CRISPR RNA-guided integrase enables high-efficiency targeted genome engineering in Agrobacterium tumefaciens.

Agrobacterium tumefaciens, the causal agent of plant crown gall disease, has been widely used to genetically transform many plant species. The inter-kingdom gene transfer capability made Agrobacterium an essential tool and model system to study the mechanism of exporting and integrating a segment of bacterial DNA into the plant genome. However, many biological processes such as Agrobacterium-host recognition and interaction are still elusive. To accelerate the understanding of this important plant pathogen and further improve its capacity in plant genetic engineering, we adopted a CRISPR RNA-guided integrase system for Agrobacterium genome engineering. In this work, we demonstrate that INsertion of Transposable Elements by Guide RNA-Assisted TargEting (INTEGRATE) can efficiently generate DNA insertions to enable targeted gene knockouts. In addition, in conjunction with Cre-loxP recombination system, we achieved precise deletions of large DNA fragments. This work provides new genetic engineering strategies for Agrobacterium species and their gene functional analyses.

Introduction

Plant genetic transformation is critical for fundamental biological research and modern crop improvement. One widely adopted transformation technology is the Agrobacterium‐mediated method. Naturally, the gram‐negative soil bacterium Agrobacterium tumefaciens exists as an opportunistic pathogen in plants, and this niche is due to the presence of a large tumour‐inducing (Ti) plasmid. During pathogenesis, A. tumefaciens transfers a segment of its DNA (transfer DNA or T‐DNA) carrying plant growth hormone genes into the host plant's genome (Banta and Montenegro, 2008; Gelvin, 2021). The overproduction of the phytohormones often leads to neoplastic growth on the infection sites, termed ‘crown gall’ disease (Păcurar et al., 2011). Furthermore, the T‐DNA harbours a set of genes required to produce opines, which are used as energy sources for specific Agrobacterium strains. For the T‐DNA delivery, Agrobacterium senses environmental signals in the form of low pH, simple sugars, and phenolic exudates from wounded plants through the aid of virulence (vir) genes encoded on the chromosome and Ti plasmid (Gelvin, 2003, 2017). Due to their ability to transfer T‐DNAs, A. tumefaciens have been modified over the years for plant transformation. The modifications included ‘disarming’ the Ti plasmid by deleting the natural T‐DNA that carries ‘plant tumour’ genes (Hood et al., 1986; Koncz and Schell, 1986; Ooms et al., 1981), and utilizing a synthetic T‐DNA(s) from a binary plasmid (Komari et al., 2006). Other modifications include the generation of ‘disarmed’ strains such as EHA101 and EHA105 from highly virulent variants (Hood et al., 1986; Hood et al., 1993; Sciaky et al., 1978); generation of the ‘super‐binary vector’ (Komari et al., 2006) and ‘helper plasmid’ (Anand et al., 2018) that carries an additional copy of essential vir genes; and inactivating the recA recombinase gene to reduce plasmid recombination (Lazo et al., 1991). More recently, auxotrophic Agrobacterium strains have been generated to minimize post‐infection bacterial overgrowth during the plant transformation processes (Aliu et al., 2020; Anand et al., 2018; Ranch et al., 2012).

While the engineered Agrobacterium strains and binary vectors are popular for plant genome engineering, they do not work equally well on all plants. For example, strains and vectors that can be used for transforming maize may not work efficiently for yam plants. In addition, the genetic and molecular basis for Agrobacterium host range determination remains unknown (Gelvin, 2003, 2021). Moreover, even though the T‐DNA can be successfully transferred and integrated into the plant genome, the information and knowledge of its integration mechanisms are still lacking. As a result, the transformation of many plant species, especially economically important crops, has been challenging (Alteter et al., 2016). With the availability of the whole genome sequences of Agrobacterium tumefaciens C58 (Goodner et al., 2001; Wood et al., 2001), an opportunity is presented to further explore its biology through gene editing and genome engineering. The traditional molecular approach for gene knockin or knockout in Agrobacterium requires homologous recombination (HR) or transposon insertional mutagenesis (Morton and Fuqua, 2012). However, these methods have a combined drawback of being laborious, time‐consuming, and inefficient; as culturable and transformable bacteria often require screening many colonies to ascertain true mutants (Aliu et al., 2020; Tamzil et al., 2021).

