Zinc finger nuclease-mediated precision genome editing of an endogenous gene in hexaploid bread wheat (Triticum aestivum) using a DNA repair template

Characterization of ZFNs targeting endogenous AHAS genes in wheat

PCR amplification, cloning and Sanger sequencing of the AHAS gene homoeologous (AHAS-6A, AHAS-6B and AHAS-6D) in the bread wheat genotype Bobwhite MBP26RH revealed about 97% and 98% identity at the nucleotide and protein levels, respectively. In silico ZFN design for the AHAS gene region encoding the S630 residue (hereafter referred to as S653 to maintain consistency with Arabidopsis nomenclature) in the protein sequence generated seven ZFNs predicted to cleave all three homoeologous gene copies. Three of the ZFN cut sites were located upstream of the S653 codon, while the remainder were positioned downstream. The ZFN cut site designed to be closest to the target was predicted to create a DSB nine-bp upstream of the S653 residue (Figure 1a).

The efficacy of each ZFN was tested against their endogenous targets using transient assays based on PEG-mediated DNA delivery to wheat mesophyll protoplasts. ZFN efficacy was assessed by measuring whether they created small insertion and deletion (indel) mutations by nonhomologous end joining (NHEJ), which has previously been shown to provide a reliable indicator for ZFN activity at endogenous loci (Shukla et al., 2009; Townsend et al., 2009). Analysis of the proportion of sequence reads showing indel mutations revealed that all ZFNs tested had activity in protoplasts and were able to create targeted DSBs in each of the homoeologous copies of the AHAS gene. Some ZFNs showed differential cleavage activity across AHAS homoeologous, while others had similar activity (Figure 1b). Examination of the frequency and pattern of indel mutations for each ZFN allowed the size of the overhangs generated by FokI cleavage at the ZFN binding site to be accurately determined. The most common deletion for the ZFNs was four bp, resulting from trimming of the single-stranded ends of the overhangs (Figure 1c).

Efficiencies for NHEJ-directed gene editing in wheat

To examine the efficiency for ZFN-mediated NHEJ-directed genome editing in wheat, two strategies for allele editing via a supplied DNA repair template (hereafter called a donor) were tested using transient assays based on particle bombardment DNA delivery to scutella of immature zygotic embryos: allele insertion and allele replacement. Our choice of tissue for transformation experiments was based on the current inability to reliably regenerate plants from wheat mesophyll protoplasts, making it necessary to use the scutella system to recover stable transformation events.

The strategy for allele insertion – the introduction of new DNA in-frame with the existing gene – used a single ZFN to induce a DSB located nine-bp upstream of the S653 codon in the AHAS homoeologous. The strategy for allele replacement – the seamless replacement of endogenous AHAS sequence with new sequence – used a pair of ZFNs to induce DSBs either side of the S653 codon. These ZFNs induced DSBs positioned 9- and 74-bp upstream and downstream of the S653 codon, respectively. Both strategies employed the codelivery of a 41-bp double-stranded linear donor DNA fragment to be inserted into ZFN-induced DSBs in the three homoeologous copies of the endogenous AHAS gene. As FokI digestion results in the generation of 5′ overhangs, each donor repair template was produced with specific 5′ overhangs to facilitate error-free ligation of the donor into the DSB created by the ZFN(s). Donor DAS000152 was designed to provide ligation overhangs compatible with those generated by cleavage of the endogenous AHAS genes by ZFN DAB109350 and to result in insertion of the 41-bp donor fragment at the site of the endogenous DSB (Figure 2a). Donor DAS000149 had ligation overhangs compatible with those generated by dual cleavage of the endogenous AHAS genes by ZFNs DAB109350 and DAB109360 and was designed to replace the intervening endogenous AHAS sequence contained between the two induced DSBs (Figure 2b).

Deep sequencing of amplicons generated by PCR amplification with primers flanking the entire target site in the endogenous AHAS genes was used to assess gene-editing outcomes in scutella cotransformed with the donor DNA and ZFN encoding plasmid(s), or with donor molecule only.

