Localized egg-cell expression of effector proteins for targeted modification of the Arabidopsis genome

Authors


(fax +972 89344181; e-mail avi.levy@weizmann.ac.il).

Summary

Targeted modification of the genome is an important genetic tool, which can be achieved via homologous, non-homologous or site-specific recombination. Although numerous efforts have been made, such a tool does not exist for routine applications in plants. This work describes a simple and useful method for targeted mutagenesis or gene targeting, tailored to floral-dip transformation in Arabidopsis, by means of specific protein expression in the egg cell. Proteins stably or transiently expressed under the egg apparatus-specific enhancer (EASE) were successfully localized to the area of the egg cell. Moreover, a zinc-finger nuclease expressed under EASE induced targeted mutagenesis. Mutations obtained under EASE control corresponded to genetically independent events that took place specifically in the germline. In addition, RAD54 expression under EASE led to an approximately 10-fold increase in gene targeting efficiency, when compared with wild-type plants. EASE-controlled gene expression provides a method for the precise engineering of the Arabidopsis genome through temporally and spatially controlled protein expression. This system can be implemented as a useful method for basic research in Arabidopsis, as well as in the optimization of tools for targeted genetic modifications in crop plants.

Introduction

A variety of methods have been developed to allow for precise modifications of genomes. These include: (i) gene targeting (GT), performed via the targeted integration of exogenous DNA by means of homologous recombination (HR), (ii) site-specific recombination at a pre-introduced site, e.g. Cre-Lox recombination, (iii) DNA cleavage executed by engineered endonucleases, resulting in error-prone, non-homologous end-joining and subsequent targeted gene mutagenesis (TGM), and (iv) oligo-mediated mutagenesis. These techniques enable precise modifications, such as gene knock-outs, mutations or insertions, at specific genomic loci. They are routinely implemented in yeast (Rothstein, 1991) and mice (Bogue, 2003), deeming them advantageous model organisms. The model species for plant research is Arabidopsis thaliana: its genome has been sequenced (Arabidopsis Genome Intiative, 2000), there are many mutant lines (Alonso et al., 2003) and it is fairly easy to transform. Its transformation does not require tissue culture, but is typically performed by floral dipping (Clough and Bent, 1998; Zhang et al., 2006). Budding plants are dipped in a solution containing Agrobacterium tumefaciens, and are transformed by the genetically engineered bacterial transfer DNA (T-DNA). Transformed cells, which successfully transmit the T-DNA to the next generation, are those that stem from the female megaspore, including the egg cell (Desfeux et al., 2000). T-DNA integration into the plant genome occurs in a non-homologous, random manner (Kim et al., 2007; Tzfira et al., 2004); it is very difficult to achieve precise targeted integration, and even more difficult to achieve homologous integration, in this model organism. Thus, the development of an efficient method allowing for targeted genetic modifications in Arabidopsis remains an important goal for plant research per se, as well as for applications in crop plants.

To date, few groups have successfully performed TGM or GT in Arabidopsis. Most of the reported TGMs in Arabidopsis have been obtained by use of engineered zinc-finger nucleases (ZFNs) (de Pater et al., 2009; Lloyd et al., 2005; Osakabe et al., 2010; Tovkach et al., 2009; Zhang et al., 2010), whereas more recently, via transcription activator-like effector nucleases (TALENs) (Boch et al., 2009; Cermak et al., 2011; Mahfouz et al., 2011). In principle, these nucleases can be engineered to cut any site along the genomic DNA, resulting in the formation of a double-strand break (DSB). Repair of this double-strand break by the non-homologous end-joining (NHEJ) mechanism usually leads to small insertions/deletions (Gorbunova and Levy, 1997), hence resulting in TGM (Urnov et al., 2010; Weinthal et al., 2010). Moreover, it has been shown in plants that DSBs induce HR-driven repair (Puchta et al., 1993). Such a repair mechanism can be used for making precise insertions or mutations, providing that the appropriate template for HR is present. A recently published example of endogenous GT executed via an HR-driven repair mechanism mediated by ZFN DSB formation involved tobacco Acetolactate synthase gene targeting, yielding herbicide resistance, and showed GT frequencies of over 2% in tobacco protoplasts (Townsend et al., 2009). Similarly, Shukla et al., 2009 performed GT of the endogenous IPK1 gene in Zea mays, conferring herbicide tolerance and decreased phytate levels: two agronomically relevant outcomes (Shukla et al., 2009). Expression of HR-promoting proteins, such as yeast RAD54, have also been applied to enhance GT frequencies in plants (Shaked et al., 2005). The efficiency and success of the noted GT and TGM methods largely depends on the spatial and temporal expression of the GT- or TGM-promoting proteins for the genetic modification to occur and pass to the next generation.