The advent of CRISPR (clustered regularly interspaced short palindromic repeats)/Cas (CRISPR‐associated protein) technology has made it easier to perform genome editing. Rodrigues et al. reported the generation of loss‐of‐function mutations of recA gene in a popular disarmed Agrobacterium strain EHA105 using CRISPR‐mediated base editing (Rodrigues et al., 2021). Recently, RNA‐guided DNA integration systems have been developed for genome engineering in prokaryotes. These integration systems have been utilized to engineer bacterial cells with new functions via the stable integration of genes and pathways into the genome (Klompe et al., 2019; Makarova et al., 2018; Strecker et al., 2019). Especially, the INsertion of Transposable Elements by Guide RNA–Assisted TargEting (INTEGRATE) system utilizing a Type 1‐F CRISPR‐Cas system from Vibrio cholerae showed highly precise and efficient bacterial genome engineering capability (Klompe et al., 2019). INTEGRATE system consists of TniQ‐Cascade complex for RNA‐guided DNA targeting and transposase proteins TnsA, TnsB, and TnsC for cargo DNA transposition (Klompe et al., 2019). Target site is determined by a dinucleotide 5’‐CC‐3′ protospacer adjacent motif (PAM) and ~ 32‐nt protospacer sequence, and DNA integration occurs about 48–52 bp downstream from the protospacer (Klompe et al., 2019). Insertion DNA can be cloned into a mini transposon cargo between the left and right end sequences (L‐end and R‐end), and the cargo DNA size is negatively correlated with the targeted DNA insertion frequency (Klompe et al., 2019; Vo et al., 2021). INTEGRATE vectors stably replicate within the bacterial cells and can be easily removed after targeted DNA integration using a negative selectable marker such as sacB gene (Gay et al., 1983; Steinmetz et al., 1983) on the vector backbone. INTEGRATE does not require target site homology as in HR or the use of selectable markers and has been reported to achieve an insertion efficiency of ~100% in gram‐negative bacteria, including Escherichia coli, Klebsiella oxytoca, and Pseudomonas putida for single gene targeting (Vo et al., 2021). Because INTEGRATE system does not generate double‐strand DNA breaks (DSB), there is little to no off‐target effect (Vo et al., 2021). In addition, although the target sequence (PAM + protospacer) is not abolished by the cargo DNA insertion, the presence of the cargo provides ‘immunity’ preventing repetitive insertions at the same target site allowing precise single‐copy insertion events (Vo et al., 2021). Furthermore, when combined with the Cre‐loxP recombination system, small and large precise genomic deletions can be achieved (Vo et al., 2021). However, as with other gene‐editing systems, determining how to strategically port the INTEGRATE system into other bacteria of scientific relevance like A. tumefaciens poses a significant challenge due to cross‐species incompatibilities and host‐specific factors (Wannier et al., 2020).

In this paper, we report optimization and successful implementation of the INTEGRATE system in A. tumefaciens. We demonstrate that single and multiplex integration can be achieved in Agrobacterium with high efficiency. In addition, we show that insertional mutants resulting from INTEGRATE are stable, exhibit the expected phenotypes, and have similar plant transformation ability compared to the wild‐type (WT) when evaluated by transient gene expression. Furthermore, we demonstrate that by combining INTEGRATE with the Cre‐loxP recombination systems, we can precisely and effectively make large T‐DNA deletions of 25 kb and 42.6 kb in the WT Ti plasmids pTiC58 and pTiBo542, respectively. This work provides a simple and efficacious genome editing tool for Agrobacterium to accelerate gene functional analysis and explore many unanswered questions in Agrobacterium biology.

Results

Optimized multiplex DNA integration in Agrobacterium

To implement the INTEGRATE system for Agrobacterium genome engineering, we tested two different plasmid backbones with broad host range origin of replication (ORI) pBBR1 and pVS1, respectively (Figure 1a), using the Agrobacterium strains listed in Table S1. As a cargo, we cloned a red fluorescent protein marker gene mCherry (Shaner et al., 2004) driven by a constitutive J23107 promoter (Iverson et al., 2016) between the two transposon ends (L‐ and R‐end) (Figures 1b, 2a,b). For easy curing of the plasmids after targeted DNA insertion, we cloned a negative selection marker gene sacB (Figure S1a, b), which confers sucrose sensitivity by converting sucrose into toxic levan (Gay et al., 1983; Steinmetz et al., 1983).

