Integration of a FT expression cassette into CRISPR/Cas9 construct enables fast generation and easy identification of transgene-free mutants in Arabidopsis

The CRISPR/Cas9 genome editing technique has been widely used to generate transgene-free mutants in different plant species. Several different methods including fluorescence marker-assisted visual screen of transgene-free mutants and programmed self-elimination of CRISPR/Cas9 construct have been used to increase the efficiency of genome edited transgene-free mutant isolation, but the overall time length required to obtain transgene-free mutants has remained unchanged in these methods. We report here a method for fast generation and easy identification of transgene-free mutants in Arabidopsis. By generating and using a single FT expression cassette-containing CRISPR/Cas9 construct, we targeted two sites of the AITR1 gene. We obtained many early bolting plants in T1 generation, and found that about two thirds of these plants have detectable mutations. We then analyzed T2 generations of two representative lines of genome edited early bolting T1 plants, and identified plants without early bolting phenotype, i.e., transgene-free plants, for both lines. Further more, aitr1 homozygous mutants were successful obtained for both lines from these transgene-free plants. Taken together, these results suggest that the method described here enables fast generation, and at the mean time, easy identification of transgene-free mutants in plants.


Introduction
Shortly after the reports that CRISPR (clustered regularly interspaced short palindromic repeats) RNA-guided Cas9 (CRISPR-associated protein 9) endonuclease is able to cleave double-stranded DNA, and can be used to generate mutations in eukaryotic cells [1,2], CRISPR/ Cas9 mediated genome-editing has been successful used for gene editing to generate mutations in several different plants including the model plant Arabidopsis and crops such as rice, tobacco and wheat [3][4][5]. Since then, the CRISPR/Cas9 genome-editing techniques including

Plant materials and growth conditions
The Columbia ecotype 'Col-0' (Col) Arabidopsis (Arabidopsis thaliana) was used for plant transformation, and as controls for bolting time assays. Seeds of Col wild type were germinated in soil pots, and grown in a growth room. T1 transgenic plants were selected by plating T1 seeds on antibiotic-containing 1/2 MS plates. Transgenic seedlings were transferred into soil pots, and grown in a growth room. As a control, seeds of Col wild type were germinated on 1/2MS plates, seedlings were transferred into soil pots and grown in a growth room. The growth conditions in the growth room have been described previously [31,32].

Constructs
The pHEE CRISPR/Cas9 vector has been described previously [33]. To insert the FT expression cassette into pHEE to generate the pHEE-FT vector, the full-length open reading frame (ORF) sequence of GmFT2a was synthesized and cloned into pUC19 under the control of the double 35S promoter, and terminated by nos [34]. The sequence of the 35S:GmFT2a-nos cassette was then amplified by PCR, and cloned into the pHEE vector at the Pme1 site by using Gibson assembly. The primers used to amplify 35S:GmFT2a-nos from the pUC19 construct were: 5'-CCTGTCAAACACTGATAGTTTGTCGACTCTAGAGGATCC-3' and 5'-GTCGT TTCCCGCCTTCAGTTTACGACGGCCAGTGAATTC -3'.

Plant transformation and transgenic plant selection
Col wild type plants have several mature flowers on the main inflorescence (~5-week-old) were used for transformation. The plants were transformed via Agrobacterium GV3101 mediated floral dipping [35]. T1 seeds collected were germinated on 1/2 MS plates containing 30 μg/ml hygromycin and 100 μg/ml carbenicillin to select transgenic plants.
Bolting time of the T1 plants was observed, and gene editing status was examined by amplifying the coding sequence of AITR1 and sequencing the PCR products obtained.

Isolation of transgene-free homozygous mutants
T2 seeds were collected from selected plants that bolted early and have the AITR1 gene edited. The seeds were then sown directly into soil pots. T2 plants with normal bolting time were selected, the absence of Cas9 T-DNA insertion was confirmed by PCR amplification of Cas9 fragment, and gene editing status was further examined/confirmed amplifying the coding sequence of AITR1 and sequencing the PCR products obtained.

