Dear Editor,

Genetic modification of pigs has many agricultural and biomedical applications. Ectopic overexpression of foreign genes is necessary in many cases to generate transgenic pigs with favorable phenotypes1. However, random integration of foreign genes often leads to unpredictable expression and unstable phenotypes2. Rosa26 is ubiquitously expressed in embryonic as well as adult tissues3. Targeting genes to the Rosa26 locus is a desirable method to create transgenic animals consistently expressing foreign genes at a high level. Insertion of a reporter or toxin gene into the Rosa26 locus has been widely used to trace or ablate specific cell lineages4, and this approach plays a fundamental role in understanding cell differentiation in vivo. Rosa26 was first identified and targeted in mouse embryonic stem cells (ESCs) in 1990s3, and then in human ESCs in 20075. With the establishment of rat pluripotent stem cells, the Rosa26 locus was also successfully identified and targeted in rats recently6. However, Rosa26 has not been tackled in large animals due to unavailability of germline-competent pluripotent stem cells. By taking advantage of recently emerging technology of gene editing mediated by transcription activator–like effector nuclease (TALEN)7, here we characterized the porcine Rosa26 (pRosa26) locus and targeted a Cre-dependent reporter gene into the pRosa26 locus. Using this approach, we also created transgenic pigs stably overexpressing a gene of interest through recombinase-mediated cassette exchange (RMCE)5.

We identified and characterized the pRosa26 locus based on a highly conserved promoter region of the Rosa26 locus in mice, rats and humans6. We searched the Ensembl porcine database using the human Rosa26 promoter and exon 1 sequence (1 036 bp) as a template and found a highly conserved region (sequence similarity > 90%) located on the pig chromosome 13. This region contains both the pRosa26 locus and the neighboring genes that have also been found in mice, rats and humans (Supplementary information, Figure S1A). The sequence alignments of porcine, mouse, rat and human Rosa26 promoter and exon 1 showed high sequence conservation (> 75%) among these species (Supplementary information, Figure S1B). In mice, rats and humans, the Rosa26 locus encodes non-coding RNAs that are ubiquitously expressed5,6,8. To determine whether the pRosa26 locus also encodes similar non-coding RNAs, we performed an online screening of the Ensembl porcine gene expression database for the pRosa26 locus, but failed to identify any expressed sequence tags (ESTs) or transcripts. Therefore, we predicted the exon 1 sequence of pRosa26 according to the multiple-sequence alignment result (Supplementary information, Figure S1B) and designed primers aligned within exon 1 to perform 3′ RACE analyses6. A non-coding RNA product of 629 bp transcribed from the pRosa26 locus was identified (Supplementary information, Figure S1C and S1D). RT-PCR and SYBR green-based quantitative PCR assays demonstrated that this non-coding RNA was expressed in a wide variety of adult tissues (Figure 1A and 1B). To determine whether the pRosa26 promoter can drive gene expression ubiquitously like the mouse promoter8, we constructed and transiently transfected a pRosa26 promoter-driven tdTomato transgene into six cell lines. At 48 h post-transfection, a high level of tdTomato expression was detected in all six cell lines, a pattern that was highly similar to the expression of the ubiquitous CMV promoter-driven tdTomato transgene (Figure 1C). Taken together, these data strongly suggest that the pRosa26 locus is an ideal site for ubiquitous expression of exogenous genes.

Figure 1
figure 1

Characterization of pRosa26 and highly efficient gene knock-in and replacement at the pRosa26 locus. (A, B) pRosa26 was expressed in a variety of organ tissues as determined by RT-PCR (A) and quantitative RT-PCR (B). For RT-PCR, the designed primers annealed in exon 1 and exon 2 and amplified a correctly spliced product of 485 bp. Porcine GAPDH was used as a control (234 bp). For qPCR, primers were specific for exon 2. PCR product of the porcine ACTB gene was used as the reference control. Data were presented as the average expression levels from three individual RT/qPCR reactions. (C) pRosa26 promoter-driven tdTomato expression in different cell lines (pRosa26 promoter region was shown in Supplementary information, Figure S1B and S1C). Transient transfection of pRosa26-tdTomato and pCMV-tdTomato vectors was performed in the indicated cell lines. (D) A diagram for TALEN-mediated knock-in of Neo-polyA-iEGFP into the pRosa26 locus. Grey triangles, wild-type (WT) loxP; white triangles, loxP2272 site; SA, splice acceptor. (E) Cre-mediated recombination activates EGFP expression by two mechanisms: Cre induces inversion of the iEGFP flanked by two loxP2272 sites (upper left) followed by excision of Neo flanked by two loxP sites; or Cre induces inversion of both Neo and iEGFP flanked by two loxP sites (upper right) followed by excision of Neo between two loxP2272 sites. (F) RMCE replaces EGFP with tdTomato. (G) Morphologically normal piglets were born from SCNT with the pRosa26-iEGFP PFFs. (H) PCR analysis confirmed the correct homologous recombination at the pRosa26 locus in the 7 piglets generated by SCNT. The pRosa26-iEGFP donor cells were used as positive control (P) and WT pig genomic DNA and water were used as negative controls. Primer pairs were shown in D. (I) EGFP activation by Cre in fibroblasts isolated from the ear tissues of cloned piglets shown in G. Cells were infected with Cre-lentivirus and the EGFP expression was observed after 48 h. (J) Embryos with constitutively activated EGFP expression were generated from SCNT of the pRosa26-EGFP PFFs obtained with a Cre plasmid transient transfection. Shown were an E35 pRosa26-EGFP embryo and its section. WT is an E45 WT embryo. (K) SCNT-generated pRosa26-tdTomato piglets through an RMCE strategy as shown in F.