Figure 1

Optimized multiplex DNA integration in Agrobacterium. (a) Schematic of INTEGRATE‐mediated targeted DNA insertions in Agrobacterium with distinct vector backbones, pBBR1 and pVS1. (b) Schematics of mCherry insertion at a target site. Target‐UP/DN, upstream and downstream sequence of a target site, respectively; forward and reverse primers for screening targeted insertion. (c) Targeted DNA integration frequency (Targeting efficiency) at three target sites estimated by PCR analysis. Targeting efficiency (%) was estimated by screening a total of 144 single colonies per target gene with three biological replicates (48 x 3 = 144). Data are shown as mean ± s.d. for n = 3 biological replicates. (d,g,j) Verification of targeted DNA insertion by PCR screening. Red arrowheads indicate representative triple targeted insertion mutant; yellow arrowheads indicate WT control band; gold square indicates mixed insertion mutant. (e,h,k) Chromatograms show the Sanger sequencing results confirming the targeted mCherry insertion for the target sites of virG (e), agpI (h), and flaA (k). TSD, target site duplication (blue coloured region). A larger band (~1.5 kb) indicates targeted mCherry insertion, whereas a smaller band (~0.5 kb) indicates no DNA integration (WT). Lanes 1–12, screened single colonies; CNT, wild‐type (WT) Agrobacterium EHA105 control. (f,i,l) Phenotypic analysis of the triple insertion mutant (EHA105VAF). The mutant strain exhibited expected mutant phenotypes: lack of vir gene induction for ΔvirG (f), faster growth rate at 25 °C for ΔagpI (i), and compromised motility for ΔflaA (l). Scale bars in (l) are 1 cm. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 2

Utilization of INTEGRATE to generate useful Agrobacterium strains for plant transformation. (a and b) Schematics of the recA genomic locus before and after targeted mCherry insertion. The pVS1‐based INTEGRATE construct is used for targeted insertion. (b) Chromatogram data confirm the targeted mCherry insertion at the recA genomic locus. The 5‐bp TSD (CTCAG) is highlighted with pink colour. PCR screening of targeted insertion at thyA of AGL1 (c), EHA101 (d), EHA105 (e), and at recA of EHA105Thy‐ (f) strains. Red arrowheads indicate representative targeted insertion mutant; yellow arrowheads indicate WT control; and gold squares show mixed insertion mutant. (g) PCR‐based quantification of integration efficiency at recA locus using two different origin of replications, pBBR1 and pVS1. Targeting efficiency (%) was estimated by screening a total of 144 single colonies with three biological replicates (48 x 3 = 148). Data are shown as mean ± s.d. for n = 3 biological replicates. (h) Thymidine‐dependent growth of auxotrophic Agrobacterium strains shown via streaked agar plates of Agrobacterium cells grown with/without 50 mg/L of thymidine. (i) Ultraviolet (UV) radiation assay for recA‐deficient A. tumefaciens strains: EHA105TR, EHA101TR, and AGL1T. EHA105 (WT) and AGL1 (WT) served as negative and positive controls, respectively. (j) Transient GUS expression assay using Arabidopsis efr‐1 seedlings to test the T‐DNA delivery capacity of Agrobacterium insertion mutants. All strains were transformed with a binary vector pTF102 with a gus reporter gene for AGROBEST assay. Wild‐type EHA105 strain without the binary vector served as a negative control. Scale bar = 2 mm. [Colour figure can be viewed at wileyonlinelibrary.com]