DNA isolation and PCR
To examine the gene editing status of AITR1, DNA was isolated from the leaves of transgenic plants, and the coding sequence of AITR1 was amplified by PCR. PCR products were recovered from gel and sequenced. The sequencing results obtained were then examined and aligned with the coding sequence of AITR1 obtained from Phytozome (http://phytozome.jgi. doe.gov/pz/portal.html). The primers used for PCR amplification of ACT2 and AITR1 coding sequence and have been described previously [30,36].
To confirm the transgene-free status of the mutants obtained, DNA was isolated from leaves of T2 transgenic plants that have a normal bolting time, and Cas9 gene fragment was amplified by PCR. The primers used for PCR amplification of Cas9 were, Cas9-F, 5'-GG ACAACGAGGAGAATGAGG-3', and Cas9-R, 5'-TGTCTCGACCAGCTGCCTCTT-3'

Bolting time assays
For the T1 plants, the date of bolting was recorded, and days after the transgenic seedlings were transferred into soil pots were calculated and used as the bolting time. Transferred Col wild type plants were used as controls.
For T2 plants, the date of bolting was recorded, and days to bolting after seed germination were calculated. Col wild type plants germinated and grown in soil pots were used as controls.

Generation of a FT expression cassette-containing CRISPR/Cas9 construct
Overexpression of FT in plants is able to promote flowering, therefore shorten the juvenile phase of the transgenic plants even in trees [27]. If used in CRISPR/Cas9 mediated gene editing, the easily visible early bolting phenotype caused by overexpression of FT may serve as an assistant selection marker for easy identification of transgene-free mutants, whereas early flowering due to shortening of the juvenile phase may reduce the length of the overall time required for generating genome edited mutants, thereby providing a method for fast generation and easy identification of genome edited transgene-free mutants.
FT promotes flowering in many plant species including the model plants Arabidopsis, crops such as rice and soybean, and fruit trees such as apple and pear [24][25][26][27]. Considering that overexpression of GmFT2a in both Arabidopsis and soybean promoted flowering in transgenic plants [37,38], we decided to integrate a GmFT2a expression cassette into CRISPR/Cas9 vector for gene editing, hoping that in addition to Arabidopsis, the construct generated may be used, in the future, for genome editing in soybean and possibly other plants.
The full-length coding sequence of GmFT2a was synthesized, and cloned into to the pUC19 vector to generate the 35S:GmFT2a construct. The whole 35S:GmFT2a-nos expression cassette was then PCR amplified and cloned into the pHEE vector [33], at the pme1 site to generate the FT expression cassette-containing CRISPR/Cas9 vector pHEE-FT ( Fig 1A). In this vector, the FT expression cassette was inserted within the T-DNA fragment of the pHEE-FT vector as an independent expression cassette. The use of the strong double 35S promoter to drive the expression of GmFT2a will enable the expression of GmFT2a at a high level in transgenic plants, therefore may lead to an easily visible early flowering phenotype in the transgenic plants.

T1 transgenic plants generated using the FT expression cassette-containing CRISPR/Cas9 construct showed an early bolting phenotype
To examine if the genome editing using the pHEE-FT vector may enable fast generation and easy identification of transgene-free mutants, we made a pHEE-FT CRISPR/Cas9 genome editing construct to target AITR1 (Fig 1B), a novel transcription factor gene that has been shown to regulate ABA signaling and abiotic tolerance in Arabidopsis [30], at two different target sites ( Fig 1C).
After selected on antibiotic-containing plates, more than 70 T1 independent transgenic plants were obtained. After transferred into soil pots, some of them failed to survive. As expected, early bolting phenotype was observed in the transgenic plants (Fig 2A). Bolting was observed in transgenic plants as early as 8 days after the seedlings were transferred from the antibiotic-containing plates into soil pots, and the bolting time for these T1 transgenic plants ranged from 8-19 days after the transfer, whereas that for the Col wild type plants ranged from 17-20 days after the transfer (Fig 2B).

Homozygous genome edited mutants were obtained in the T1 transgenic plants
The pHEE-FT-AITR1 CRISPR/Cas9 construct was made to target two sites in AITR1, therefore fragment deletion should be expected in the transgenic plants if both sites can be edited. Because we intended to use early bolting phenotype as an assistant marker for transgene-free mutant isolation, transgenic plants with an early bolting phenotype, i.e., bolted 8 or 9 days after being transferred, were chosen for fragment deletion examination by PCR. Yet we included a few plants with medium or normal bolting time in this experiment to examine if this is a co-relationship between early bolting phenotype and genome editing status.  FT expression cassette enables fast generation and easy identification of genome edited transgene-free mutants Indeed, smaller PCR product band was obtained (Fig 3A). However, among about 50 T1 plants examined, only two plants, i.e., lines 14 and 59 with early and medium bolting phenotype, respectively yield two PCR bands, but none of them produced only one smaller band ( Fig  3B). All other plants produced only one larger PCR band with expected site for the full-length coding sequence of AITR1 (Fig 3A). These results suggest that both target sites in AITR1 can be edited, but homozygous fragment deletion mutants were not obtained in the plants examined. We therefore sequenced the full-length AITR1 PCR products obtained from some of the plants, including 15 early bolting and 2 normal bolting plants that produced only one PCR product band, to see if we may get homozygous mutants with only one AITR1 target site edited. Indeed, two of the early bolting plants, lines 11 and 15 were identified as homozygous mutants with a single nucleotide insertion at the second target site of AITR1 (Fig 3B). One of the T1 plants with normal bolting phenotype, line I7 was edited at the second site with a single nucleotide insertion, but was a biallelic mutant (Fig 3B). We also found that another 9 early bolting plants sequenced were edited in one of the target sites but were heterozygous (Fig 3B). It should note that among the single site edited mutant, only one was edited at the first target site (Fig 3B).