Full size image

To generate transgenic pigs with stable and ubiquitous overexpression of exogenous genes and also reliable reporters that can be used for stem cell and human disease studies, we attempted to introduce a Cre-dependent EGFP reporter into the pRosa26 locus. As porcine ESCs for gene targeting are not available, we applied the targeting strategy to pig fetal fibroblasts (PFFs) and performed somatic cell nuclear transfer (SCNT) to generate transgenic pigs1. We first applied a traditional homologous recombination-mediated targeting strategy. We selected 425 colonies, but none of them was correctly targeted. Thus, the efficiency of the traditional strategy appeared too low for us to obtain targeted clones.

We then applied the TALEN technology to improve the targeting efficiency. Six TALENs (Supplementary information, Figure S2A) composed of 14.5-16.5 repeats were designed based on published guidelines7. The activity of the TALENs was tested using a single-strand annealing (SSA) assay9 (Supplementary information, Figure S2B and S2C) and a T7 endonuclease I (T7EI) assay7 (Supplementary information, Figure S2D and S2E). The TALEN pairs with the highest activity (Figure 1D and Supplementary information, Figure S2F) were used to target the pRosa26 locus. The targeting vector (Figure 1D) contains an 1.2 Kb 5′ arm and a 5.6 Kb 3′ arm for homologous recombination. These homologous arms span 6.8 Kb of the pRosa26 locus containing the putative first exon and the following intron of pRosa26. An expression cassette comprising a viral splice acceptor (SA), a promoterless neomycin-resistance (Neo) gene, and an inverted EGFP (iEGFP) gene was inserted between the homologous arms. The loxP and mutant loxP2272 sites were arranged to flank the Neo and iEGFP genes as indicated in Figure 1D, which could result in removal of the Neo gene and inversion of the iEGFP after Cre-mediated recombination. This Cre-mediated rearrangement would place EGFP directly under the control of the endogenous pRosa26 promoter (Figure 1E). Another advantage of putting heterotypic loxP sites in the knock-in vector is that virtually any gene of interest could be inserted into the pRosa26 locus by RMCE.

For gene targeting, PFFs were electroporated with either the linear or circular targeting construct accompanied with TALEN pairs. After selection with G418 (1 mg/ml from day 10 to day 14), 192 cell clones were expanded and screened by PCR analysis. A total of 46 out of 96 clones (pRosa26-iEGFP) derived from linear construct electroporation and 14 out of 96 clones obtained from circular construct electroporation were correctly targeted based on 5′- and 3′-arm PCR analysis and EGFP expression induced by Cre-lentivirus infection (Figure 1D and Supplementary information, Figure S3A-S3E). The TALEN generates site-specific DNA double-strand breaks (DSBs), which can be repaired by nonhomologous end joining (NHEJ) or homology-directed repair7. Once DSBs are erroneously repaired by NHEJ, unpredictable gene mutations will be generated. Therefore, we next examined the modification patterns in both pRosa26 alleles among these 60 correctly targeted cell clones using PCR-based DNA sequencing. We found that 39 clones had knock-in mutation in one allele and NHEJ-mediated mutation in the other, but no homozygous knock-in clones were identified. We chose the clones with only one Neo-iEGFP knock-in pRosa26 allele but no NHEJ-mediated mutation for further validation. These cells exhibited a normal karyotype (Supplementary information, Figure S3F) and expressed a high level of EGFP when infected with the Cre-lentivirus (Supplementary information, Figure S3C and S3D). Two correctly targeted PFF cell lines obtained from the linear and circular constructs, respectively, were chosen as donor cells for SCNT. A total of 1 088 cloned embryos were generated and transferred into four surrogate mothers. Two surrogates were confirmed pregnant by ultrasound examination one month after the embryo transfer.