To test the effects of plasmid ORIs and the feasibility of multiplexed DNA insertions in Agrobacterium, we selected three target genes, virG (transcriptional activator of vir genes), apg1 (bacteriophytochrome protein), and flaA (flagella associated protein), based on their non‐lethality and known null mutant phenotypes (Chesnokova et al., 1997; Gelvin, 2003; Merritt et al., 2007; Xue et al., 2021). Because the presence of the INTEGRATE vectors (harbouring a constitutive promoter driving mCherry reporter) conferred red fluorescence even in the absence of targeted insertion, we used PCR screening to identify insertion mutants, as insertion of the mCherry cassette in the genome generates larger amplicons (>1.5 kb) compared to the WT (~ 0.5 kb, Figure 1b; Table S2). To estimate the targeted DNA insertion frequencies, we screened a total of 144 individual colonies per target gene (48 colonies x 3 independent experiments) and classified them into three groups: WT (no DNA insertion); Targeted insertion (targeted DNA insertion without WT band); Mixed insertion (targeted DNA insertion and WT band). As shown in Figure 1c, the INTEGRATE vector with pVS1 ORI performed much better than the pBBR1 ORI vector at all three target sites: Targeted insertion frequencies without WT band were 3.5–22.2% and 20.1–75.0% for pBBR1 and pVS1 ORIs, respectively (Figure 1c; paired t‐test, t = 3.0, df = 8, P < 0.05). Differences in total targeted insertion frequencies including the colonies with Mixed insertion were even greater with 8.3–42.4% for pBBR1 ORI and 45.1–98.6% for pVS1 ORI, respectively (Figure 1c; paired t‐test, t = 5.6, df = 8, P < 0.001). Pure insertion mutants were readily obtained from the Mixed insertion colonies by re‐plating after serial dilution as described in the methods. The three tested gRNAs exhibited varying levels of efficiency ranging from 0.7% (gRNA1 with pBBR1 ORI) to 75% (gRNA3 with pVS1 ORI; Figure 1c). Interestingly, gRNA2 exhibited smaller difference in Targeted insertion frequency (22.2% vs. 32.6% = 1.5‐fold) compared to the other two gRNAs (gRNA1, 3.5% vs. 20.1% = 5.7‐fold; gRNA3, 12.5% vs. 75.0% = 6.0‐fold). Consistent with the previous report (Vo et al., 2021), we also observed that gRNA efficiencies were dependent on the relative position of the spacer within the crRNA array: i.e., the first spacer (gRNA3, flaA) exhibited highest efficiency followed by the second (gRNA2, agpI) and the third (gRNA1, virG) spacers (Figure 1c; Figure S1). Importantly, with pVS1 ORI, colonies with targeted insertions were readily obtained with an average of 42% at all three target sites (triple insertion mutants) highlighting the efficacy of our INTEGRATE system for targeted DNA insertions in Agrobacterium.

To confirm the targeted DNA insertion into the target sites, PCR products were purified and subjected to Sanger sequencing. Targeted mCherry insertion was confirmed from all three target sites (Figures 1d,g,j), and the integration site was 47–54 bp downstream from the protospacer with either forward (T‐RL) or reverse cargo orientation (T‐LR), which is consistent with the previous report (Vo et al., 2021) (Table S3). Sanger sequencing also revealed that all targeted insertion events were accompanied by a 5‐bp target site duplication (TSD) at the L‐ or R‐end (Figures 1e,h,k), a well‐known feature generated by transposition of Tn7‐like elements (Peters et al., 2017), further confirming that targeted DNA insertion generated insertional mutations. We then cured the INTEGRATE vectors of the triple insertional mutant strain (EHA105VAF) deficient of virG, agp1, and flaA by plating the cells on a solid YEP medium amended with 5% sucrose. Emerging single colonies with antibiotics sensitivity were further tested for targeted insertion at all three target sites by PCR and Sanger sequencing before being used for phenotyping analysis.

For the virG null mutant phenotype, we used a vir gene induction assay. Agrobacterium utilizes VirA/G two‐component system to regulate vir gene expression (Gelvin, 2003, 2017). A well‐known Agrobacterium vir gene inducer is acetosyringone (AS), which is a phenolic exudate from wounded plants (Ashby et al., 1987) and is commonly used for Agrobacterium‐mediated plant transformation. AS has the ability to interact with transmembrane VirA protein, which in turn phosphorylates VirG. Once phosphorylated, VirG acts as a transcriptional activator for other vir genes (Winans et al., 1988). We generated a reporter plasmid, pEA183 (Table 1; Figure S2b) that has a green fluorescent protein (EGFP) gene under a VirG‐inducible virB promoter (PvirB) as described in the methods. The reporter pEA183 was introduced into the wild type EHA105 and the triple mutant strain EHA105VAF. To monitor the virG activity in these strains, we measured the green fluorescence with and without induction by AS. As shown in Figure 1f, the control strain (WT/eGFP) showed induction of the EGFP expression, whereas only a basal level of EGFP expression was observed in the EHA105VAF mutant strain (virG‐/eGFP), indicating that the triple mutant strain indeed lacks the ability to activate vir gene expression.

