Efficient TALEN‐mediated gene editing in wheat

Plant genome editing is a major advance in the production of novel plant genotypes. There is, however, only a single previous report of applying this technology to hexaploid wheat using TALEN-mediated gene editing to produce heritable modifications (Wang et al., 2014). Here we describe highly efficient TALEN editing of a uidA transgene and a second endogenous gene, lr21Ψ, in bread wheat with efficiencies exceeding previous TALEN and CRISPR/Cas9 reports in this species. This article is protected by copyright. All rights reserved.

Plant genome editing is a major advance in the production of novel plant genotypes. There is, however, only a single previous report of applying this technology to hexaploid wheat using TALEN-mediated gene editing to produce heritable modifications (Wang et al., 2014). Here, we describe highly efficient TALEN editing of a uidA transgene and a second endogenous gene, lr21Ψ, in bread wheat with efficiencies exceeding most previous TALEN and CRISPR/Cas9 reports in this species.
A TALEN pair (Figure 1a) targeting the E. coli uidA gene was co-transformed into embryos from a wheat cultivar Fielder line segregating for a Ubi-uidA transgene. Twelve lines were produced containing both TALENs and at least one copy of uidA. T 0 genomic DNAs of these plants were restricted with BclI, as this enzyme cleaves the target site, and the target region was then PCR amplified ( Figure 1b). Products were amplified from three of the 12 (25%) DNAs (plants P21, P38 and P45), and Sanger sequencing identified deletions of 3, 13 and 4 bp, respectively, that destroyed the BclI site ( Figure 1b).
T 1 analysis showed that P38 and P45 were hemizygous for uidA and that only edited uidA alleles were transmitted to progeny (nine edited-uidA: six null and 10 edited-uidA: 5 null), consistent with an editing event subsequently dominating the majority of germinal tissue in each plant. In contrast, Mendelian inheritance occurred in P21 T 1 progeny (6 wt-uidA: 12 presumed heterozygotes (p-hets): six edited-uidA; X 2 P = 1.0), indicating the parent plant was uidA/uidA and a single edited allele predominated in germinal tissue. No GUS staining occurred in P38 and P45 T 1 plants containing 13-and 4-bp deletion alleles due to loss of uidA function, whereas P21 progeny homozygous for a 3-bp deletion allele showed GUS staining, presumably because the uidA ORF remained in-frame ( Figure 1c).
Most gene editing reports have produced targeted gene knockouts. Herein, we attempted to reactivate a pseudogene (lr21Ψ) present in Fielder wheat. The functional Lr21 gene (GenBank AH012974) encodes a nucleotide binding site leucine-rich repeat protein (NLR) that provides race-specific resistance to leaf rust disease caused by Puccinia triticina (Huang et al., 2009). lr21Ψ differs to Lr21 by 3 nonsynonymous SNPS (498 G/D, 854 M/I, 1055 R/S) and a single base deletion that destroys the gene ORF (Huang et al., 2009). Previously, a recombinant allele encoding Fielder lr21Ψ 5 0 regulatory sequences gave functional Lr21 resistance (Huang et al, 2009). Given this functional expression and near sequence identity between lr21Ψ and Lr21, we reasoned that restoring the lr21Ψ ORF by editing the 1-bp deletion site may reconstitute a functional resistance gene.
Forty T 0 plants were produced containing a TALEN pair ( Figure 1d) targeting the lr21Ψ 1-bp deletion site ( Figure 1e). DNAs from each plant were PCR amplified using primers flanking the lr21Ψ target site and amplicons MiSeq sequenced with an average of 2266 AE 800 (standard deviation) reads analysed per sample. Seventy-three different edited alleles were identified amongst T 0 amplicons with 71 encoding deletions (1-11 bp; Figure 1f) and two encoding small indels.
Substantial lr21Ψ editing occurred with 85% of T 0 plants (34/40) having between 15% and 100% of amplicons edited. On average, 55% AE 38% (standard deviation) of amplicons from each T 0 plant were edited. Individual allele frequencies within DNA samples ranged from 0.25% to 98.41% of amplicons, and most alleles (50/ 73; 68%) were common to at least two plants. The eight most common modified alleles with their respective frequencies are shown ( Figure 1e). Nucleotides in the middle of the target site were more commonly deleted than those adjacent to TALEN binding sites ( Figure 1g). Target site analysis of 3 Fielder control DNAs showed a very low error rate in this analysis with 0%, 0% and 0.12% of amplicons differing to the wild-type lr21Ψ sequence.
Two edited alleles inherited in T 1 progeny encoded restored lr21Ψ ORFs (Figure 1i). One, present both L14 and L20 T 1 families, encoded a 5-bp deletion ( Figure 1h) which restored the gene ORF, albeit with the loss of two amino acids ( Figure 1i). The second allele, also inherited by L20 progeny, had an 11-bp This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. deletion ( Figure 1h) and encoded an in-frame ORF with four codons deleted at the editing site (Figure 1h, i). However, Lr21 resistance to P. triticina was not recovered in seedlings homozygous for either of these alleles or amongst 443 T 1 seedlings from 10 other active lr21Ψ TALEN lines which potentially contained additional allelic variants (Figure 1k).