We hereby describe an approach whereby GT- and TGM-promoting proteins are expressed in a spatiotemporal-specific pattern that coincides with the use of the floral-dip transformation method. For this purpose, we developed a system in which the GT-/TGM-promoting proteins are specifically expressed in the Arabidopsis egg cell, the target of Agrobacterium transformation with the floral-dip method, by exploiting the egg apparatus-specific enhancer (EASE), described by Yang et al. (2005). EASE is a characterized 77-bp enhancer sequence that efficiently controls specific gene expression in the egg apparatus of Arabidopsis (Yang et al., 2005). The work described here shows that ZFN and RAD54 proteins, expressed downstream to EASE, catalyze TGM and GT at the desired time and location, thereby resulting in induced, germline-restricted genetic modifications, which were, in turn, transmitted to the next generation.

Results

Proteins expressed downstream to EASE are localized to the egg cell

Two EASE-regulated, egg cell-localized protein expression systems were established in Arabidopsis for egg cell-specific expression of genome-modifying proteins (both are variations of the original system developed by Yang et al., 2005). The first system combines EASE and the trans-activation system (Moore et al., 1998), in which a driver transgene, EASE:LhG4, produces a chimeric transcription factor in the egg cell that activates the responder transgene, Op:mRFP or Op:ScRAD54 (Op stands for the Lac operator sequence that is bound by LhG4). Microscopic analysis of mRFP reporter expression in EASE>>mRFP plants confirmed the egg cell-specific expression pattern of EASE in the Arabidopsis ovule (Figure 1). An EASE:LhG4 line showing strong mRFP expression was then selected for further crossing with Op:ScRAD54 lines.

Figure 1.

 A transactivation system for expressing recombination proteins in the egg cell.
Confocal microscope images of an ovule at the flowering stage, before anthesis, collected from an EASE>>mRFP plant line. EASE>>mRFP denotes the transactivation of mRFP by the chimeric LhG4 activator expressed under the EASE enhancer. Panels from left to right: DIC; DIC, DAPI (blue) and mRFP (red) overlay; DIC and mRFP (red) overlay. A high mRFP signal (red) is detected in the egg cell area. Scale bar: 20 μm. The constructs used for the transactivation in this plant are shown below the picture and are described in detail in the Experimental procedures.

The second system positioned the gene of interest along the same binary vector as, and directly downstream to, EASE (Figure 2a, EASE:EGFP). Arabidopsis plants were transformed with EASE:EGFP using the floral-dip method. Ovules were dissected and visualized 2 and 6 days following transformation. At 2 days post-transformation, only one of 48 examined ovules showed weak expression of EGFP localized to the egg cell. Six days following transformation, 7% of the 59 ovules analyzed contained EGFP expression localized to the egg cell (Figure 2). The seeds from these plants were collected and EASE:EGFP T-DNA integration was assessed by measuring resistance to glufosinate (Basta). Only 0.44% of the 14 400 seeds tested contained integrated EASE:EGFP sequences, indicating a significant difference between the number of ovules demonstrating transient expression and the number that underwent stable genomic integration of the T-DNA. These results indicate that EASE can be applied for transient gene expression in the egg cell. Transient expression, as seen by the EASE:EGFP reporter gene, can be detected within approximately 2 days of transformation, and increases further over the following 4 days. Fluorescent proteins expressed under EASE via the transactivation and transient expression systems, localized to the region of the egg cell (Figures 1 and 2), indicate the efficacy of both approaches in targeting protein expression to the egg cell.

Figure 2.

