Genetic dissection of mutagenic repair and T-DNA capture at CRISPR-induced DNA breaks in Arabidopsis thaliana

A practical and powerful approach for genome editing in plants is delivery of CRISPR reagents via Agrobacterium tumefaciens transformation. The double-strand break (DSB)-inducing enzyme is expressed from a transferred segment of bacterial DNA, the T-DNA, which upon transformation integrates at random locations into the host genome or is captured at the self-inflicted DSB site. To develop efficient strategies for precise genome editing, it is thus important to define the mechanisms that repair CRISPR-induced DSBs, as well as those that govern random and targeted integration of T-DNA. In this study, we present a detailed and comprehensive genetic analysis of Cas9-induced DSB repair and T-DNA capture in the model plant Arabidopsis thaliana. We found that classical nonhomologous end joining (cNHEJ) and polymerase theta-mediated end joining (TMEJ) are both, and in part redundantly, acting on CRISPR-induced DSBs to produce very different mutational outcomes. We used newly developed CISGUIDE technology to establish that 8% of mutant alleles have captured T-DNA at the induced break site. In addition, we find T-DNA shards within genomic DSB repair sites indicative of frequent temporary interactions during TMEJ. Analysis of thousands of plant genome-T-DNA junctions, followed up by genetic dissection, further reveals that TMEJ is responsible for attaching the 3' end of T-DNA to a CRISPR-induced DSB, while the 5' end can be attached via TMEJ as well as cNHEJ. By identifying the mechanisms that act to connect recombinogenic ends of DNA molecules at chromosomal breaks, and quantifying their contributions, our study supports the development of tailor-made strategies toward predictable engineering of crop plants.

Distinct mechanisms for genomic attachment of the 5' and 3' ends of Agrobacterium T-DNA in plants

Agrobacterium tumefaciens, a pathogenic bacterium capable of transforming plants through horizontal gene transfer, is nowadays the preferred vector for plant genetic engineering. The vehicle for transfer is the T-strand, a single-stranded DNA molecule bound by the bacterial protein VirD2, which guides the T-DNA into the plant's nucleus where it integrates. How VirD2 is removed from T-DNA, and which mechanism acts to attach the liberated end to the plant genome is currently unknown. Here, using newly developed technology that yields hundreds of T-DNA integrations in somatic tissue of Arabidopsis thaliana, we uncover two redundant mechanisms for the genomic capture of the T-DNA 5' end. Different from capture of the 3' end of the T-DNA, which is the exclusive action of polymerase theta-mediated end joining (TMEJ), 5' attachment is accomplished either by TMEJ or by canonical non-homologous end joining (cNHEJ). We further find that TMEJ needs MRE11, whereas cNHEJ requires TDP2 to remove the 5' end-blocking protein VirD2. As a consequence, T-DNA integration is severely impaired in plants deficient for both MRE11 and TDP2 (or other cNHEJ factors). In support of MRE11 and cNHEJ specifically acting on the 5' end, we demonstrate rescue of the integration defect of double-deficient plants by using T-DNAs that are capable of forming telomeres upon 3' capture. Our study provides a mechanistic model for how Agrobacterium exploits the plant's own DNA repair machineries to transform it.

Gene targeting in polymerase theta-deficient Arabidopsis thaliana

Agrobacterium tumefaciens-mediated transformation has been for decades the preferred tool to generate transgenic plants. During this process, a T-DNA carrying transgenes is transferred from the bacterium to plant cells, where it randomly integrates into the genome via polymerase theta (Polθ)-mediated end joining (TMEJ). Targeting of the T-DNA to a specific genomic locus via homologous recombination (HR) is also possible, but such gene targeting (GT) events occur at low frequency and are almost invariably accompanied by random integration events. An additional complexity is that the product of recombination between T-DNA and target locus may not only map to the target locus (true GT), but also to random positions in the genome (ectopic GT). In this study, we have investigated how TMEJ functionality affects the biology of GT in plants, by using Arabidopsis thaliana mutated for the TEBICHI gene, which encodes for Polθ. Whereas in TMEJ-proficient plants we predominantly found GT events accompanied by random T-DNA integrations, GT events obtained in the teb mutant background lacked additional T-DNA copies, corroborating the essential role of Polθ in T-DNA integration. Polθ deficiency also prevented ectopic GT events, suggesting that the sequence of events leading up to this outcome requires TMEJ. Our findings provide insights that can be used for the development of strategies to obtain high-quality GT events in crop plants.