As particle bombardment-mediated methods deliver DNA to only a small number of scutella cells, the vast majority of cells are expected to contain unmodified AHAS gene sequences. We therefore assessed editing frequency by calculating the percentage of sequence reads showing evidence for integration of the donor DNA at the targeted DSB as a proportion of the total number of reads showing evidence of ZFN-mediated mutations. We also determined the proportion of reads showing evidence for perfect and imperfect donor integration, where perfect integration denotes seamless donor integration at the endogenous DSB and imperfect integration indicates the additional presence of indels caused by NHEJ-associated exonuclease activity at the repair junctions.

About 0.02% of sequence reads generated from scutella cotransformed with ZFN DAB109350 and donor DAS000152 showed evidence for ZFN-induced indel mutations or donor integration. About 36% of these reads showed evidence for donor integration, while the remainder had indel mutations. The majority (86%) of reads with donor integration had perfect gene editing. The frequency of observed gene-editing outcomes was similar across the A, B and D genomes. No reads with donor integration were observed for scutella transformed only with donor DAB000152 (Table 1). Of the sequence reads generated for scutella cotransformed with ZFNs DAB109350 and DAB109360 and donor DAS000149, about 0.006% showed evidence for ZFN-induced indels and donor integration. About 6% of these reads showed evidence for donor integration, while the remainder had indel mutations. Twenty-two (219/1010) per cent of the reads containing indels showed evidence for cleavage at both ZFN sites and loss of the intervening sequence between the DSBs; the remainder (78%; 791/1010) had an indel corresponding to the creation of a DSB by only one ZFN. The vast majority (99%) of reads with donor integration had perfect gene editing. Again, the frequency of observed gene-editing outcomes was similar across the three wheat subgenomes. No reads with donor integration were observed for scutella transformed only with donor DAS000149 (Table 1).

Generation and characterization of wheat plants with AHAS(S653N) via direct selection

To demonstrate ZFN-mediated NHEJ-directed editing of an endogenous gene to create a selectable trait, the AHAS genes in wheat were targeted for the introduction of the S653N mutation known to confer resistance to imidazolinone class herbicides. Scutella of immature zygotic wheat embryos were cotransformed with ZFN encoding plasmid and a linear double-stranded DNA donor using particle bombardment. Two sets of cotransformation were performed. The first delivered ZFN DAB109350 and donor DAS000267, while the second delivered ZFNs DAB109350 and DAB109360 and donor DAS000268 (Figure 2). Four selection schedules were used to recover stably transformed T₀ wheat plants having resistance to imazamox, an imidazolinone class herbicide (Table 2).

The double-stranded 95-bp donor DAS000267 (Figure 2) was designed to integrate via the NHEJ pathway at a DSB created in an AHAS homoeolog by ZFN DAB109350 and comprised two parts. The 5′ end contained sequence homologous to the endogenous AHAS gene encoded in the D genome, starting from the target ZFN cleavage site and ending at the AHAS stop codon but containing six intentional base changes: two mutations encoded the S653N mutation (AGC->AAT) and four synonymous base changes to prevent ZFN recognition and cleavage of the integrated donor. The 3′ end of the donor contained a unique 28-bp sequence that could be used for diagnostic PCR to detect the integrated donor in regenerated wheat plants. The donor molecule had protruding 5′ ends to provide ligation overhangs compatible with those produced by ZFN DAB109350.

The double-stranded 79-bp donor DAS000268 (Figure 2) was designed to replace the intervening sequence between the cleavage sites in the endogenous AHAS genes for ZFNs DAB109350 and DAB109360. The donor contained sequence homologous to the endogenous AHAS gene encoded in the D genome, starting from the cleavage site for ZFN DAB109350 and ending at the cleavage site for ZFN DAB109360 but contained ten intentional base changes. Six base changes were located at the 5′ end of the donor: two encoded the S653N mutation (AGC->AAT) and four encoded synonymous mutations. Four base changes were located at the 3′ end of the donor and positioned in noncoding sequence. The additional base changes were again used to prevent ZFN recognition and cleavage of integrated donor. The donor molecule had protruding 5′ ends to provide ligation overhangs compatible with those generated by the pair of ZFNs.