To verify whether the expression of EGFP in the cloned pRosa26-iEGFP fetuses could be induced by the Cre-recombinase, a pregnant surrogate was sacrificed at day 35 after the embryo transfer. Six fetuses were collected to isolate PFFs. The PFFs were transiently transfected with a Cre expression plasmid (Figure 1E) and the expression of EGFP in those cells were observed 48 h later. EGFP-expressing (pRosa26-EGFP) cells were sorted by flow cytometry. PCR analysis confirmed the correct removal of the Neo gene and the inversion of iEGFP in the pRosa26-EGFP cells (Supplementary information, Figure S4A). Karyotype analysis showed that these cells contained normal chromosomes (Supplementary information, Figure S4B). The other surrogate mother was allowed to develop to term and delivered seven pRosa26-iEGFP piglets (three live piglets (Figure 1G) and four stillbirths). PCR analysis showed that all the seven piglets were derived from the SCNT donor cells (Figure 1H). We also examined whether the cloned piglets could express EGFP after Cre-mediated recombination. Fibroblasts isolated from the ear tissues of the live piglets were infected with Cre-lentivirus. As shown in Figure 1I, a high level of EGFP expression was observed in these cells (Figure 1I).

To further examine the expression pattern of EGFP in vivo after Cre-mediated recombination, pRosa26-EGFP PFF donor cells were used for the second SCNT. The pregnancy of a surrogate was terminated 35 days after the embryo transfer and six fetuses were obtained. EGFP was indeed ubiquitously expressed in all organs of the six fetuses (Figure 1J). In addition, flow cytometry analysis revealed that all fibroblasts isolated from the fetuses expressed EGFP (Supplementary information, Figure S4C and S4D). All the results above indicate that the pRosa26 locus is an excellent site for ubiquitous expression of exogenous genes. In addition, the Cre-inducible pRosa26-iEGFP reporter line could serve as an ideal tool to trace cell lineages in pigs.

The insertion of heterotypic loxP sites along with EGFP into pRosa26 introduced a homing site (Figure 1E), which could allow us to replace the EGFP cassette at the pRosa26 locus with any gene of interest by RMCE without drug selection. To test the feasibility of this approach, we engineered an exchange vector (Figure 1F) containing a promoterless tdTomato flanked by 5′ loxP and 3′ loxP2272 sites. This exchange vector, together with a Cre expression plasmid (Figure 1F), was electroporated into pRosa26-EGFP fetal fibroblasts. Forty-eight hours after electroporation, tdTomato-expressing cells started to be observed (Supplementary information, Figure S5A). And tdTomato+/EGFP− (pRosa26-tdTomato) cells were sorted 5 days later. PCR analysis further confirmed the successful replacement of EGFP with tdTomato (Supplementary information, Figure S5B). Karyotype analysis showed that the sorted cells contained normal chromosomes (Supplementary information, Figure S5C). Again, pRosa26-tdTomato cells were used as donor cells for SCNT, and four pRosa26-tdTomato piglets (2 live piglets and 2 stillbirths) were obtained. tdTomato fluorescence was directly observed in piglets using goggles10 (Figure 1K). Flow cytometry analysis showed that tdTomato was expressed in all pRosa26-tdTomato piglet ear fibroblasts (Supplementary information, Figure S5D and S5E) and a large majority (98.5%) of blood cells with nuclei (Supplementary information, Figure S5F). To examine the expression pattern of tdTomato in vivo, one piglet was sacrificed and tdTomato expression was observed in all the organs examined under a fluorescence stereomicroscope (Supplementary information, Figure S5G and S5H). Quantitative PCR assays showed that the expression pattern of tdTomato was similar to that of the pRosa26 non-coding RNA in various adult tissues (Supplementary information, Figure S5I). Taken together, these results indicate that transgenic pigs stably overexpressing a gene of interest could be generated through RMCE in the pRosa26 locus without drug selection. The stable and ubiquitous expression of EGFP or tdTomato at the targeted pRosa26 locus is in stark contrast to the expression of tdTomato in a pCMV-tdTomato transgenic pig that we generated through random integration. In that case, only 18% of pig ear-derived fibroblasts expressed tdTomato and the expression levels also varied significantly among individual cells (Supplementary information, Figure S6).

In summary, we have for the first time identified and characterized the porcine Rosa26 locus. Similar to the findings in mice, rats and humans, pRosa26 was also expressed in all tissues examined and the pRosa26 promoter was capable of driving gene expression in all cell lines tested. We genetically manipulated the locus and created a Cre-inducible EGFP reporter pig line, which could be used as a reliable porcine reporter model for lineage tracing studies. We further confirmed that foreign genes inserted in the pRosa26 locus could be expressed ubiquitously. More importantly, the RMCE technology could be used to produce transgenic pigs stably overexpressing any gene of interest at the pRosa26 locus. The pRosa26-targeted pigs, together with the RMCE strategy reported in this study, will serve as an excellent platform for generating genetically modified pigs for biomedical and agricultural applications.