Table 1

Agrobacterium strains and plasmids used and produced in this study

Item	Description	Reference
Strains
AGL1T	thyA mCherry insertion mutant derived from AGL1	This study; Figure S3
Bo542ΔT‐DNA	Disarmed Bo542 strain	This study; Figure 3
C58ΔT‐DNA	Disarmed C58 strain	This study; Figure 3
EHA101TR	recA mCherry insertion mutant derived from EHA101Thy‐	This study; Figure 2i
EHA105TR	recA mCherry insertion mutant derived from EHA105Thy‐	This study; Figure 2i
EHA105VAF	virG, agpI and flaA mCherry insertion mutant derived from EHA105	This study; Figure 1
Plasmids
pEA106	Inducible PvirB‐mCherry construct	This study; Addgene #187872
pEA183	Inducible PvirB‐eGFP construct	This study; Figure S2b; Addgene #187873
pEA186	Empty pVS1‐based INTEGRATE vector with mCherry cargo; dual BsaI‐sites for new spacer cloning; sacB gene for plasmid curing; spectinomycin resistant.	This study; Addgene #187874
pKL2310	Empty pVS1‐based INTEGRATE vector with loxP cargo; cargo can be replaced by PstI and XhoI digestion; dual BsaI‐sites for new spacer cloning; spectinomycin resistant.	This study; Addgene #187875
pKL2315	Cre recombinase expression vector; sacB gene for plasmid curing; pVS1 ORI; kanamycin resistant.	This study; Figure S4c; Addgene #187876

We then tested if the triple mutant strain showed agp1 null mutant phenotype via a temperature‐sensitive growth assay. Agp1/2 are two phytochrome photoreceptor genes encoded in A. tumefaciens C58 genome which are responsible for sensing lights in blue, red, and far‐red wavelengths (Lamparter et al., 2021). Previous studies in Agrobacterium have shown that Agp1/2 can inhibit cell propagation under nutrient deprivation (Xue et al., 2021), as the single and double knockout mutants exhibited faster growth and reached higher cell densities than the WT strain during an expanded growth period. Because the histidine kinase domain of Agp1 is temperature‐dependent, exhibiting lower phosphorylation activity as temperature increases above 25 °C (Njimona and Lamparter, 2011), we grew the mutant and WT strains at 25 °C and monitored cell density over the course of 49 h. Consistent with the previous report (Xue et al., 2021), we observed that the agp1‐deficient strain grew faster than the WT and the difference became evident after 21 h (Figure 1i; Student t‐test, t = 3.6, df = 5, P < 0.05), suggesting growth inhibition by the Agp1 in the WT strain under nutrient deprivation conditions.

Lastly, we tested the flaA mutant phenotype using the swimming motility assay. Bacteria depend on external appendages, including pili and flagella, for movement through their environment. Swimming motility in Agrobacterium is mediated by flagella, which is encoded by four discrete flagellin gene homologues (flaABCD), and mutants of the flagellin genes have been reported nonmotile and attenuated for virulence (Chesnokova et al., 1997; Merritt et al., 2007). We assessed the swimming motility of the triple mutant strain on 0.25% swim agar plates (Figure 1l). Compared to the WT strain, the triple mutant strain demonstrated irregular colony shape and reduced motility, consistent with the flaA knockout mutant phenotype observed in previous reports (Chesnokova et al., 1997; Deakin et al., 1999). Together, these results confirm that our improved INTEGRATE system with pVS1 ORI is an efficient tool for multiplex targeted mutagenesis. The system can accelerate gene functional analyses in Agrobacterium and related bacteria.

Utilization of INTEGRATE to engineer Agrobacterium strains

Using the improved INTEGRATE system with pVS1 ORI (Figures 2a,b), we sought to engineer Agrobacterium strains to enhance plant transformation. As an essential biotechnology tool for plant gene functional analyses and crop trait improvements, the easy removal of Agrobacterium cells after the co‐cultivation period is one of the highly desired traits. Auxotrophic Agrobacterium strains have been proven to be useful for plant transformation (Anand et al., 2018; Ranch et al., 2012). We previously generated thymidine auxotrophic strains EHA101Thy‐ and EHA105Thy‐ using a suicide vector and homologous recombination approach (Aliu et al., 2020). However, this traditional mutagenesis approach suffered a low efficiency in generating thymidine knockout mutants (Aliu et al., 2020). Additionally, the recombination‐based mutagenesis cannot be used for Agrobacterium strains with suppressed functions in homologous recombination. For example, AGL1 strain is one of the widely used public strains for plant transformation due to insertional mutagenesis of the recA gene (Lazo et al., 1991), which encodes a recombinase enzyme essential for homologous recombination (Chen et al., 2008). The AGL1 strain is known to have greater plasmid stability when the binary vector carries DNA sequences with homologies.