 Cellular localization of EASE-EGFP following floral-dip transformation.
(a) Illustration of an Arabidopsis ovule at the eight-nuclei stage (adapted from Yadegari and Drews, 2004). The location of the egg cell and the synergid cell is indicated as ec and sc, respectively. The right section of panel (a) is a scheme of the construct used for EGFP expression under the EASE enhancer, with the Basta resistance gene (BaR).
(b) Arabidopsis plants (7-weeks old) were transformed, using the floral-dip method, with Agrobacterium carrying the EASE-EGFP construct. Six days following transformation, Arabidopsis ovules were dissected from ovaries of flowers, one stage before anthesis. Ovules were kept in double-distilled water and cellular localization was analyzed using a laser confocal microscope, as described above. Panels from left to right: EGFP (green); DIC; and an overlay of DIC + EGFP (green). Scale bar: 20 μm.
(c) Quantification of transformation efficiency in transient versus stable transformation. Arabidopsis plants were transformed with the EASE-EGFP construct using the floral-dip method, and transient transformation efficiency was measured by counting the number of ovules exhibiting EGFP expression 6 days following transformation. Stable transformation efficiency was measured by counting seedlings from seeds of the same transformation, which were resistant to Basta selection. The bar with an asterisk indicates a significant difference in the efficiency of transient versus stable transformation (χ2 = 51.24; < 0.0001).

Developmental timing of ZFN activity and TGM frequency under EASE

In order to determine the timing of expression of genes regulated by EASE, we modified a system used by Tovkach et al., 2009, for the assessment of ZFN efficiency in planta. The system is based on NHEJ repair of a DSB made in a GUS gene (Tovkach et al., 2009) mutated to contain a QQR-ZFN (Kim et al., 1997) binding site immediately after the start ATG, followed by a stop codon and then another QQR-ZFN binding site (Figure 3a). Certain indels, resulting from the repair of the DSB, will lead to an in-frame amendment of the GUS gene, and allow for its expression. For our assay, we transformed Arabidopsis plants with T-DNA containing a mutated GUS cassette: pRCS2. [KAN][QQR-TS*::GUS] (Figure 3a). The QQR-ZFN was chosen for testing the EASE system because it is a well-characterized ZFN (Bibikova et al., 2001; Kim et al., 1997) that has been shown previously to be functional in planta. Two stable lines with a single locus of the mutated GUS cassette were transformed with T-DNA, bearing QQR-ZFN downstream of EASE (EASE:QQR-ZFN) and demonstrating hygromycin resistance. X-Gluc staining allowed for monitoring of the precise timing of the DNA cleavage performed by the EASE-driven QQR-ZFN, thereby suggesting the timing of the nuclease and repair activities. The interpretation of the outcomes of the assay is as described by Bechtold et al. (2003). In short, an entire blue seedling and a blue endosperm (i.e. blue seed coat) indicate a very early cleavage and repair event, transpiring before the division of the gametophyte into the eight-nuclei stage (Figure 2a). An entire blue seedling lacking a blue seed coat indicates a cleavage event that occurred in either the egg cell or the very early zygotic stage. Finally, a ‘chimeric’ seedling containing blue sectors indicates somatic cleavage events that occurred following a number of embryo divisions.

Figure 3.

 Developmental timing of EASE expression.
(a) Seedlings were grown from seeds of Arabidopsis plants containing: top, a single locus of the mutated GUS cassette (uidA), which includes a stop codon downstream to the CaMV 35S promoter (35SP), and situated in between two QQR-ZFN recognition sequences (QQR-TS), as well as the 35S terminator (35ST) downstream to the GUS (downstream from the mutated GUS there is a kanamycin resistance gene, NptII); bottom, a QQR-ZFN cassette under EASE, with hygromycin resistance (Hpt).
(b) Seedlings were stained with X-Gluc staining solution, to detect NHEJ repair of the mutated GUS gene. A blue seedling represents a mutated seedling.
(c) Partial sequences of PCR fragments encompassing the QQR-ZFN break site from seven blue seedlings, compared with the wild type (WT). Underlined sequence, start ATG of the uidA gene; bold sequence, DdeI restriction site; italicized sequence, nucleotides inserted during NHEJ; colon, represents a deleted nucleotide; gray sequence, the QQR-TS.

The first experiment was carried out on approximately 15 000 seedlings originating from seeds that were collected from plants bearing the mutated GUS and transiently transformed with EASE:QQR-ZFN. X-Gluc staining of 4-day-old seedlings revealed no blue staining. The second experiment was performed on second-generation seedlings that originated from plants that were selected on kanamycin and hygromycin, indicating the presence of the mutated GUS cassette and the EASE:QQR-ZFN cassette, respectively. A total of approximately 9000 seedlings containing both the mutated GUS cassette and the genome integrated EASE:QQR-ZFN were tested. In this case, seven seedlings contained blue staining throughout the entire seedling, with no staining of the seed coat (Figure 3b). No other staining patterns were observed, indicating that QQR-ZFN under EASE is active in the egg cell or the very early zygote, at early stages of development.

