing F0 NBT:Gal4 and UAS:XFD-VenusGFP founder frogs (supplementary material Fig. S3I-K). These data showed the use of pTransgenesis to investigate the role of signalling pathways under strict spatial or temporal control. A more detailed examination of this phenotype is beyond the scope of this report and will be addressed in a separate study.

Another widely used binary transgene approach is via Cre-loxP recombination, in which a floxed transgene is excised by Cre, resulting in the conditional gene activation of another transgene, lying downstream of the distal loxP site (Lakso et al., 1992; Mosimann et al., 2011; Werdien et al., 2001). Using the pTransgenesis resource, we have successfully established via I-SceI transgenesis the first transgenic X. tropicalis line amenable to Cre-loxP recombination. The line expresses cyan fluorescent protein (CFP) ubiquitously in the absence of Cre recombinase (supplementary material Fig. S4A-E), but in the presence of Cre, recombination leads to the replacement of CFP with RFP. We demonstrated the functionality of this transgenic line, by injecting synthetic Cre-recombinase mRNA into one of the two blastomeres at the two-cell stage in the F2 generation of this line, causing the expected hemispheric RFP expression by the neurula stage (supplementary material Fig. S4F-N). This transgenic line will be valuable in long-term fate-mapping studies in X. tropicalis.

The pTransgenesis resource is designed such that the p3 coding sequence library can be easily adapted to other model species, namely, by including species-compatible selection p1 markers and p2 promoters and, if necessary, species-compatible p4 transgenesis constructs (Fig. 4A). In this section, we outline experiments showing the compatibility of pTransgenesis in Drosophila, zebrafish, and HeLa cells.

For zebrafish, we first confirmed that the p1 γ-crystallin RFP, p2 CMV construction, and a p3 fluorescent protein (Midori-Ishi Cyan) properly express in this model organism. We injected a vector with recombined p1 γ-crystallin RFP, p2 CMV and p3 Midori-Ishi Cyan constructs with or without Tol2 mRNA. We found, like others, that the p1 γ-crystallin RFP screenable marker was functional in the zebrafish eye lens (Fig. 4B, arrow), and the CMV promoter was able to drive widespread Midori-Ishi expression in the fish body (Fig. 4B) (Davidson et al., 2003). Furthermore, we cloned the zebrafish HB9 promoter (Flanagan-Steet et al., 2005) into the p2 position, recombined this with the p1 γ-crystallin RFP screenable marker and p3 VenusGFP (supplementary material Fig. S5A,B) and confirmed its activity in zebrafish using Tol2-mediated transgenesis (supplementary material Fig. S5C-E) and in Xenopus using REMI transgenesis (supplementary material Fig. S5F). Notably, we generated a transgenic zebrafish line with this construct using Tol2-mediated transgenesis, showing the functionality of the p4 Tol2 elements in the pTransgenesis system (supplementary material Fig. S5D,E).

To test the pTransgenesis system in mammalian cells, we created a p1 PGK:Puromycin Resistance cassette, which allows selection following application of puromycin to culture media. We linked this selection marker, or the p1 γ-crystallin RFP construction (negative control), to the p2 CMV and p3 HyPerYFP constructs (Fig. 4C,D). Expression of HyPerYFP was observed in ∼50% of cells 48 hours after transfection using either construct (Fig. 4C,D, middle left panels). However, following addition of puromycin in the culture media, only the cells bearing the p1 PGK:PuroR cassette survived and continued to express HyPerYFP (Fig. 4D, middle right panel). Cells were successfully cultured for over a month under selection conditions.

We also tested the pTransgenesis resource in Drosophila. For this purpose, we created a p1 mini-white plasmid, which allows selection for red eyes (Tang and Sun, 2002) and a p4 vector containing sequences enabling P-element mediated transgenesis (Rubin and Spradling, 1982). By linking the p1 mini-white, p2 UAS, p3 VenusGFP and p4 P-element constructions, we created transgenic flies via P-element mediated transgenesis selectable by their red eyes (Fig. 4F). We crossed this pTransgenesis-engineered Drosophila line with previously established transgenic engrailed-Gal4, srp-Gal4 and elav-Gal4 driver lines (Bruckner et al., 2004; Landgraf et al., 1999; Millard and Martin, 2008) (Fig. 4G-K). Control stage 15-16 embryos lacking Gal4 expression showed no fluorescence besides the expected autofluorescence of the gut and cuticle (Fig. 4H, arrowheads) (Bainbridge and Bownes, 1981). By contrast, embryos expressing Gal4 showed the expected Gal4-UAS-driven VenusGFP expression in hemocytes (srp-Gal4, Fig. 4I), motor neurons (elav-Gal4, Fig. 4J) or engrailed domain segments (engrailed-Gal4, Fig. 4K).

Together, these data show that pTransgenesis vectors are compatible with other vertebrate, invertebrate and mammalian cell culture models. Compatibility between pTransgenesis vectors and previously established resources using Multisite Gateway cloning depends on the plasmid ‘RL’-site design of each resource. For example, the previously reported R4-R3 ROSA26 locus targeting (Kappas et al., 2008) and Tol2-transposon-containing vectors (Kwan et al., 2007; Villefranc et al., 2007) function as p4 vectors in the pTransgenesis system. Moreover, the promoters generated in other zebrafish studies are compatible in the p2 promoter position (Fisher et al., 2006). However, unlike earlier projects, pTransgenesis is not designed for the rapid creation of fusion proteins (Akbari et al., 2009; Villefranc et al., 2007) and, thus, these applications are better served by their original resources.

In conclusion, the pTransgenesis resource markedly streamlines the process of creating DNA constructions for engineering transgenic organisms and provides a straightforward framework for the distribution of constructs across various model organisms. An additional salient feature of the pTransgenesis resource is the substantial reservoir of constructs, which readily allows a wide variety of experimental approaches. Finally, the cross-species utility of pTransgenesis will allow researchers to evaluate more easily whether biological processes, which are increasingly being studied with a view to application, are evolutionarily conserved.

We thank Sarah Woolner, Tim Mohun, Shane Woods, Matthew Ronshaugen, Jonathan Slack, Caroline Beck, Robert Grainger, Scott Fraser, Igor Samokhvalov, Koichi Kawakami, Makoto Ikeya, Mototsugu Eiraku, Richard Hawley, Ali Ramezani, Paul Krieg, Lyle Zimmerman, Roberto Mayor and John Gurdon for their kind donations of reagents.