Our previous attempt to generate thymidine auxotrophic AGL1 strain using the traditional recombination mutagenesis approach was not successful, because the recA mutation suppressed targeted thyA gene deletion via homologous recombination (Aliu et al., 2020). Using our improved INTEGRATE system, we attempted to generate thymidine auxotrophic AGL1 strain (AGL1T) as well as both thymidine auxotrophic/recA‐deficient (TR) EHA101 and EHA105 strains (Figure 2). The mCherry cargo used in the multiplex DNA insertions described above was also used here for single gene insertions (Figures 2a,b). A 32‐bp spacer targeting recA (atu1874) or thyA (atu2047) were cloned into the crRNA repeats as described in the methods. The resulting constructs (pEA189 and pEA190, Table S1) were introduced into EHA101, EHA105, and AGL1 strains via electroporation. After re‐plating (see Methods), emerging single colonies were screened by PCR to identify single colonies with targeted insertion (Figures 2c‐f), and Sanger sequencing analysis verified the precise insertion of the mCherry cargo into the target site (Figure 2b). As shown in Figure 2c, single colonies with targeted insertion were readily obtained for all target sites (thyA and recA) and strains (AGL1, EHA101, and EHA105). In addition, consistent with the multiplexing experiments, INTEGRATE vector with pVS1 ORI showed a much higher insertion frequency than the pBBR1 vector for recA gene (61.1% vs. 0.7%, Figure 2g), further demonstrating much improved targeted DNA insertion efficiency using the pVS1 ORI.

The resulting thyA mutant strains were tested for their thymidine dependence. Consistent with our previous observation (Aliu et al., 2020), all thymidine auxotrophic strains (AGL1T, EHA101T, and EHA105T) only grew in the presence of 50 mg L−1 thymidine in contrast to the corresponding wild‐type strain (Figure 2h), thus confirming successful knockout of the thyA gene by insertional mutagenesis. To monitor the stability of the insertional mutants generated by the INTEGRATE system, bacterial plate cultures were streaked every two days for 20 days. Colonies formed only in the presence of supplemented thymidine with a constitutive expression of red fluorescent protein, indicating that these strains did not lose the inserted mCherry cargo. In addition, Sanger sequencing analysis on the colonies from the 8th‐streaked plate showed that the target site in the AGL1T strain did not have additional sequence changes (Figure S3).

The recA gene encodes a protein that is essential for DNA repair (Chau et al., 2008). To test the recA insertion mutants (EHA101TR and EHA105TR), we subjected the Agrobacterium strains to UV‐B irradiation to observe their UV susceptibility. Consistent with the previous report (Chau et al., 2008), EHA101TR, EHA105TR, AGL1, and AGL1T exhibited increased sensitivity to UV irradiation compared to the wild‐type EHA105 strain (Figure 2i), confirming that these strains are indeed recA‐deficient mutants.

Lastly, we tested if these strains could deliver T‐DNAs into plant cells as well as their WT strains using a transient transgene expression assay, AGROBEST (Wu et al., 2014) (Figure 2j). A binary vector containing a GUS reporter gene, pTF102 (Frame et al., 2002), was introduced into the mutant and corresponding WT strains via electroporation. These strains were then used for infecting Arabidopsis thaliana T‐DNA insertion mutant efr‐1 (SALK 044334) (see Methods). Transient GUS staining results showed that there was no noticeable difference among the mutants and their parental strains as all tested strains showed strong GUS expression in the Arabidopsis seedlings (Figure 2j). These results indicate that our auxotrophic and recA‐deficient Agrobacterium strains (EHA101TR, EHA105TR, and AGL1T) generated by the INTEGRATE system are stable and suitable for plant transformation. Furthermore, their engineered traits would be advantageous for controlling Agrobacterium overgrowth or handling large plasmids with repetitive sequences.