In order to confirm that the seven blue seedlings resulted from independent cleavage events, we performed PCR directly on the seedlings, using primers complementary to the QQR-ZFN break-site surrounding region, followed by sequencing (Figure 3c). Six of the seven seedlings had unique indels, indicating that they originated from independent mutagenesis events, each of which allowed for restoration of the reading frame of the GUS gene.

The rate of mutagenesis evaluated via the GUS staining assay was of seven out of 9000 seedlings: 7.8 × 10−4. However, this method of assessment results in an underestimation of the actual TGM rate, as it only monitors events that led to GUS expression. PCR using primers complementary to the surrounding region of the QQR-ZFN break site was performed on randomly selected second-generation seedlings, in order to allow for a more accurate determination of the rate of EASE-regulated, QQR-ZFN-driven targeted mutagenesis (Figure 4a). The PCR products were then incubated with the DdeI restriction enzyme, which has a unique recognition sequence within the QQR-ZFN break site (Figure 4a, c). In cases of QQR-ZFN-driven DSB repair, indels abolishing the DdeI restriction site are formed (Figure 4b, see lane under arrow). This mode of analysis yielded one of 366 seedlings with a nullified DdeI restriction site, corresponding to a mutagenesis rate of 2.7 × 10−3. Whereas the present sample size is too small for performing a statistical analysis, the calculated mutagenesis rate is approximately threefold that determined by means of the GUS staining assay. These findings compare favorably with the expectation that only one-third of the QQR-ZFN-induced indels will result in a valid reading frame and in GUS expression. The PCR-based assay still underestimates the actual rate of TGM, as it does not consider large deletions that could either obstruct PCR performance or result in small, undetectable PCR products. Additionally, break events later undergoing precise repair are not detectable using this assay.

Figure 4.

 Expression of QQR-ZFN in the egg cell leads to targeted mutagenesis.
(a) PCR was performed on genomic DNA extracted from randomly selected Arabidopsis seedlings of the same origin as those described in Figure 3. Arrows mark primers used for PCR. DdeI marks the site of restriction for this enzyme, used for mutagenesis analysis.
(b) PCR products were digested with DdeI and separated on agarose gel. The vertical arrow points to a lane with DNA from a seedling that was mutated by QQR-ZFN.
(c) Partial sequence of PCR fragment encompassing the QQR-ZFN break site, generated from a randomly mutated seedling (lower sequence) compared with the wild type (WT, upper sequence). Underlined sequence, start ATG of the uidA gene; bold sequence, DdeI restriction site; colon, deleted nucleotide; gray sequence, the QQR-TS.

Expression of ScRAD54 in the egg cell enhances GT events

We tested the effect of egg cell-specific ScRAD54 expression on GT, using the Cru3 GT assay, which was described in detail in Shaked et al. (2005). In the present study, the Cru3 GT assay was modified by replacing both the GFP with mRFP, fused to nucleotides 37–1217 of the Cru3 open reading frame (ORF), and the NOS terminator with the OCS terminator. The terminator was switched because of a more convenient cloning procedure. Successful Cru3 GT events result in easily detectable red fluorescent seeds. An additional assay alteration incorporated in the present model involved the use of plants expressing EASE-regulated ScRAD54, rather than relying on the 35S promoter. Driver lines, expressing the LhG4 activator under EASE, were selected for Op:ScRAD54 expression, based on the intensity and specificity of expression of a Op:mRFP reporter transactivated by the same driver.

Gene targeting rates in EASE>>ScRAD54 Arabidopsis lines were compared with those of wild-type Arabidopsis plants (WT) (Figure 5). In WT plants, only three GT events were isolated from the approximately 6 320 000 seeds screened in 27 independent experiments. The rate of GT in these experiments, namely the ratio of homologous to non-homologous integration, was 3.45 × 10−5, e.g. three GT events out of approximately 63 200 transformed seeds (see GT rate calculation in the Experimental procedures). The EASE>>ScRAD54 lines produced six GT events isolated from the approximately 2 400 000 seeds that were screened and collected from three independent lines. The calculated GT rate out of an estimated 24 000 transformed seeds in these experiments is 2.5 × 10−4. These results indicate that specific expression of RAD54 in the egg cell is associated with a statistically significant approximately 10-fold increase in GT rate [P2) = 0.0002].

Figure 5.