Eight potential lr21Ψ TALEN off-target sites are present in the Chinese Spring wheat genome, and two of these sites, corresponding to genes TraesCS1B02G002400 and TraesC-S1A02G006700, were successfully amplified from Fielder DNA. MiSeq sequencing of amplicons from all 40 T 0 DNAs identified two potential editing events, a 6-bp and a 1-bp deletion, at   ORFs as shown in panels (h) and (i). T 1 progeny of these plants were grown and genotyped for zygosity of alleles with restored ORFs (wt = wild lr21Ψ, p-het = presumed heterozygous, ed = homozygous for the edited allele), and these plants then challenged with P. triticina. Heterozygotes are described as presumed as plants producing PCR products with mixed sequencing traces could also be biallelic or chimeric. All seedlings were fully susceptible and indistinguishable to the Fielder control shown on the left. (l) Potential lr21Ψ TALEN off-target sites (OTS) amplified from the Fielder wheat genome. OTS1 and OST2 correspond to annotated wheat genes TraesCS1B02G002400 and TraesCS1A02G006700, respectively. Other off-target sites are present in the Chinese Spring genome sequence that could not be amplified from Fielder presumably due to sequence polymorphism existing between these two cultivars. TALEN binding sites are highlighted in blue on the lr21Ψ and OTS sequences. Mismatched nucleotides at the TALEN binding sites of OTS sequences are highlighted in red. Immediately beneath each OTS are variant sequences identified amongst 40 T 0 DNAs with frequencies indicated. TraesCS1B02G002400 at low frequency (0.025% and 0.04%) and a single 1-bp deletion allele at TraesCS1A02G006700 (0.18% of amplicons; Figure 1l). Little, if any, off-target editing therefore occurred given these allele frequencies are similar to the background observed in Fielder controls at the lr21Ψ site. It was of interest to compare the editing efficiencies of lr21Ψ TALENs in wheat protoplasts with stable transgenics, given protoplast assays are often used for editing studies. Fielder protoplasts were co-transformed with Ubi-YFP and the lr21Ψ TALEN pair. After 2 days of incubation, when 20% of protoplasts showed YFP expression, DNA was extracted and the lr21Ψ target site amplified and sequenced (4400 amplicons). A similar, although nonidentical, spectrum of editing events occurred at the lr21Ψ target when compared with transgenic wheat plants. Of the 28 alleles amplified from protoplasts, 21 (75%) were also present in wheat T 0 DNAs. Amongst the eight most frequent edited alleles, five were common to both protoplast and T 0 DNAs (compare Figure 1e and j). Similar editing events were therefore produced in protoplasts and transgenic plants using this TALEN pair.
These data show highly efficient wheat TALEN editing of a transgene (25%) and an endogenous gene (85%) in T 0 wheat plants and modified alleles having high heritability. Previously, the wheat Mlo gene was TALEN edited with 3.4%-6.0% efficiency in T 0 plants although with multiple homoeologous loci simultaneously modified in some plants and co-inherited in T 1 progeny (Wang et al., 2014). Both our and Wang's studies used maize polyubiquitin promoters for TALEN expression, so the large editing efficiency differences between studies suggest variation in TALEN target site accessibility. Alternatively, we used Agrobacterium transformation rather than biolistics which may cause significant TALEN expression differences. It is noteworthy that Wang saw higher editing in protoplasts compared with biolistic transgenics.
A variety of stable and transient CRISPR/Cas9 editing approaches have also been used in wheat with T 0 editing efficiencies usually around 1%-10%, (Howells et al., 2018 and references therein;Zhang et al., 2019;Kumar et al., 2019;Kelliher et al., 2019;Okada et al., 2019), which is low compared with rice . While CRISPR/Cas is technically simpler than TALENs and far more amenable to multiplex targeting, not all target sites are efficiently edited in wheat, likely due to sgRNA and genomic target site structural constraints (Yarrington et al., 2018;Graf et al., 2019). Potentially, this is problematic if an unamenable target site is critical to edit. TALEN editing therefore provides an efficient alternative, without a PAM sequence requirement, and using both platforms will maximize wheat editing opportunities.
While the lr21Ψ ORF was successfully restored, resistance gene function was not, possibly due to editing footprints. While this pseudogene reactivation attempt was unsuccessful, others may succeed if less constrained proteins are targeted where editing footprints may be tolerated. Potentially, pseudogene reactivation could be beneficial in introducing a functional allele in breeding programs, which is laborious if wild relatives or unimproved germplasm is the only other available source.