Large DNA fragment deletions

In conjunction with Cre‐loxP recombination system, we utilized INTEGRATE system to demonstrate an efficient protocol to disarm two natural Ti plasmids (pTiBo542 and pTiC58) by deleting the T‐DNAs. The loxP site was inserted into the mini cargo. Two spacer sequences targeting the flanking regions of the T‐DNAs of the two pTi plasmids were cloned into the crRNA array in the INTEGRATE vector with pVS1 backbone resulting in pKL2308 (for pTiBo542) and pKL2313 (for pTiC58) (Figures S4a,b; Table S1).

As illustrated in Figure 3a, a large fragment deletion (42.6 kb) was conducted in two steps. The first step was to insert the loxP sites into the two target sites. Because of the improved efficiency with the pVS1 ORI, homogeneous Agrobacterium colonies with simultaneous insertions at both target sites were obtained from the Bo542 cells. Four out of 36 colonies carried homogeneous targeted insertion at both target sites (Table S4). However, targeted insertion efficiency was lower in the C58 cells as only 11% of screened colonies (3/28) had loxP insertion at either target site, but none had simultaneous insertions at both targets (Table S1). An extra re‐plating step was performed using a colony with loxP insertion at target 2 to obtain homogenous colonies with insertions at both target sites with a 70% frequency (14/20, Table S4). Sanger sequencing analysis was used to identify deletion‐ready colonies, which have loxP insertions at both target sites with the same orientation. Three out of four Bo542/2308 colonies with targeted insertions had forward cargo orientation (T‐RL) at both target sites (Table S5), thus they were eligible for Cre recombinase‐mediated deletion. Likewise, 8 out of the 14 C58/2313 colonies with loxP insertions at both target sites were analysed by Sanger sequencing, and all of them had the same loxP orientation (T‐RL) at both targets (Table S5).

Figure 3

Large DNA fragment deletion in conjunction with Cre‐loxP recombination system. (a) Schematic diagrams of two‐step deletion of natural T‐DNAs from pTiBo542. O, opine metabolism genes; TL, left T‐DNA; TR, right T‐DNA; V, virulence genes. (b) Illustration of loxP insertions into two target sites flanking the T‐DNAs in pTiBo542 and pTiC58. The chromatograms show the Sanger sequencing results for the strains Bo542ΔT‐DNA (top) and C58ΔT‐DNA (bottom) strains after recombination. TSD, target site duplication. T1, target 1; T2, target 2; F, forward primer; R, reverse primer. (c and d) Tumorigenicity assay of Bo542ΔT‐DNA (c) and C58ΔT‐DNA (d) strains on Kalanchoe leaves. Grey arrows, wild‐type strain; red arrows, disarmed strains. Scale bars in (c and d) are 1 cm. [Colour figure can be viewed at wileyonlinelibrary.com]

The second step was to induce recombination between the two loxP sites using Cre recombinase. Competent cells were prepared from the Bo542/2308–3 and C58/2313–9 colonies. A plasmid expressing Cre recombinase (pKL2315, Table 1, Figure S4c) was introduced via electroporation. Recombination between the two loxP sites was screened by PCR using the forward primer from target 1 (Table S2) and reverse primer from target 2 (Table S2). Recombination was detected from the first‐generation colonies in both Bo542/2308–3 and C58/2313–9 strains but the efficiency was lower in Bo542, presumably due to the larger distance between the two loxP sites (42.6 kb) compared to C58 (25 kb) (Table S4). An extra replating step in Bo542/2308–3 ensured obtaining homogeneous colonies without remaining target 2 sites (Table S4), and Sanger sequencing analysis confirmed targeted T‐DNA deletion by recombination between the two loxP sites as shown in Figure 3b. The signature 5‐bp TSD from the loxP cargo insertion at the target 2 remained after recombination. Because the Cre recombinase vector (pKL2315) carries a negative selection marker sacB gene, it was easily cured by spreading Agrobacterium cells on a solid YEP medium amended with 5% sucrose. The resulting colonies were susceptible to kanamycin, indicating that the Agrobacterium ΔT‐DNA mutant strains were free of pKL2315.

To verify the deletion of the natural T‐DNAs, the mutants (Bo542ΔT‐DNA and C58ΔT‐DNA) and their WT strains were inoculated onto Kalanchoe leaves for tumorigenicity assay. Three weeks after inoculation, only the WT strains incited tumours indicating that these disarmed strains indeed lost the oncogenes encoded on the T‐DNAs (Figures 3c,d). These results demonstrate that INTEGRATE/Cre‐loxP recombination system makes a powerful tool to generate precise large fragment deletions in Agrobacterium cells.