 Expression of ScRAD54 in the egg cell enhances gene targeting (GT).
Frequencies of Cru3 GT were compared between the wild type (WT) and EASE>>ScRAD54 lines. The GT frequency was approximately 10-fold higher in EASE>>ScRAD54 than in the WT. *This difference was statistically significant: P2) = 0.0002. The constructs used for the transactivation in EASE>>ScRAD54 plants are shown below the graph.

Upon further analysis, we found that all three GT events detected in WT plants were non-precise targeting events. These three events showed precise copying of the 5′ genomic Cru3 region upstream of the 5′ region of homology, but vector integration via non-homologous end joining at the 3′ junction. Southern blot analysis of their progeny indicated that the WT allele was not modified in two of the events; the vector underwent ectopic gene targeting by copying the 5′Cru3 region and integrating at an ectopic genomic location, without replacing the WT allele. In the third event the WT allele was changed, but gene replacement was not precise. The six GT events originating from the EASE>>ScRAD54 lines included one true and precise GT event, as implicated from sequencing of both 5′ and 3′ junctions between the vector and the Cru3 genomic locus. Moreover, Southern blot analysis revealed that the WT allele was lost in 25% of the progeny of this specific event, and the band size of the targeted allele indicated precise gene replacement (Figure 6c). In parallel, five non-precise GT events were obtained, all of which yielded the expected 5′ junction, but not the 3′ junction (Figure 6a) when analyzed by PCR. Four of these non-precise GT events showed no change of the WT allele in their progeny, thus suggesting ectopic events (Figure S1). In the fifth non-precise GT event, the WT allele was changed, but not as expected for homologous gene replacement. These results show that although no true GT events were detected in WT plants, in EASE>>ScRAD54 plants we obtained a rate of 4.17 × 10−5 bona fide GT.

Figure 6.

 Validation of Cru3 gene targeting (GT) events originating from expression of ScRAD54 in the egg cell.
(a) PCR fragments of genomic DNA from six EASE>>ScRAD54 GT events. 5′ and 3′ indicate PCR amplification of the 5′ or 3′Cru3 junction between genomic DNA and the GT vector, respectively. One of the six events yielded the expected size for both 5′ and 3′ junctions (lane 1); the other five events yielded reactions at the 5′ junction only (lanes 2–6).
(b) Scheme of the structure of the gene targeting vector, the wild-type (WT) allele and the expected targeted genomic locus. The T-DNA gene targeting vector includes: LB, T-DNA left border; RB, T-DNA right border; and NPT II, the kanamycin resistance selection marker. The probe used in Southern blot analysis detects a 6000-bp fragment in the Cru3 WT allele, following digestion of genomic DNA with HindIII. WT allele, Cru3 WT allele; promoter, Cru3 promoter; M, Cru3 ATG initiation codon; probe, 1062-bp region used for Southern blot probing, located 78 bp downstream to the genomic region that is included in the GT vector. In the targeted allele, digestion with HindIII and hybridization with the same probe detects a 4600-bp fragment. Targeted allele, Cru3 allele that underwent HR with the GT vector.
(c) Southern blot of true GT event progeny. The parent plant yielded both 5′ and 3′ junctions by PCR, as seen in panel (a, lane 1). Genomic DNA was digested with HindIII and hybridized with the 1062-bp probe described in panel (b). WT, Cru3 WT allele; TGT, true gene targeting of Cru3 allele.

Discussion

Currently, the ability to perform targeted modifications of the Arabidopsis genome is quite limited. This work describes a system designed to improve existing methods used for TGM and GT in Arabidopsis. Egg cell-specific expression of TGM- and GT-promoting proteins regulated by the EASE element (Yang et al., 2005) allowed for localized protein activity in the transformed germinal tissue when using the floral-dip transformation method (Clough and Bent, 1998). The present study shows that egg cell-localized expression can be achieved by transient or stable transformation of a reporter protein regulated in cis or trans by EASE. Further evidence of the specificity of EASE, and of its utility in the context of targeted genetic manipulations, comes from the genetic data shown in Figure 3 confirming the highly accurate catalytic activity of EASE:QQR-ZFN in the female megaspore. However, transient expression of the QQR-ZFN did not facilitate increased TGM, although earlier works in tobacco and maize reported stimulation of GT following transient ZFN expression (Shukla et al., 2009; Townsend et al., 2009). Thus, gene targeting must be optimized for each combination of species, cell type and delivery method, and the absence of effect here may result from properties of the EASE-driven transient expression in Arabidopsis, like expression in egg cells and/or Agrobacterium-mediated transient expression during floral dipping. Transient expression of a TGM or GT effector protein would have provided a more time-efficient system, bypassing the need to preselect stable transformants and eliminating the need to remove the effector proteins after fulfilling their role. Therefore, it is worth pursuing the attempts to develop effective transient expression of ZFNs or of the new generation of TALENs proteins (Boch et al., 2009; Cermak et al., 2011). Conversely, stable expression of EASE:QQR-ZFN gave rise to a TGM rate of 0.27%, which fell in line with the range reported in previous ZFN-based systems reporting of TGM of either endogenous or non-endogenous targets in Arabidopsis plants (de Pater et al., 2009; Lloyd et al., 2005; Osakabe et al., 2010; Zhang et al., 2010). In three of these studies the ZFN sequence was placed under an inducible promoter in order to reduce the risk of ZFN-related toxicity. In this study, the EASE system eliminates the need for induction and restricts expression to the desired cell types, thereby avoiding ZFN toxicity in other tissues. An additional advantage of the EASE system is that every mutagenesis event obtained with the EASE system stemmed from an independent germinal event, generating a multiplicity of new alleles. It must also be stressed that the presented results were collected from randomly selected, EASE-QQR-ZFN-expressing lines. It is therefore likely that higher TGM rates can be reached with lines that are selected for having strong ZFN activity.

