Allele exchange at the EPSPS locus confers glyphosate tolerance in cassava

Abstract

Effective weed control can protect yields of cassava (Manihot esculenta) storage roots. Farmers could benefit from using herbicide with a tolerant cultivar. We applied traditional transgenesis and gene editing to generate robust glyphosate tolerance in cassava. By comparing promoters regulating expression of transformed 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS) genes with various paired amino acid substitutions, we found that strong constitutive expression is required to achieve glyphosate tolerance during in vitro selection and in whole cassava plants. Using strategies that exploit homologous recombination (HR) and nonhomologous end-joining (NHEJ) DNA repair pathways, we precisely introduced the best-performing allele into the cassava genome, simultaneously creating a promoter swap and dual amino acid substitutions at the endogenous EPSPS locus. Primary EPSPS-edited plants were phenotypically normal, tolerant to high doses of glyphosate, with some free of detectable T-DNA integrations. Our methods demonstrate an editing strategy for creating glyphosate tolerance in crop plants and demonstrate the potential of gene editing for further improvement of cassava.

Keywords: cassava; gene editing; gene replacement; herbicide tolerance.

Figure 1

Glyphosate tolerance of cassava plants transformed with various EPSPS expression cassettes. (a) Quantification of shikimate accumulation in a colorimetric assay for the effect of glyphosate on EPSPS function in leaf discs derived from independent transformations with each of the EPSPS gene models. RGB images were converted to LAB space and average intensity of the ‘A’ channel was quantified for each sample. (b) Bar graphs indicating herbicide injury score of plants derived from independent transformation events of each gene model 23 days after application of 50 mg active ingredient (AI) of glyphosate isopropylamine salt to each plant. Impact of herbicide application was assessed three times per week on a scale of 1–7 for damage to the shoots where 1 = no damage to 7 = plant death. (c, d) Photographs showing representative phenotypes of plants derived from three independent transformations of the gene models expressing TIPA (c) and GAAT (d) mutant EPSPS enzymes from the 2x35S promoter 21 days after application of 50 mg AI of glyphosate isopropylamine salt to each plant. For all figures, numbers below bar graphs and photographs indicate the identity of independent transgenic events.

Figure 2

EPSPS editing strategies and molecular characterization of recovered plants. (a) Scaled map of the EPSPS locus, repair templates configured for exploitation of the HR pathway alone or for both the HR and NHEJ pathways, and the T‐DNA structures. Cas9 was expressed from the 2x35S promoter, while sgRNA #7 and sgRNA #11 were expressed from the Arabidopsis U6 and 7SL promoters, respectively. GVR = geminivirus replicon (b) PCR characterization of glyphosate‐resistant plants derived from the NHEJ or HR repair template on standard T‐DNA (H055), or the HR repair template on standard T‐DNA (H056) and on the GVR (H060). Gene targeting was assayed with the TC414/VLP476 primers under conditions that would not amplify the WT allele. Heterozygosity of the editing events was confirmed in all plants by detection of the WT allele with primers VLP51/VLP23. Identification of all amplicons was verified by Sanger sequencing (not shown).

Figure 3

Glyphosate tolerance of plants with edited EPSPS alleles. (a) Quantification of shikimate accumulation. (b) Graph indicating average herbicide injury score of plants derived from EPSPS editing events with the H055 and H056 vectors. (c) Photograph showing representative phenotypes of plants derived from EPSPS editing events with the H055 and H056 vectors 21 days after application of 50 mg AI of glyphosate isopropylamine salt. Data shown in a, b and c were generated as described in Figure 1.