Cobombardment of about 4000 scutella with ZFN DAB109350 and donor DAS000267 for each imazamox selection strategy resulted in the regeneration of 12 wheat seedlings, with 2–5 plants recovered from each strategy (Table 3). PCR amplification across the target sites in the endogenous AHAS genes confirmed allele insertion via stable donor integration at the target site in at least one of the endogenous AHAS genes in each of the recovered wheat plants (Figure 3).

For deliveries with ZFNs DAB109350 and DAB109360 and donor DAS000268, about 2000 scutella were cobombarded for each imazamox selection strategy. Two wheat seedlings were regenerated from the second selection strategy (Table 3), both of which were confirmed by PCR analysis to have stable donor integration in at least one copy of the endogenous AHAS genes. However, the PCR results for both wheat seedlings suggested the intervening endogenous sequence between the two ZFN binding sites had not been excised, as a larger-than-expected PCR fragment was observed (Figure 3). Excision of intervening endogenous AHAS sequence was expected to result in the amplification of only wild-type fragment sizes.

To characterize the subgenomic location and gene-editing outcomes in the 14 regenerated T₀ wheat seedlings, the PCR products derived from amplification across the target sites in the AHAS gene homoeologous were cloned and Sanger sequenced. Up to 120 independent plasmid clones from each T₀ plant were sequenced to ensure each AHAS homoeoallele was characterized. A consensus sequence for each allele was generated and used to determine the subgenomic origin and sequence for each edited allele.

Targeted allele insertion at the endogenous AHAS loci was detected for 11 of the 12 regenerated T₀ wheat plants that were transformed using ZFN DAB109350 and donor DAS000267 (Table 4). The remaining plant (Event 8) contained only wild-type alleles and was considered an escape from selection. The 11 insert-positive plants all contained at least one allele spliced gene that was predicted to produce a functional AHAS protein containing the S653N mutation, and explained the observed resistance of the wheat plant to imazamox. The 11 plants showed a range of gene-editing outcomes including (i) perfect monoallelic allele insertion in the A, B or D genome, for example Events 6, 10 and 11; (ii) simultaneous allele insertion in multiple subgenomes, for example Events 1, 4 and 9; (iii) perfect biallelic allele insertion, for example Event 9; and (iv) different biallelic editing outcomes at the same locus, for example Events 1, 4 and 12. Examples of T₀ plants with perfect and imperfect allele insertion are shown in Figure 4.

Similarly, both of the regenerated wheat plants that were transformed using ZFNs DAB109350 and DAB109360 and donor DAS000268 (Table 4) had targeted gene editing. One plant had monoallelic editing in the A and D genomes, while the other plant had monoallelic editing in the A genome. Neither plant had the expected allele replacement but rather showed allele insertion via donor integration.

Generation and characterization of wheat plants with AHAS(S653N) via indirect selection

The previous experiments demonstrated that direct gene editing to confer a selectable herbicide resistance trait in wheat was feasible across multiple alleles. To demonstrate the feasibility for ZFN-mediated NHEJ-directed editing using a supplied DNA repair template at a nonselectable locus, scutella from immature zygotic wheat embryos were cotransformed with ZFN DAB109350, donor DAS000267 and a plasmid encoding a wheat codon optimized version of the phosphinothricin acetyl transferase (PAT) gene. This allowed the selection and regeneration of stably transformed wheat plants using the selection agent phosphinothricin.

Cobombardment of 2320 embryos led to the regeneration of 170 stably transformed wheat plants selected using phosphinothricin. PCR performed across the target sites in the endogenous AHAS genes confirmed two (1.2%) of the transformed wheat plants had integrated donor at one or more of the AHAS loci. Digital droplet PCR performed to detect ZFN-induced indels identified an additional five (2.9%) T₀ plants with modified AHAS alleles. Clones of the two wheat plants with integrated donor were tested for their ability to produce roots when grown on rooting media containing imazamox and were confirmed to be resistant to the AHAS inhibitor.

Characterization of the subgenomic location and nature of gene modifications in the seven wheat plants by sequencing of PCR products derived from amplification across the AHAS target sites confirmed that the two wheat plants showing imazamox resistance had at least one functional gene copy encoding the AHAS (S653N) protein (Table 5). Of the remaining plants, four had NHEJ-derived indels in one of the alleles in the A genome, and one had indels in one allele in the each of the B and D genomes (Table 5).