Homozygous genome edited transgene-free mutants were obtained in the T2 plants
Having shown that both homozygous and heterozygous mutants can be obtained in early bolting T1 transgenic plants (Fig 3), we decided to further examine whether transgene-free mutants can be easily obtained in T2 generation base on phenotypic observation. Four representative early bolting mutant lines, i.e., lines 6, 9, 14 and 15 were selected for the experiments, and T2 seeds were germinated directly in soil pots. Line 14 was chosen because PCR results indicated that both target sites of AITR1 were edited in this line, where as line 15 was a homozygous, and lines 6 and 9 were heterozygous mutants based on sequencing results.
Segregation on bolting phenotype was observed in T2 plants (Fig 4A). Multiple plants without early bolting phenotype were obtained for all four lines, and the segregation ratio for early and normal bolting plants was about 3:1 (Fig 4B). Quantitative analysis for lines 14 and 15 that the average bolting time were about 15 and 25 days after germination for the early and the normal bolting plants, respectively (Fig 2B).
We then examined if the plants without early bolting phenotype were transgene-free plants by amplification of Cas9 gene. As shown in Fig 5, no PCR products were obtained in all the normal bolting plants examined, whereas PCR products were obtained in early bolting plants. These results suggest that transgene-free plants can be easily identified by bolting phenotype.
By using PCR amplification, we found that two of the normal flowering plants in line 14 produced only a small band (Fig 6A), indicating that they were homozygous mutants. Sequence results showed that a 428bp fragment in AITR1 was deleted in this mutant, leading to a few amino acid substitutions and premature stop after the 89 th amino acid residue ( Fig  6A). One of the two randomly selected early bolting plants also produced a small band, however, as expected, it was not a transgene-free mutant (Fig 6A).
We then sequenced AITR1 PCR products from two of the transgene-free plants produced by line 15 to confirm its homozygous genome editing status, and the normal bolting plants produced by lines 6 and 9 to identify homozygous mutants. We found that line 15 was indeed a homozygous mutant with a heritable single nucleotide insertion occurred in the second target site. We also successfully identified one homozygous mutant plant from line 6, and three from line 9. Interestingly, in all the four homozygous mutants identified from lines 6 and 9, the AITR1 gene was edited the same way as that in line 15. In these homozygous mutants, the single nucleotide insertion in AITR1 led to a few amino acid substitutions, and premature stop occurred after the 232 nd amino acid residue (Fig 6B).