CRISPR/Cas9-Induced Double-Strand Break Repair in Arabidopsis Nonhomologous End-Joining Mutants

Double-strand breaks (DSBs) are one of the most harmful DNA lesions. Cells utilize two main pathways for DSB repair: homologous recombination (HR) and nonhomologous end-joining (NHEJ). NHEJ can be subdivided into the KU-dependent classical NHEJ (c-NHEJ) and the more error-prone KU-independent backup-NHEJ (b-NHEJ) pathways, involving the poly (ADP-ribose) polymerases (PARPs). However, in the absence of these factors, cells still seem able to adequately maintain genome integrity, suggesting the presence of other b-NHEJ repair factors or pathways independent from KU and PARPs. The outcome of DSB repair by NHEJ pathways can be investigated by using artificial sequence-specific nucleases such as CRISPR/Cas9 to induce DSBs at a target of interest. Here, we used CRISPR/Cas9 for DSB induction at the Arabidopsis cruciferin 3 (CRU3) and protoporphyrinogen oxidase (PPO) genes. DSB repair outcomes via NHEJ were analyzed using footprint analysis in wild-type plants and plants deficient in key factors of c-NHEJ (ku80), b-NHEJ (parp1 parp2), or both (ku80 parp1 parp2). We found that larger deletions of >20 bp predominated after DSB repair in ku80 and ku80 parp1 parp2 mutants, corroborating with a role of KU in preventing DSB end resection. Deletion lengths did not significantly differ between ku80 and ku80 parp1 parp2 mutants, suggesting that a KU- and PARP-independent b-NHEJ mechanism becomes active in these mutants. Furthermore, microhomologies and templated insertions were observed at the repair junctions in the wild type and all mutants. Since these characteristics are hallmarks of polymerase θ-mediated DSB repair, we suggest a possible role for this recently discovered polymerase in DSB repair in plants.

T-DNA integration in plants results from polymerase-θ-mediated DNA repair

Agrobacterium tumefaciens is a pathogenic bacterium, which transforms plants by transferring a discrete segment of its DNA, the T-DNA, to plant cells. The T-DNA then integrates into the plant genome. T-DNA biotechnology is widely exploited in the genetic engineering of model plants and crops. However, the molecular mechanism underlying T-DNA integration remains unknown1. Here we demonstrate that in Arabidopsis thaliana T-DNA integration critically depends on polymerase theta (Pol θ). We find that TEBICHI/POLQ mutant plants (which have mutated Pol θ), although susceptible to Agrobacterium infection, are resistant to T-DNA integration. Characterization of >10,000 T-DNA-plant genome junctions reveals a distinct signature of Pol θ action and also indicates that 3' end capture at genomic breaks is the prevalent mechanism of T-DNA integration. The primer-template switching ability of Pol θ can explain the molecular patchwork known as filler DNA that is frequently observed at sites of integration. T-DNA integration signatures in other plant species closely resemble those of Arabidopsis, suggesting that Pol-θ-mediated integration is evolutionarily conserved. Thus, Pol θ provides the mechanism for T-DNA random integration into the plant genome, demonstrating a potential to disrupt random integration so as to improve the quality and biosafety of plant transgenesis.

Epigenetic alterations at genomic loci modified by gene targeting in Arabidopsis thaliana

Gene Targeting (GT) is the integration of an introduced vector into a specific chromosomal site, via homologous recombination. It is considered an effective tool for precise genome editing, with far-reaching implications in biological research and biotechnology, and is widely used in mice, with the potential of becoming routine in many species. Nevertheless, the epigenetic status of the targeted allele remains largely unexplored. Using GT-modified lines of the model plant Arabidopsis thaliana, we show that the DNA methylation profile of the targeted locus is changed following GT. This effect is non-directional as methylation can be either completely lost, maintained with minor alterations or show instability in the generations subsequent to GT. As DNA methylation is known to be involved in several cellular processes, GT-related alterations may result in unexpected or even unnoticed perturbations. Our analysis shows that GT may be used as a new tool for generating epialleles, for example, to study the role of gene body methylation. In addition, the analysis of DNA methylation at the targeted locus may be utilized to investigate the mechanism of GT, many aspects of which are still unknown.