This work clearly demonstrates the potency of the EASE system in the enhancement of GT in Arabidopsis. The GT system described here is a modification of the system described by Shaked et al. (2005), including improved seed selection by using mRFP as a reporter protein in place of GFP. Additionally, as a result of RAD54 gene silencing, when placed under the 35S promoter (Lieberman and A. A. Levy, unpublished data), it was necessary to monitor the activity of genes by means of fluorescent reporter proteins. Although slightly lower than the rate described by Shaked et al. (2005), the enhancement of GT rate obtained with the EASE system was significant, and provides the advantage of lowering the risk of silencing, as observed with the 35S promoter.

Overall, this work contributes to the optimization of TGM and GT procedures in Arabidopsis, in pursuit of the development of efficient, routine and precise modification methods for the Arabidopsis genome. EASE-driven expression can be applied to a broad range of technologies involving the expression of proteins, such as ZFNs, TALENs or other homologous recombination-enhancing proteins, together with the delivery of a GT vector via floral dipping. Moreover, lessons learned on GT and TGM in Arabidopsis, can assist the development of similar approaches in crop plants.

Experimental Procedures

Plant growth and transformation

Plants were grown in a growth room with a controlled environment and maintained at 19°C, with 16 h of daylight. The transformation of A. thaliana (ecotype Columbia) was carried out by the floral-dip method as previously described (Clough and Bent, 1998). Cultures of A. tumefaciens were grown overnight in lysogeny broth (LB) at 28°C, until they reached a stationary phase. Before each transformation experiment, cells were concentrated to an OD600 of ≈1.8 in 5% sucrose, later supplemented with half-strength MS salts and Silwet L-77 (Momentive Performance Materials, http://www.momentive.com), added to a final concentration of 0.02% before dipping.

Constructs

All of the restriction enzymes used were purchased from New England Biolabs (http://www.neb.com).

GT vector.  The pGTmRFP gene-targeting vector was designed to contain the Cru box corresponding to the 1210-bp genomic sequence at the 5′ end of the Cruciferin3 gene. The Cru box lacks the ATG initiation codon as well as the 36 bp downstream of it. The mRFP reporter was fused in-frame, downstream of the Cru box, and immediately upstream of the transcription termination sequence of the octopine synthase gene (OCS). A 2492-bp fragment identical to the genomic region downstream to the sequence of the Cru box in the Cruciferin3 gene, of which 773 bp corresponded to the 3′ end of the Cruciferin3 gene (ciferin), followed by 1719 bp of downstream DNA, flanked the mRFP reporter. The cassette was cloned into the pART27 binary vector, which contains a gene that confers kanamycin resistance in plants (Appendix S1).

EASE:LhG4 vector.  In this expression vector, LhG4 is under the control of the 35S minimal promoter plus EASE, arranged in four tandem repeats of its 77-bp sequence (Yang et al., 2005). In addition, the vector contains an OCS terminator at the 3′ end of LhG4. This cassette was cloned into the pMLBArt binary vector, which contains a gene conferring glufosinate (BASTA) resistance in plants.

Op:mRFP vector.  In this vector, the mRFP ORF is under the control of the 35S minimal promoter plus the Lac operator arranged in seven tandem repeats (Moore et al., 1998). An OCS terminator was positioned at the 3′ end of mRFP. This cassette was cloned into the pART27 binary vector, which contains a gene that confers kanamycin resistance in plants.