Discussion
Removal of the Cas9 T-DNA from the transgenic plants is likely necessary to get stable and heritable mutants, expecially for commercial use of the genome-edited crops [7,10], whereas FT has been used to accelerate the plant breeding process by reducing juvenile phase of many plants [15,27,29]. By inserting the GmFT2a expression cassette into the pHEE CRISPR/Cas9 vector (Fig 1), we established a method for fast generation and easy identification of genomeedited transgene-free mutants in Arabidopsis (Fig 7).
In this method, the easily visible early flowering phenotype can be used as an indicator of plants with Cas9 T-DNA. In the T1 generation, only plants with early bolting phenotypes should be selected and genome editing status should be examined to identify mutations. In the T2 generation of the genome-edited T1 plants, only plants without early flowering phenotypes, i.e., transgene-free plants should be selected, and genome editing status should be examined to ensure that genome-edited transgene-free homozygous mutants will be obtained (Fig 7).
By using the pHEE-FT vector to generate a CRISPR/Cas9 genome editing construct for simultaneously targeting two sites in the ABA signaling and abiotic stress tolerance regulator gene AITR1 [30], we successfully obtained genome-edited transgene-free aitr homozygous mutants (Fig 6). In the T1 generation, about two thirds of the early bolting plants examined have at least one target site edited (Fig 3), indicating that insertion of the GmFT2a expression cassette into the CRISPR/Cas9 vector did not affect the editing efficiency of Cas9. However, we noted that including the two plants with fragment deletion, only three plants had mutations at the first target site, suggesting that the two targets selected have different editing efficiency. Had both target sites had high editing efficiency, we should have been able to obtain homozygous fragment deletion mutants in T1 generation. Nevertheless, we obtained homozygous mutants with mutations that occurred at only one target site from the T1 plants (Fig 3), and we isolated transgene-free homozygous fragments deletion mutants in T2 generations (Fig 6).
Even though only the double-stranded DNA cleave based CRISPR/Cas9 genome-editing system was examined in this study, the concept used in this study may also be applied to the nucleotide substitution based CRISPR/Cas9 genome-editing system to facility transgene-free mutant isolation.
It should note that target site editing was also observed in T1 transgenic plants with medium or normal bolting time (Fig 3), suggesting that editing efficiency may not be positively correlated with early bolting phenotypes. This is likely because FT and Cas9 in the vector were in two different expression cassettes, and were driven by different promoters (Fig 1), thus their expression levels may not always positively related in the transgenic plants. Consider that mutants can be obtained from early bolting T1 plants based on PCR results only or combined with sequencing (Fig 2), and transgene-free plants can be easily obtained from offsprings of the early bolting T1 plants based solely on bolting phenotype segregation (Fig 5), only T1 plants with early bolting phenotypes should be selected for next step analysis when using this method to generate for genome edited transgene-free mutants.
In addition to enabling easy identification of transgene-free mutants based on phenotypic observation, the early bolting phenotype also reduce the length of the juvenile phase, which led to reduced overall time length required for generating genome edited transgene-free mutants. In our case, the different bolting time between early bolting transgenic plants and Col wild type plants were more than 10 days even in soil grown T2 generations (Fig 2). Because early bolting plants were selected in T1, but plants with normal bolting time were selected in T2, the overall time length required for generating genome edited transgene-free mutants in Arabidopsis will be reduced by at least 10 days. In some cases, transgene-free mutants may not be able to be identified in T2 generations, and had to be identified on T3 generations. In that case, identifying mutants from offsprings of the gene-edited early bolting T2 plants may save more time than from that of the transgene-free heterozygous T2 mutants. However, it should be considered that early flowering Arabidopsis plants are weak, thus need to be taken good care of in order to obtain seeds. It should also be noted that because overexpression of FT promoted flowering, CRISPR/Cas9 vector with FT expression cassette is not suitable for editing genes involved in the regulation of flowering time in plants.
In this study, the effects of integration of FT expression cassette into CRISPR/Cas9 vector to accelerate transgene-free mutant isolation was tasted only in Arabidopsis, however,  6, 9 and 15 (B). DNA was isolated from leaves collected from at least 2 individual normal bolting T2, and used as a template to amplify the coding sequence of AITR1 and Cas9. Amplification of ACT2 was used as a control. The PCR products were recovered from gel and sequenced. M, 250 bp DNA maker. Sequencing results were compared with coding sequence of AITR1 to check the editing status. ORF of the AITR1 sequences in the aitr1 mutants were identified on ORFfinder (https://www.ncbi.nlm.nih.gov/orffinder/), and corresponding amino acid sequences were used for alignment with AITR1 amino acid sequences. Triangle indicates fragment deletion, arrow indicates nucleotide insertion, and underlines in the sequence results indicate the PAM sites. Numbers in the alignment indicate the position of amino acid relative to the first Met of AITR1. https://doi.org/10.1371/journal.pone.0218583.g006 considering that the juvenile phase for most crops such as rice and soybean lasts for months, and that for most of the trees including fruit trees such as apple and pear lasts for years, whereas overexpression of FT greatly reduced the length of juvenile phase in most of the plants examined, including all the plants such as the model plants Arabidopsis, crops including rice and soybean, and fruit trees such as apple and pear [25][26][27][28][39][40][41], the method described here may benefit even more for genome editing based breeding for the plants with a long juvenile phase.
On the other hand, transgenic plants of different plant species may need to be selected in different antibiotics, and multiple genes may need to be edited simultaneously, by replacing the antibiotic gene and/or the sgRNA clone cassette, the pHEE-FT vector reported here may be used for genome editing for a single or multiple genes in different plant species.