Utility of I-SceI and CCR5-ZFN nucleases in excising selectable marker genes from transgenic plants

Objectives Removal of selection marker genes from transgenic plants is highly desirable for their regulatory approval and public acceptance. This study evaluated the use of two nucleases, the yeast homing endonuclease, I-SceI, and the designed zinc finger nuclease, CCR5-ZFN, in excising marker genes from plants using rice and Arabidopsis as the models. Results In an in vitro culture assay, both nucleases were effective in precisely excising the DNA fragments marked by the nuclease target sites. However, rice cultures were found to be refractory to transformation with the I-SceI and CCR5-ZFN overexpressing constructs. The inducible I-SceI expression was also problematic in rice as the progeny of the transgenic lines expressing the heat-inducible I-SceI did not inherit the functional gene. On the other hand, heat-inducible I-SceI expression in Arabidopsis was effective in creating somatic excisions in transgenic plants but ineffective in generating heritable excisions. The inducible expression of CCR5-ZFN in rice, although transmitted stably to the progeny, appeared ineffective in creating detectable excisions. Therefore, toxicity of these nucleases in plant cells poses major bottleneck in their application in plant biotechnology, which could be avoided by expressing them transiently in cultures in vitro. Electronic supplementary material The online version of this article (10.1186/s13104-019-4304-2) contains supplementary material, which is available to authorized users.


Introduction
Selection marker genes are indispensable tools in genetic engineering. Their presence in transgenic crops, however, could be detrimental [1], requiring methods for removing them from the plant. The most desirable outcome is to precisely delete the marker genes without creating offtarget mutations. The Cre-lox site-specific recombination system is highly successful in achieving that goal [2][3][4], but it leaves a reactive footprint, the functional lox site, in the genome, rendering it non-reusable for the next round of transformation [5,6].
The double-stranded break (DSB) repair mechanism has long been proposed as an alternative approach for excising marker genes, which can be repeatedly used in the same transgenic line as this mechanism destroys the target site by creating insertion-deletions (indels). Several nucleases, including meganucleases, ZFN, and CRISPR/Cas have been used for creating concomitant DSBs to achieve transgene deletions in the plant cells [7][8][9][10][11]. However, their applications in generating markerfree plants needs more investigation. This study evaluated the effectiveness of codon-optimized I-SceI [12] and CCR5-ZFN [13] in excising genes in rice and Arabidopsis using overexpression and inducible expression approaches. These two nucleases were chosen because they have been successfully used in plant genome engineering [10,[14][15][16].
In this study, the expression of I-SceI and CCR5-ZFN appeared to be deleterious as indicated by the failure to transform rice with the overexpression constructs, indicating their activity on non-canonical target sites. The inducible expression was ineffective in creating excisions Open Access BMC Research Notes *Correspondence: vibhas@uark.edu 4 Dept. of Horticulture, University of Arkansas, Fayetteville, AR, USA Full list of author information is available at the end of the article in plants and/or transmitting them to the progeny. Retransformation approach, on the other hand, was successful in creating targeted excision in cultures in vitro. Therefore, the use of nucleases in plants is hampered by their genotoxic property and lower efficiencies, but retransformation of in vitro cultures could serve as a practical solution for creating targeted excisions, which could then be regenerated into plants. However, several 'excision events' will have to be screened for precise targeted excisions and the potential off-target mutations.

DNA constructs, plant transformation, and treatments
All constructs were prepared using the standard molecular biology techniques. The synthetic coding sequences of I-SceI and CCR5-ZFN were provided by Drs. Holger Puchta (Karlsruhe, Germany) and Joseph Petolino (Dow Agro Sciences, Inc.), respectively. Agrobacteriummediated and biolistics-mediated rice (Nipponbare) transformations have been described earlier [9,17]. Arabidopsis (Col-0) transformation was done using the floral-dip method [18]. Heat-shock treatments of rice in vitro cultures, cut leaves or the seedlings was done by placing the tissues in the petri-dish or wrapped in aluminum foil in an incubator maintained at 42 °C for 3 h, followed by 72 h of recovery before scarifying the tissue for DNA/RNA isolation. For Arabidopsis, seedlings in the germination media (MS media without sucrose) were placed in 40 °C for 3 h followed by 48 h of recovery.

Molecular analysis
The PCR primers were designed using Primer Blast tool and verified in the IDT oligo-analyzer for the hairpin, self and heterodimer structures. They were also checked by BLAST to look for any potential non-specific sites in the rice and Arabidopsis genomes. Primers used in the present study are given in Additional file 1: Table S1. PCR was performed at 94 °C for 4 min followed by 40 cycles of 1 min at 58-60 °C and 1-2 min at 72 °C depending on the amplicon size (unless otherwise stated) using Emerald Amp PCR master mix (TaKaRa Inc.). All the PCR assays included the non-transformed rice or Arabidopsis genomic DNA as the negative control to screen for any non-specific amplification. For gene expression analysis, total RNA isolated using RNaesy kit (Qiagen Inc.) was subjected to real-time PCR using Super Script III one step qRT-PCR kit (Invitrogen) using manufacturer's instructions. Relative expression was calculated against wild-type using 2 ΔΔCt method [19], and the Ct values were normalized against internal control, Ubiquitin or Phytoene desaturase genes. The purified PCR products were sequenced at Eurofin Genomics USA. Genomic DNA of selected lines were also analyzed on Southern blot using P32-labeled DNA probes.