Mendelian inheritance of AHAS gene edits

To investigate whether the ZFN-induced gene edits and indels could be stably transmitted to the next generation, the T₀ wheat seedlings regenerated from the direct and indirect selection transformation strategies were allowed to self-pollinate and their T₁ progenies were analysed. Digital droplet PCR assays designed to measure the number of wild-type, indel and donor-integrated AHAS alleles present in each plant showed the expected Mendelian inheritance in the T₁ generation for each selfed T₀ plant. These results were further confirmed by sequencing the PCR products derived from amplification across the target sites in the AHAS gene homoeologous for T₁ plants derived from T₀ plants having more than one AHAS gene edit or indel (Table S2). PCR analysis for the presence of integrated ZFN expression vector was not performed.

To confirm stable inheritance of herbicide resistance, T₁ seed from each T₀ plant with gene editing resulting in a functional AHAS(S653N) protein was germinated in tissue culture, genotyped and transferred to rooting media containing imazamox after removing the seedlings roots. After 2 weeks of growth on rooting media, T₁ plantlets having at least one AHAS(S653N) allele grew long roots and had normal phenotypes. Plantlets lacking any AHAS(S653N) alleles had very short roots and became chlorotic, a phenotype also observed for control wild-type plants (data not shown).

By testing T₁ seedlings derived from wheat plants generated via the indirect selection transformation strategy with a digital droplet PCR assay designed to detect the presence of the PAT selectable marker, we were also able to identify T₁ plants that carried AHAS gene edits and indel mutations but lacked the PAT selectable marker. These results confirmed that selectable marker-free plants carrying only the desired DNA sequence changes in a nonselected gene can be obtained through genetic segregation.

AHAS gene sequencing

PCR primers were designed (Fwd 5′ AYC AGA TGT GGG CGG CTC AGT AT and Rvs 5′ GGG ATA TGT AGG ACA AGA AAC TTG CAT GA) using published genomic sequences from GenBank (Accession Numbers AY210405.1, AY210407.1, AY210406.1, AY210408.1, FJ997628.1, FJ997629.1, FJ997631.1, FJ997630.1, FJ997627.1, AY273827.1) to amplify the homoeologous copies of the AHAS genes from Triticum aestivum genotype Bobwhite MPB26RH, the donor material used for wheat transformation. The PCR products were purified and cloned using pGEM-T Easy Vector (Promega) into E. coli JM109 cells. Plasmid DNA was extracted using a DNAeasy Plasmid DNA Purification Kit (Qiagen) and Sanger sequenced using BigDye v3.1 chemistry (Applied Biosystems) on an ABI3730XL automated capillary electrophoresis platform. Sequence analysis was performed using Sequencher software (GeneCodes).

ZFN design and construct assembly

Zinc finger proteins directed against the amplified DNA sequences for the homoeologous copies of the AHAS genes were designed as described in Urnov et al. (2005). Target binding sites for the designed zinc fingers are shown in Table S1. The AHAS targeting zinc finger designs were incorporated into zinc finger expression vectors encoding a protein having at least one finger with a CCHC structure (see US Patent Publication No. 2008/0182332 for further detail). In particular, the last finger in each protein had a CCHC backbone for the recognition helix. The noncanonical zinc finger-encoding sequences were fused to the nuclease domain of the type IIS restriction enzyme FokI (corresponding to amino acids 384–579 of the sequence of Wah et al., 1998) via a four amino acid ZC linker and an opaque-2 nuclear localization signal derived from Zea mays to form zinc finger nucleases. Expression of the ZFN proteins was driven by a constitutively expressed ubiquitin gene (Ubi-1) promoter, including the 5′ untranslated region, from Z. mays (Toki et al., 1992) and flanked by the 3′ UTR (comprising the transcriptional terminator and polyadenylation site) from a peroxidase (Per5) gene from Z. mays (US Patent Publication No. 2004/0158887). The self-hydrolysing 2A encoding nucleotide sequence from Thosea asigna virus (Szymczak et al., 2004) was added between the two ZFN monomer protein coding sequences generating a fusion gene that contains both monomers. Plasmid vectors were assembled using In-Fusion Advantage Technology (Clontech). Plasmid DNA for each ZFN construct for delivery to wheat cells was prepared from cultures of E. coli using the Pure Yield Plasmid Maxiprep System (Promega), following the supplier's instructions.