Op:ScRAD54 vector.  In this vector, the ScRAD54 ORF is under the control of the 35S minimal promoter plus the Lac operator arranged in seven tandem repeats (Moore et al., 1998). In addition, an OCS terminator was positioned at the 3′ end of ScRAD54. This cassette was cloned into the pMLBArt binary vector, which contains a gene conferring glufosinate (BASTA) resistance in plants.

EASE:EGFP and EASE:QQR-ZFN vectors.  The LhG4 gene in the EASE:LhG4 cassette was replaced by sequences encoding either EGFP (Clontech, http://www.clontech.com) or QQR-ZFN (originally from a plasmid kindly provided by Dr Gary Drews, University of Utah) by subcloning with XhoI and XbaI restriction enzymes. The EASE:EGFP cassette was subcloned, using NotI, into the pMLBArt binary vector, which contains a gene conferring glufosinate (BASTA) resistance in plants. The EASE:QQR-ZFN cassette was subcloned into the pCambia1300 binary vector (Cambia, Brisbane, Australia), using NheI and NsiI, which cut the EASE:QQR-ZFN, and PstI and XbaI, which cut the pCambia1300. The final vector conferred hygromycin resistance in plants.

Confocal microscopy

Arabidopsis ovules were collected from ovaries of flowers at one stage before anthesis. Ovules were kept in double-distilled water and cellular localization was analyzed using a laser confocal microscope, Olympus IX81 FV1000 Spectral (Olympus, http://www.olympus.com), equipped with an objective lens FV1000 UPLAPO (40× oil NA: 1.35). mRFP was detected using a Diode laser (excitation, 559 nm; emission, 575–620 nm). 4′,6-Diamidino-2-phenylindole (DAPI) staining was detected using a Diode laser (excitation, 405 nm; emission, 415–435 nm). EGFP was detec-ted using an Argon laser (excitation, 488 nm; emission, 500–545 nm).

EASE:QQR-ZFN activity timing assay

A cassette including a mutated GUS gene with QQR-ZFN binding sites surrounding a stop codon, directly following the start ATG of the GUS gene, and conferring kanamycin resistance in plants – pRCS2[Kan][QQR-TS*::GUS] (Tovkach et al., 2009) (Figure 3a) – was kindly provided by Dr Tzvi Tzfira. This cassette was transformed into Arabidopsis plants, as described above, and plants were selected on 50 μg ml−1 kanamycin. Plants with a single locus T-DNA insertion were identified using Southern blotting, performed as described in Shaked et al. (2005), but with the following modifications: the genomic DNA was digested with PvuII and detection was carried out with a probe that detects the kanamycin resistance gene. The probe was prepared by PCR amplification with the following primers: nptII-Fwd, 5′-GGATTGCACGCAGGTTCTCC-3′; and nptII-Rev, 5′-TATTCGGCAAGCAGGCATCG-3′, using the pART27 binary vector as template. The single locus lines (lines Numbers 7 and 15) were transformed with Agrobacterium containing the pCambia1300-EASE-QQR-ZFN cassette. To monitor the timeframe of EASE:QQR-ZFN activity, seeds were collected from plants that transiently expressed EASE:QQR-ZFN and germinated in the presence of kanamycin. In all, approximately 15 000 T1 seedlings, which were subjected to X-Gluc staining, were generated. In addition, 18 plants derived from these same plant lines were selected on medium containing both kanamycin and hygromycin, to isolate those that were stably transformed by both the GUS mutant and the EASE:QQR-ZFN constructs. Seeds were collected from these plants and germinated on medium containing both kanamycin and hygromycin. Approximately 9000 T2 seedlings, resistant to both antibiotics, were subjected to X-Gluc staining. For staining, seeds were sown on half-strength MS-Agar plates with the appropriate antibiotics, incubated for 72 h at 4°C and then grown for 4 days at 26°C, under 16 h of daylight. The seedlings were then covered with Approximately 6 ml X-Gluc staining solution containing 1 mg ml−1 X-Gluc substrate (Gold Biotechnology, http://www.goldbio.com) dissolved in a solution of 50 mm sodium phosphate buffer (pH 7.0), 0.1% Triton X-100, 0.1% dimethylsulphoxide, 5 mm potassium ferricyanide and 5 mm potassium ferrocyanide trihydrate (Sigma-Aldrich, http://www.sigmaaldrich.com) and incubated for 24 h at 37°C. Samples were then placed in 70% ethanol for de-staining for another 48 h. In all of the experiments, a plate with seeds collected from GUS+ plants served as a positive control. Following complete de-staining, the seedlings were visually examined and then further analyzed under a Nikon SMZ1500 stereomicroscope (Nikon, http://www.nikon.com). For sequencing of the QQR-ZFN break site in the seedlings, an approximately 1-mm2 piece from a cotyledon or first leaf was placed in TerraTM-PCR Direct Polymerase Mix (Clontech) with primers surrounding the break site (35SF forward primer, 5′-GACCCTTCCTCTATATAAG-3′ and GUSr reverse primer 5′-GGGATAGTCTGCCAGTTC-3′). The PCR was carried out according to the manufacturer’s instructions, and resulted in a 680-bp product from the WT locus. The PCR fragments were sequenced at the Weizmann Institute DNA sequencing facility using the 35SF primer.

