The Structural Features of Thousands of T-DNA Insertion Sites Are Consistent with a Double-Strand Break Repair-Based Insertion Mechanism

Abstract

Transformation by Agrobacterium tumefaciens, an important tool in modern plant research, involves the integration of T-DNA initially present on a plasmid in agrobacteria into the genome of plant cells. The process of attachment of the agrobacteria to plant cells and the transport of T-DNA into the cell and further to the nucleus has been well described. However, the exact mechanism of integration into the host's DNA is still unclear, although several models have been proposed. During confirmation of T-DNA insertion alleles from the GABI-Kat collection of Arabidopsis thaliana mutants, we have generated about 34,000 sequences from the junctions between inserted T-DNA and adjacent genome regions. Here, we describe the evaluation of this dataset with regard to existing models for T-DNA integration. The results suggest that integration into the plant genome is mainly mediated by the endogenous plant DNA repair machinery. The observed integration events showed characteristics highly similar to those of repair sites of double-strand breaks with respect to microhomology and deletion sizes. In addition, we describe unexpected integration events, such as large deletions and inversions at the integration site that are relevant for correct interpretation of results from T-DNA insertion mutants in reverse genetics experiments.

The complex architecture of plant transgene insertions

“…We did not recover any LB sequences at the chromosome/T-DNA junction, in line with literature reports that usually 73 -113 bp are missing from the LB sequence inwards [15,27]. Internal T-DNA sequence deletions were also seen at breakpoints within the insertion (Fig.…”

Enhancing the GABI-KatArabidopsis thalianaT-DNA Insertion Mutant Database by Incorporating Araport11 Annotation

“…The confirmation sequences obtained during the validation steps at GABI-Kat while confirming insertion predictions in the T2 generation are generally of much better quality. These data allow a more precise deduction of the exact insertion position (Kleinboelting et al 2015). The established method in SimpleSearch and other FST databases was to keep the insertion position predicted on the basis of the original FST.…”

“…Table 2 presents the increase in terms of data which have been added to SimpleSearch in the last 5 years. Of the 5,177 new FSTs, 4,281 were regular FSTs, 746 were FSTs derived from a confirmation sequence made for the second border of the insertion to determine both junctions between T-DNA and the genome (see Kleinboelting et al 2015), and the remaining ones were the FSTs generated from confirmation sequences as mentioned above. A list of all submitted FSTs along with the predicted insertion site is given in Supplemental Table 1.…”

“…An insertion is expected to be different from another one in the same line if the distance between the two predicted insertion positions is at least 20 kbp (Kleinboelting et al 2015). The gain of 6,653 predicted insertion alleles (from 88,580 (September 15, 2011) to 95,233 (August 15, 2016)) is in part due to data from the Ecker group (O’Malley et al in preparation).…”

“…The inserted T-DNA often disrupts the gene at the insertion site (Wang et al 1984, Peralta and Ream 1985, Gelvin 1998, Krysan et al 1999). While the exact mechanism of integration is still debated, a process mediated by plant repair pathways seems likely (Tzfira et al 2004, Gelvin 2010, Kleinboelting et al 2015). Due to the (largely) randomness of the insertion process with respect to the insertion position in the nuclear genome (Szabados et al 2002, Li et al 2006, Kim et al 2007), a large number of mutants needs to be generated in order to achieve knockout lines for most of the genes.…”