Donor synthesis

Each DNA strand of the donor templates DAS000149, DAS000152, DAS000267 and DAS000268 was synthesized as gBlock gene fragments (Integrated DNA Technologies). Double-stranded donor was prepared by mixing equimolar amounts of each DNA strand, heating at 95 °C for 1 min and cooling to room temperature.

Protoplast transient assays for assessing ZFN efficacy and specificity

Mesophyll protoplasts from the wheat genotype Bobwhite MPB26RH were prepared for transfection using polyethylene glycol (PEG)-mediated DNA delivery as follows.

Etiolated leaf material harvested from seed germinated and grown in the dark at 24 °C for 21–28 days was submersed in about 10 mL of leaf digest mix (0.6 m mannitol, 10 mm MES, 1.5% w/v cellulase R10, 0.3% w/v macerozyme, 1 mm CaCl₂, 1% bovine serum albumin, 0.025% v/v pluronic acid, 5 mm β-mercaptoethanol, pH 5.7) and sliced transversely into 1- to 2-mm segments using a scalpel blade. Additional leaf digest mix was added to a final volume of 10 mL per gram fresh weight of leaf material and subjected to vacuum (20″ Hg) pressure for 30 min, followed by incubation at 28 °C with rotational shaking at 30 r.p.m. for 4 h. Mesophyll protoplasts released from the leaf material were collected by passing the digestion suspension through a 100-μm sieve, followed by 70- and 40-μm sieves. The digested leaf material was washed three times with 10 mL wash buffer (20 mm KCl, 4 mm MES, 0.6 m mannitol, pH 5.6) to maximize protoplast yield. The filtered protoplasts were pelleted by centrifugation at 70 g and 15 °C for 10 min and the supernatant removed. The protoplast pellet was washed once in 7 mL of wash buffer and finally resuspended in 1 mL wash buffer. The yield of mesophyll protoplasts was estimated using a Neubauer haemocytometer. Evans Blue stain was used to determine the proportion of live cells.

PEG-mediated DNA transfection was performed by pelleting about 1 million protoplasts by centrifugation at 70 g and 15 °C for 10 min. The protoplasts were resuspended in 600 μL of wash buffer containing 70 μg of plasmid DNA, comprising 30 μg of a ZFN construct and 40 μg of a reporter construct for expression of red fluorescent protein (DsRed). An equal volume of 40% PEG solution (40% w/v PEG 4000, 0.2 m mannitol, 0.1 m Ca(NO₃)₂, pH 5.6) was slowly added to the protoplast suspension with simultaneous mixing by gentle rotation of the tube. The protoplast suspension was allowed to incubate for 15 min at room temperature occasional gentle mixing. Six millilitres of wash buffer was then slowly added to the protoplast suspension, with simultaneous gentle mixing to maintain a homogenous suspension. The protoplasts were pelleted by centrifugation at 70 g and 15 °C for 10 min and the supernatant removed. The protoplast pellet was resuspended in 2 mL wash buffer before an additional 7 mL of wash buffer was added. Centrifugation was repeated to pellet to the protoplasts and the supernatant removed. The protoplast pellet was resuspended in 2 mL Qiao's media (0.44% w/v MS plus vitamins, 3 mm MES, 0.0001% w/v 2,4-D, 0.6 m glucose, pH 5.7) and transferred to a sterile 3-cm Petri dish before incubation in the dark for 24 °C for 72 h. The transfection efficiency was determined using FACS (BD-Biosciences), as the proportion of protoplasts having red fluorescent protein expression.