DAYSLEEPER: a nuclear and vesicular-localized protein that is expressed in proliferating tissues

BACKGROUND

DAYSLEEPER is a domesticated transposase that is essential for development in Arabidopsis thaliana [Nature, 436:282-284, 2005]. It is derived from a hAT-superfamily transposon and contains many of the features found in the coding sequence of these elements [Nature, 436:282-284, 2005, Genetics, 158:949-957, 2001]. This work sheds light on the expression of this gene and localization of its product in protoplasts and in planta. Using deletion constructs, important domains in the protein were identified.

RESULTS

DAYSLEEPER is predominantly expressed in meristems, developing flowers and siliques. The protein is mainly localized in the nucleus, but can also be seen in discrete foci in the cytoplasm. Using several vesicular markers, we found that these foci belong to vesicular structures of the trans-golgi network, multivesicular bodies (MVB's) and late endosomes. The central region as well as both the N- and the C-terminus are essential to DAYSLEEPER function, since versions of DAYSLEEPER deleted for these regions are not able to complement the daysleeper phenotype. Like hAT-transposases, we show that DAYSLEEPER has a functionally conserved dimerization domain [J Biol Chem, 282:7563-7575, 2007].

CONCLUSIONS

DAYSLEEPER has retained the global structure of hAT transposases and it seems that most of these conserved features are essential to DAYSLEEPER's cellular function. Although structurally similar, DAYSLEEPER seems to have broadened its range of action beyond the nucleus in comparison to transposases.

Poly(ADP-ribose)polymerases are involved in microhomology mediated back-up non-homologous end joining in Arabidopsis thaliana

Besides the KU-dependent classical non-homologous end-joining (C-NHEJ) pathway, an alternative NHEJ pathway first identified in mammalian systems, which is often called the back-up NHEJ (B-NHEJ) pathway, was also found in plants. In mammalian systems PARP was found to be one of the essential components in B-NHEJ. Here we investigated whether PARP1 and PARP2 were also involved in B-NHEJ in Arabidopsis. To this end Arabidopsis parp1, parp2 and parp1parp2 (p1p2) mutants were isolated and functionally characterized. The p1p2 double mutant was crossed with the C-NHEJ ku80 mutant resulting in the parp1parp2ku80 (p1p2k80) triple mutant. As expected, because of their role in single strand break repair (SSBR) and base excision repair (BER), the p1p2 and p1p2k80 mutants were shown to be sensitive to treatment with the DNA damaging agent MMS. End-joining assays in cell-free leaf protein extracts of the different mutants using linear DNA substrates with different ends reflecting a variety of double strand breaks were performed. The results showed that compatible 5'-overhangs were accurately joined in all mutants, that KU80 protected the ends preventing the formation of large deletions and that PARP proteins were involved in microhomology mediated end joining (MMEJ), one of the characteristics of B-NHEJ.

ZFN-mediated gene targeting of the Arabidopsis protoporphyrinogen oxidase gene through Agrobacterium-mediated floral dip transformation

Previously, we showed that ZFN-mediated induction of double-strand breaks (DSBs) at the intended recombination site enhanced the frequency of gene targeting (GT) at an artificial target locus using Agrobacterium-mediated floral dip transformation. Here, we designed zinc finger nucleases (ZFNs) for induction of DSBs in the natural protoporphyrinogen oxidase (PPO) gene, which can be conveniently utilized for GT experiments. Wild-type Arabidopsis plants and plants expressing the ZFNs were transformed via floral dip transformation with a repair T-DNA with an incomplete PPO gene, missing the 5' coding region but containing two mutations rendering the enzyme insensitive to the herbicide butafenacil as well as an extra KpnI site for molecular analysis of GT events. Selection on butafenacil yielded 2 GT events for the wild type with a frequency of 0.8 × 10⁻³ per transformation event and 8 GT events for the ZFNs expressing plant line with a frequency of 3.1 × 10⁻³ per transformation event. Molecular analysis using PCR and Southern blot analysis showed that 9 of the GT events were so-called true GT events, repaired via homologous recombination (HR) at the 5' and the 3' end of the gene. One plant line contained a PPO gene repaired only at the 5' end via HR. Most plant lines contained extra randomly integrated T-DNA copies. Two plant lines did not contain extra T-DNAs, and the repaired PPO genes in these lines were transmitted to the next generation in a Mendelian fashion.