Expression of I-SceI and ZFN in rice
The overexpression constructs consisting of ZmUbi1 promoter for I-SceI or ZFN expression (Fig. 1a)  hygR lines were generated that were PCR-positive for ZFN gene. However, only 3 of these set a low number of seeds (10-30 seeds/line), indicating high rate of sterility in ZFN rice plants. The PCR analysis of the T1 plants from these three lines revealed lack of inheritance of the ZFN gene (Additional file 2: Figure S1). Therefore, strong expression of ZFN also generated toxicity in rice cells that severely hampered inheritance of the ZFN gene. The BLASTn analysis, (using default parameters-input: 33 or 18 bp; e-value threshold: 10; match/mismatch score: 1, − 3; gapopen: − 5 and gapextend: − 3) of 18 bp I-SceI and 33 bp CCR5 sites did not reveal match in the rice or Arabidopsis genome. The online tools for predicting offtarget of I-SceI are lacking, but five I-SceI like sites [20] were also used in the BLASTn analysis, none of which found a 100% match in the rice or Arabidopsis genome.
Off-target prediction of the CCR5-ZFN by Prognos tool [21] found 12 highly probable sites in the rice genome. Next, inducible expression constructs consisting of GmHSP17.5E gene promoter expressing I-SceI or ZFN (Fig. 1b) were co-transformed with hygR gene into Nipponbare callus. Seven I-SceI and 8 ZFN lines were recovered, indicating curbed toxicity of the inducible I-SceI and ZFN in rice. Expression analysis was conducted on heat-shock-treated (HS) cut leaves obtained from the greenhouse grown plants. Five HS-ISceI lines and seven HS-ZFN lines showed several fold increase in the expression with respect to the untreated control, confirming proper regulation of these nucleases in the rice plant (Fig. 1c, d). The HS-ZFN lines showed normal growth and fertility, and transmitted ZFN activity to the progeny. The HS-ISceI lines, on the other hand, did not transmit I-SceI gene to the progeny and showed poor growth and high sterility, indicating toxicity of the basal expression of the inducible I-SceI gene to the somatic and germ cells.

Characterization of inducible ZFN activity in excising marker gene in rice plants
While the experiments with HS-ISceI had to be discontinued due to problematic heritability of I-SceI gene, HS-ZFN lines were cross-pollinated with CCR5 target lines developed by transformation of Nipponbare rice with pBP5 that contains three gene cassettes, GFP, HPT and NPT, with a pair of 33 bp CCR5 sites flanking the HPT cassette (Fig. 2a). Targeting of CCR5 sites by ZFN could lead to the excision of HPT and fusion of the distal ends creating indels at the targeted sites (Fig. 2b). Five healthy F1 plants representing three different ZFN lines (lines #3, #6, #7; Fig. 1b) and two different CCR5-target lines ( Fig. 2c) were heat-shocked and grown to maturity in the greenhouse. All F1 plants expressed GFP and the HSinduced ZFN activity, confirming the presence of CCR5 target and ZFN constructs; however, excision of the HPT cassette was undetectable by PCR across CCR5 sites (data not shown). Several F2 seedlings that were positive for GFP and ZFN were also heat-shocked and sacrificed for DNA isolation, but none showed the excision site (≤ 1.3 kb) in the PCR, while the presence of intact target site (3.5 kb) was evident in a number of them (Fig. 2d). Hence, HS-induced ZFN activity appeared suboptimal in creating detectable excisions in rice. This observation corroborates with that of Lu et al. [22], who reported low frequency targeting by heat-inducible ZFN in poplar.