Genomic DNA was extracted from the transfected mesophyll protoplasts using the DNAeasy Plant DNA Extraction Mini Kit (Qiagen) following the manufacturer's instructions, except that tissue disruption was not required and the protoplasts were added directly to the lysis buffer. The DNA region surrounding the ZFN target site was amplified using two rounds of PCR to enable incorporation of amplicon-specific barcodes for Illumina sequencing. The primers for the first PCR round were designed to amplify the AHAS homoeoloci and to include nucleotide variants that distinguished between AHAS gene copies originating from the different wheat subgenomes. These primers were tagged at their 5′ end with sequences to enable Illumina barcode indexing in the subsequent PCR step (AHAS.F1 5′ aca ctc ttt ccc tac acg acg ctc ttc cga tct GAG ACC CCA GGG CCA TAC TTG; AHAS.R3 5′ gtg act gga gtt cag acg tgt gct ctt ccg atc tCA AGC AAA CTA GAA AAC GCA TGG, where lower case indicates Illumina sequences). The second round of PCR was performed using Illumina's commercially available indexing primers. The resulting amplicons were purified using Agencourt Ampure XP magnetic beads (Beckman-Coulter) with a DNA-to-bead ration of 1:1.7. The purified amplicons were titrated for sequencing by Illumina short read technology using a PCR-based library quantification kit (KAPA) according the manufacturer's instructions (KAPA Biosystems). The samples were sequenced on an Illumina GAII_x or HiSeq2000 instrument (Illumina) according to the manufacturer's instructions.

Zinc finger nuclease efficacy was determined as the proportion of sequence reads arising from each AHAS homoeolocus that contained NHEJ-derived indels. The observed efficacy was expressed in parts per thousand (ppk) activity after normalizing for protoplast transfection efficiency. Only reads with >1-nucleotide indel within a 10-bp window centred over the expected site of cleavage were classified as NHEJ mutations. The frequency distribution of indel sizes was determined from the most prevalent NHEJ mutations arising within the expected cleavage site, and the predominant overhang generated by each ZFN determined from the most common mutation. The ranking order of the ZFNs by their observed cleavage efficiency was used to select ZFNs with the best cleavage activity for subsequent experiments.

Scutella transient assays for assessing NHEJ-directed gene-editing efficiencies

Twenty scutella of immature zygotic wheat embryos from the donor genotype Bobwhite MPB26RH were cobombarded with each ZFN vector and donor template combination. Briefly, for allele insertion, DNA for donor DAS000267 (or DAS000152) was mixed at a 5:1 molar ratio with plasmid DNA for ZFN DAB109350, and for allele replacement, DNA for donor DAS000268 (or DAS000149) was mixed at a 5:1:1 molar ratio with plasmid DNA for ZFNs DAB109350 and DAB109360. DNA was precipitated onto gold particles for biolistic-mediated DNA delivery as follows: 50 μL of 40 mg/mL of 0.6 μm colloidal gold particles (Bio-Rad) was added to 5 μL of DNA solution (1 μg DNA/μL), followed by sequential addition of 50 μL of 2.5 M CaCl₂ and 20 μL of 0.1 m spermidine. The mixture was vortexed for 1 min. The DNA-coated gold particles were pelleted by centrifugation and the supernatant removed. The gold particles were washed once in 1 mL of absolute ethanol and resuspended in 100 μL of absolute ethanol. Immediately following resuspension, 10 μL of DNA-coated gold particles was placed centrally onto a macrocarrier membrane and allowed to air dry. Particle bombardment was performed using the PDS-100/HE Particle Gun Delivery System (Bio-Rad) with a 900 psi rupture disc and the following settings: gap 2.5 cm, stopping plate aperture 0.8 cm, target distance 6.0 cm, vacuum 91.4–94.8 kPa, vacuum flow rate 5.0 and vent flow rate 4.5. Postbombardment the scutella were cultured on callus induction media in the dark at 26 °C for 7 days. Genomic DNA was isolated from the 20 scutella using the DNAeasy Plant DNA Extraction Mini Kit (Qiagen) according to the manufacturer's instructions. The DNA region surrounding the ZFN target site was amplified using two rounds of PCR and subjected to deep amplicon sequencing, as described above. The frequency of ZFN-induced gene editing was assessed as the proportion of sequence reads arising from each AHAS homoeolocus having donor integration. The proportion of sequence reads having NHEJ-derived mutation was calculated as described above.