The SLEEPER genes: a transposase-derived angiosperm-specific gene family

BACKGROUND

DAYSLEEPER encodes a domesticated transposase from the hAT-superfamily, which is essential for development in Arabidopsis thaliana. Little is known about the presence of DAYSLEEPER orthologs in other species, or how and when it was domesticated. We studied the presence of DAYSLEEPER orthologs in plants and propose a model for the domestication of the ancestral DAYSLEEPER gene in angiosperms.

RESULTS

Using specific BLAST searches in genomic and EST libraries, we found that DAYSLEEPER-like genes (hereafter called SLEEPER genes) are unique to angiosperms. Basal angiosperms as well as grasses (Poaceae) and dicotyledonous plants possess such putative orthologous genes, but SLEEPER-family genes were not found in gymnosperms, mosses and algae. Most species contain more than one SLEEPER gene. All SLEEPERs contain a C2H2 type BED-zinc finger domain and a hATC dimerization domain. We designated 3 motifs, partly overlapping the BED-zinc finger and dimerization domain, which are hallmark features in the SLEEPER family. Although SLEEPER genes are structurally conserved between species, constructs with SLEEPER genes from grapevine and rice did not complement the daysleeper phenotype in Arabidopsis, when expressed under control of the DAYSLEEPER promoter. However these constructs did cause a dominant phenotype when expressed in Arabidopsis. Rice plant lines with an insertion in the RICESLEEPER1 or 2 locus displayed phenotypic abnormalities, indicating that these genes are functional and important for normal development in rice. We suggest a model in which we hypothesize that an ancestral hAT transposase was retrocopied and stably integrated in the genome during early angiosperm evolution. Evidence is also presented for more recent retroposition events of SLEEPER genes, such as an event in the rice genome, which gave rise to the RICESLEEPER1 and 2 genes.

CONCLUSIONS

We propose the ancestral SLEEPER gene was formed after a process of retro-transposition during the evolution of the first angiosperms. It may have acquired an important function early on, as mutation of two SLEEPER genes in rice, like the daysleeper mutant in A. thaliana gave a developmental phenotype indicative of their importance for normal plant development.

Programmed Cell Death in the Leaves of the Arabidopsis Spontaneous Necrotic Spots (sns-D) Mutant Correlates with Increased Expression of the Eukaryotic Translation Initiation Factor eIF4B2

From a pool of transgenic Arabidopsis (Arabidopsis thaliana) plants harboring an activator T-DNA construct, one mutant was identified that developed spontaneous necrotic spots (sns-D) on the rosette leaves under aseptic conditions. The sns-D mutation is dominant and homozygous plants are embryo lethal. The mutant produced smaller rosettes with a different number of stomata than the wild-type. DNA fragmentation in the nuclei of cells in the necrotic spots and a significant increase of caspase-3 and caspase-6 like activities in sns-D leaf extracts indicated that the sns-D mutation caused programmed cell death (PCD). The integration of the activator T-DNA caused an increase of the expression level of At1g13020, which encodes the eukaryotic translation initiation factor eIF4B2. The expression level of eIF4B2 was positively correlated with the severity of sns-D mutant phenotype. Overexpression of the eIF4B2 cDNA mimicked phenotypic traits of the sns-D mutant indicating that the sns-D mutant phenotype is indeed caused by activation tagging of eIF4B2. Thus, incorrect regulation of translation initiation may result in PCD.