Targeted excisions by retransformation
The failure in scoring targeted excisions in the F1 hybrids and their progeny derived from the crosses between HS-ZFN and CCR5-target lines raised questions whether ZFN expression was sufficient and the target locus was accessible to ZFN activity. To address these questions, reciprocal transformations were done, i.e., transformation of ZFN-expressing line with pBP5, and transformation of CCR5-target lines with pHS:ZFN. Retransformation of HS-ZFN line #7 with pBP5 generated 19 geneticin-resistant calli events that expressed GFP, indicating stable integration of the target construct in the genome. PCR across CCR5 sites found that 17 of these lines showed both full-length HPT cassette (3.5 kb) and the excision site (≤ 1.3 kb) in the room temperature (RT) samples, 4 of which showed strong presence of excision site in the heat-shock (HS) samples (Fig. 2e). These data suggest that basal ZFN activity from HS:ZFN gene could induce targeting at CCR5 sites but the targeting efficiency increased upon HS treatment. Four regenerated plants were obtained from these callus lines that also showed the ~ 1.3 kb excision site (Fig. 2e). Similarly, transformation of the CCR5-target lines with pHS:ZFN vector, produced 9 calli events, 4 of which showed ~ 1.3 kb excision band in HS-treated calli (Fig. 2f ). Sequencing of five excision sites (≤ 1.3 kb) from these experiments found complete or partial excision of HPT cassette with large indels (> 1.5 kb) spreading into the adjacent sequences (Fig. 2g). In summary, HS-induced ZFN activity is capable of creating targeted excisions in rice cultures in vitro.

Inducible I-SceI mediated marker excision in Arabidopsis
Since I-SceI expression was highly toxic in rice, further experiments with inducible I-SceI were carried out in Arabidopsis. For this purpose, pEP4b construct was developed that contains a pair of I-SceI target sites flanking the GFP cassette, the kanamycin resistance (NPT) cassette, and the HS-inducible I-SceI expression cassette (Fig. 3a). The excision of the GFP cassette in this construct would result in fusion of I-SceI and NPT cassette with indels in between (Fig. 3b). Transformation of Arabidopsis Col-0 with pEP4b generated 11 kanamycin resistant T1 lines that contained a full-length integration of pEP4b construct in the PCR assay (Fig. 3c). Fertility in these T1 plants was substantially low, indicating I-SceI toxicity in the germline (≤ 10× lower compared to that of the healthy Arabidopsis plants). Germination of T2 seedlings on kanamycin-containing (50 mg/l) media displayed gradual lethality and receding GFP expression in all lines; however, seedlings could be rescued on a kanamycin-free medium and grown to maturity. This indicates that large indels possibly occurred at the target sites, eliminating NPT and GFP activity. The rescued T2 seedlings were analyzed by PCR to determine the target and excision sites, indicated by 3.0 and 1.2 kb products, respectively ( Fig. 3a, b). The majority of T2 progeny either failed to show these PCR products or showed their weak presence, indicating large indels at the target site in the majority of the tissue. Two T2 lines showed strong presence of ~ 1.2 kb band (Fig. 3d: white arrows), which was sequenced and found to contain the near-precise excision of GFP cassette with very small indels at the target sites (Fig. 3e). The analysis of T3 seedlings, however, suggested that the observed excision site in the T2 parents was not transmitted to the progeny as none showed the 1.2 kb band (Fig. 3d). In summary, HS-ISceI was able to generate targeted excisions in the Arabidopsis seedlings, but inheritance of the excision site was questionable.

Conclusions
Potential genotoxicity of I-SceI and CCR5-ZFN appears to be a major bottleneck in their application in plant biotechnology. However, retransformation of in vitro

Limitations
The main limitation of this study is that rice and Arabidopsis genomes could contain off-target sites of I-SceI and CCR5-ZFN nucleases that would prohibit the application of these nucleases in these plant species. A larger set of nucleases, e.g., newly designed ZFNs or TALENs should be tested to determine if other nucleases can be used successfully in achieving marker excision in these plant species.

Additional files
Additional file 1: Table S1. Primers used in this study.

Funding
Funding from Arkansas Bioscience Institute, Arkansas NASA-EPSCoR, and USDA-NIFA 2014-02849 supported the project activities. The funding agencies had no role in the design of the study and collection, analysis, and interpretation of data, and in writing the manuscript.

Availability of data and materials
The vectors generated in this study can be requested from the corresponding author. All data generated and analyzed during this study are included in this published article and its additional information.
Ethics approval and consent to participate Not applicable. ISceI -NPT Fig. 3 Characterization of HS-inducible I-SceI in Arabidopsis. a I-SceI target construct, pEP4b, in pPZP200 binary vector contains HS-inducible I-SceI, GFP, and NPT expression units with 18 bp I-SceI target sites (gray bars) flanking the GFP cassette. b Predicted structure of the target site upon precise excision of GFP cassette with indels at the targeted site (dotted bar). PCR primer positions and the fragment sizes are shown by blue arrows. c PCR analysis of the first generation transgenic (T1) lines using primers located in I-SceI and NPT cassettes with pEP4b and wild-type Col-0 as controls. d PCR analysis of three generations: T1 parents, T2, and T3 progeny to detect excision of GFP cassette. White arrows indicate bands that were purified and subjected to Sanger sequencing. e DNA sequences of ~ 1.2 kb predicted excision bands were aligned with the pEP4b reference to determine indels at the targeted sites. Red and blue fonts represent the two I-SceI sites with predicted breakpoints (^). Dotted lines indicate deletions and green small letters show insertions