Wheat transformation

Particle bombardment transformation of scutella from immature zygotic embryos from the donor genotype Bobwhite MPB26RH was used to generate gene-edited wheat plants as described above. Postbombardment the scutella were cultured in the dark on callus induction medium at 24 °C for 2 weeks. The resultant calli were subcultured once onto fresh callus induction medium and kept in the same conditions for a further 2 weeks. Somatic embryogenic callus (SEC) was transferred onto plant regeneration medium and cultured for 2 weeks at 24 °C under a 16/8-h (light/dark) hour photoperiod in a growth room. Regenerated plantlets were transferred onto rooting medium and cultured under the same conditions for 2–3 weeks. Chemical selection agent (imazamox or phosphinothricin) was added to each culture medium as described in the Results section. To increase stringency for the selection of regenerated plants having the S653N mutation, the roots of regenerated plants were removed and the plants were again subcultured on rooting media containing imazamox under the same conditions. Plantlets rooting after a second round of rooting media were transferred to compost-based potting mix and grown under glasshouse containment conditions. T₁ seed was harvested from individual plants, following bagging of individual spikes prior to flowering to prevent out-crossing.

Characterization of AHAS-edited wheat plants and Mendelian inheritance

PCR was used to characterize the subgenomic location, zygosity and outcomes for ZFN-mediated gene editing in the transformed wheat plants. The DNA region surrounding the ZFN target sites were amplified using the AHAS.F1/AHAS.R3 primer pair (see above) but without the 5′ Illumina barcode indexing sequences. The resulting amplicons were separated and visualized on 2% agarose gel, or cloned into pGEM-T plasmid vector (Promega) and Sanger sequenced as described above. Up to 120 independent plasmid clones were sequenced to ensure each AHAS homoeoallele was characterized. Sequencher^® software (GeneCodes) was used to generate a consensus sequence for each AHAS homoeoallele and to determine its subgenomic origin.

The Bio-Rad QX100™ Droplet Digital PCR (ddPCR) system (Bio-Rad) was used to quantitatively assess the number of wild-type, NHEJ indel and donor-integrated AHAS alleles present in the transformed wheat plants and their offspring. ddPCR increases the sensitivity and dynamic range of quantitative PCR by separating each reaction into a large number of noninteracting partitions (Mazaika and Homsy, 2014). A TaqMan™ assay was performed using two ZEN™ Double Quenched hydrolysis probes (Integrated DNA Technologies) that were positioned within the same amplicon. The first hydrolysis probe (5′ FAM-CCA GGG CCA TAC TTG TTG GAT ATC) detected the presence of any AHAS homoeoallele, while the second probe (5′ HEX-TTG GAA TCA TAG GCA GCA CGT) was positioned over the ZFN cut site and designed to discriminate between wild-type, NHEJ mutations and donor integration. The forward and reverse primer pair for this dual TaqMan assay were 5′ GTG ACG AAG AAG AGC GAA G and 5′ GCC ATC ACC CTC CATG AT, respectively. PCR droplets were classified as wild-type, NHEJ-derived indel or S653N edited based on the ratio of HEX:FAM using QuantaSoft software (Bio-Rad). To ensure correct droplet classification, control assays were performed using synthetic DNA templates corresponding to the expected wild-type, NHEJ indel and S653N donor-integrated alleles. A duplex TaqMan digital droplet PCR assay was used to detect the presence of the PAT selectable marker. This assay comprised a primer pair (Fwd 5′ GAT CAA TTG TTC GAC AGT GAA GGT, Rvs 5′ TCA AAA TGA CTG GCC TAA TCA GAT AA) and ZEN™ Double Quenched hydrolysis probe (5′ HEX-TCC GCT GGG CAT AAT TCC AAT GAG C) to amplify the known endogenous single copy gene, EST BE497897, from chromosome 6B of hexaploid wheat (Qi et al., 2004) and a primer pair (Fwd 5′ CCC GCC CCT CTC CTC TTT C, Rvs 5′ CTC CCG CGC ACC GAT CTG) and ZEN™ Double Quenched hydrolysis probe (5′ FAM-AAG CCG CCT CTC GCC CAC CCA) targeting a region of the Oryza sativa Actin (Act1) promoter present in a wheat codon optimized version of the phosphinothricin acetyl transferase gene (PAT). The number of copies of the exogenous PAT gene was determined using the CNV functionality in the QuantaSoft software.