ZFN-induced mutagenesis and gene-targeting in Arabidopsis through Agrobacterium-mediated floral dip transformation

Zinc-finger nucleases (ZFNs) are artificial restriction enzymes, custom designed for induction of double-strand breaks (DSBs) at a specific locus. These DSBs may result in site-specific mutagenesis or homologous recombination at the repair site, depending on the DNA repair pathway that is used. These promising techniques for genome engineering were evaluated in Arabidopsis plants using Agrobacterium-mediated floral dip transformation. A T-DNA containing the target site for a ZFN pair, that was shown to be active in yeast, was integrated in the Arabidopsis genome. Subsequently, the corresponding pair of ZFN genes was stably integrated in the Arabidopsis genome and ZFN activity was determined by PCR and sequence analysis of the target site. Footprints were obtained in up to 2% of the PCR products, consisting of deletions ranging between 1 and 200 bp and insertions ranging between 1 and 14 bp. We did not observe any toxicity from expression of the ZFNs. In order to obtain ZFN-induced gene-targeting (GT), Arabidopsis plants containing the target site and expressing the ZFN pair were transformed with a T-DNA GT construct. Three GT plants were obtained from approximately 3000 transformants. Two of these represent heritable true GT events, as determined by PCR, Southern blot analysis and sequencing of the resulting recombined locus. The third plant showed an ectopic GT event. No GT plants were obtained in a comparable number of transformants that did not contain the ZFNs. Our results demonstrate that ZFNs enhance site-specific mutagenesis and gene-targeting of Agrobacterium T-DNA constructs delivered through floral dip transformation.

Manipulation of starch granule size distribution in potato tubers by modulation of plastid division

Starch granule size is an important parameter for starch applications in industry. Starch granules are formed in amyloplasts, which are, like chloroplasts, derived from proplastids. Division processes and associated machinery are likely to be similar for all plastids. Essential roles for FtsZ proteins in plastid division in land plants have been revealed. FtsZ forms the so-called Z ring which, together with inner and outer plastid division rings, brings about constriction of the plastid. It has been shown that modulation of the expression level of FtsZ may result in altered chloroplast size and number. To test whether FtsZ is also involved in amyloplast division and whether this, in turn, may affect the starch granule size in crop plants, FtsZ protein levels were either reduced or increased in potato. As shown previously in other plant species, decreased StFtsZ1 protein levels in leaves resulted in a decrease in the number of chloroplasts in guard cells. More interestingly, plants with increased StFtsZ1 protein levels in tubers resulted in less, but larger, starch granules. This suggests that the stoichiometry between StFtsZ1 and other components of the plastid division machinery is important for its function. Starch from these tubers also had altered pasting properties and phosphate content. The importance of our results for the starch industry is discussed.

Glucosylation activity and complex formation of two classes of reversibly glycosylated polypeptides

Reversibly glycosylated polypeptides (RGPs) have been implicated in polysaccharide biosynthesis. In plants, these proteins may function, for example, in cell wall synthesis and/or in synthesis of starch. We have isolated wheat (Triticum aestivum) and rice (Oryza sativa) Rgp cDNA clones to study the function of RGPs. Sequence comparisons showed the existence of two classes of RGP proteins, designated RGP1 and RGP2. Glucosylation activity of RGP1 and RGP2 from wheat and rice was studied. After separate expression of Rgp1 and Rgp2 in Escherichia coli or yeast (Saccharomyces cerevisiae), only RGP1 showed self-glucosylation. In Superose 12 fractions from wheat endosperm extract, a polypeptide with a molecular mass of about 40 kD is glucosylated by UDP-glucose. Transgenic tobacco (Nicotiana tabacum) plants, overexpressing either wheat Rgp1 or Rgp2, were generated. Subsequent glucosylation assays revealed that in RGP1-containing tobacco extracts as well as in RGP2-containing tobacco extracts UDP-glucose is incorporated, indicating that an RGP2-containing complex is active. Gel filtration experiments with wheat endosperm extracts and extracts from transgenic tobacco plants, overexpressing either wheat Rgp1 or Rgp2, showed the presence of RGP1 and RGP2 in high-molecular mass complexes. Yeast two-hybrid studies indicated that RGP1 and RGP2 form homo- and heterodimers. Screening of a cDNA library using the yeast two-hybrid system and purification of the complex by an antibody affinity column did not reveal the presence of other proteins in the RGP complexes. Taken together, these results suggest the presence of active RGP1 and RGP2 homo- and heteromultimers in wheat endosperm.