EASE:QQR-ZFN-targeted mutagenesis assay

To analyze the targeted mutagenesis rate of EASE:QQR-ZFN in our system, approximately 400 seeds from 18 different plants of line number 7, described above, that have the EASE-QQR-ZFN integrated into the genome, were selected on kanamycin and hygromycin. Three hundred and eighty seedlings were subjected to PCR using the TerraTM-PCR Direct Polymerase Mix (Clontech), as described above. PCR reaction samples (5 μl) were then restricted overnight at 37°C, using DdeI (New England Biolabs, http://www.neb.com). The restriction reactions were separated on 1.7% agarose gels and bands were visualized. A PCR product that showed an unrestricted band was sequenced to confirm that mutagenesis had occurred.

GT assay

Approximately 1000 WT plants and approximately 360 EASE>>ScRAD54 plants were transformed with the pGTmRFP vector by floral dipping (Zhang et al., 2006). Seeds were collected and red fluorescent seeds were selected under a Nikon SMZ1500 stereomicroscope adapted to the X-CITER 120PC Q light source system (Lumen Dynamics Group Inc, http://www.ldgi.com), which delivers a rich spectral fluorescence excitation and includes filter sets for Texas red (excitation 560/40, DM595 DCLP, emission 630/60). The frequency of pGTmRFP non-homologous integration was estimated from the transformation frequency: approximately 2000 seeds were plated from each GT experiment, and seedlings were selected for resistance to kanamycin. The GT rate was defined as the ratio of homologous to non-homologous integration. It was calculated as the number of fluorescent seeds that underwent validation as GT events (see below) divided by the total number of seeds that were transformed.

In order to confirm precise homologous recombination events, we performed PCR reactions on genomic DNA extracted from three or four young leaves of plants that derived from red fluorescent seeds, as previously described (Shaked et al., 2005). The following primers were used to confirm the 5′Cru3 GT junction: PRO_Cru_961_F 5′-ACTGAAGTCCCCATGCAAACC-3′ and mRFP_ ORF_REV 5′-CCTTGATGACGTCCTCGGAG-3′. The expected size of the PCR product from this reaction is 1400 bp. The following primers were used to confirm the 3′Cru3 GT junction: OCS_3′_Forw, 5′-AGGACCGGCATGCAAGCTAG-3′; and Cru_down_68981, 5′-GGC- TTTAATGCGTGTTTGGTCA-3′. The expected size of the PCR product from this reaction is 2612 bp.

Southern blotting was used to confirm precise GT and replacement of the WT Cruciferin3 allele. The assay was carried out as previously described, but with a 1062-bp Cifgs probe generated by PCR on WT Arabidopsis (ecotype Columbia), using the following primers: Cif_gs_F, 5′-TGACCAAACACGCATTAAAGCC-3′; and Cif_gs_R, 5′-CGAACGGTGATGGATCCACG-3′. The recognition location of the probe is depicted in Figure 6(a).

Acknowledgements

We would like to thank Ueli Grossniklauss and Nir Ohad for providing us with useful advice on egg cell-specific expression, and Tzvi Tzfira for fruitful discussions and his generosity with reagents. We thank Vladimir Kiss for help with the confocal microscopy, Michal Kenan-Eichler for construction of the new RFP GT cassette, and Michal Lieberman-Lazarovitch and members of the Levy lab for useful discussions. This work was supported by grants from BARD and by the EU-FP6 TAGIP and EU-FP7 RECBREED projects.

Ancillary