Discussion

We successfully implemented V. cholerae INTEGRATE system as an efficient genome engineering tool for Agrobacterium. A set of new vectors were developed with different origin of replication (pBBR1 and pVS1), transposon cargos (mCherry and loxP), and a negative selectable marker gene (sacB) to optimize the INTEGRATE system for targeted DNA integration and large fragment deletion in Agrobacterium (Table 1). These vectors can be easily modified to clone new DNA cargo or spacers using the existing restriction enzyme sites. For example, pEA186 and pKL2310 are empty (no guide RNA) pVS1‐based INTEGRATE vectors with mCherry and loxP cargo, respectively. They can be used for targeted insertion in Agrobacterium strains. In addition, Agrobacterium strains generated from this study (Table 1) can be useful for plant genetic transformation to control post co‐cultivation overgrowth or to enhance plasmid DNA stability.

Targeted insertional mutagenesis is a powerful tool to study gene functions. We showed the efficacy of this approach by targeting 5 different genes: thyA, recA, virG, flaA, and agp1 located on the A. tumefaciens chromosome and Ti plasmid, respectively. In particular, the thyA and recA genes were selected to generate useful Agrobacterium strains for plant transformation. Auxotrophic strains are increasingly useful not only for more controlled growth during plant transformation but also for easing the biosafety issues associated with the genome engineering tools such as CRISPR/Cas systems. Because the auxotrophic strains cannot survive outside of the laboratory conditions, using these strains to deliver genome‐editing reagents can decrease the risk of environmental release. In addition, the recA mutant strains can be useful when handling large plasmid DNAs with repetitive sequences or constructs with same promoters driving multiple genes, which often recombine within bacterial cells exhibiting plasmid instability problems. Prior to this study, three thyA‐deficient mutant strains have been reported including EHA101Thy‐ (Aliu et al., 2020), EHA105Thy‐ (Aliu et al., 2020), and LBA4404Thy‐ (Ranch et al., 2012). No successful attempt has been reported for generating thyA mutant using the homologous recombination approach from AGL1 strain, which has a deficiency in DNA repair and recombination due to the insertional mutation in the recA gene (Farrand et al., 1989; Lazo et al., 1991). Recently, Tamzil et al. reported the generation of threonine auxotrophic AGL1 strain using Tn5 transposon (Tamzil et al., 2021). Unlike RNA‐guided INTEGRATE system demonstrated here, the random insertional mutagenesis approach requires screening a large number of mutants as Tamzil et al. obtained 20 putative auxotrophs from 3315 mutants, demonstrating the cumbersome procedure of the traditional insertional mutagenesis approach (Tamzil et al., 2021). The recA mutant strains such as AGL1 are desired for plant transformation due to the increased plasmid stability within Agrobacterium cells, but AGL1 has been the only public recA mutant strain for almost 30 years until Rodrigues et al. generated recA‐deficient EHA105 strain using a cytosine base editor (Rodrigues et al., 2021). Using the INTEGRATE system, we readily generated both single and double knockout mutants lacking thyA and recA genes for the popular Agrobacterium strains EHA101, EHA105, and AGL1 strains (Table 1).

With further harnessing, INTEGRATE system can be scaled‐up for genome‐wide gene functional analyses. As demonstrated in other CRISPR‐Cas systems using large libraries of guide RNAs (Koike‐Yusa et al., 2014; Sanjana, 2017; Sharon et al., 2018), INTEGRATE system could be utilized for high‐throughput insertional mutagenesis assays to identify unknown Agrobacterium genes or genetic elements that play an important role for plant transformation or host range determination. Another useful aspect of the INTEGRATE system is the capability to precisely delete large DNA fragments in conjunction with the Cre‐loxP recombination system. A precise large genomic DNA fragment deletion approach could be suitable for synthetic biology approaches to modify a specific pathway by deleting an entire operon or introducing a set of genes up to 10 kb (Vo et al., 2021) to build new traits. Furthermore, as demonstrated here, novel Ti plasmids can be readily disarmed and utilized for plant transformation purposes. This can be especially useful for exploring novel Agrobacterium strains for wider host ranges, as diverse Agrobacterium strains have been sequenced. Comparative genomic and transcriptomic studies can provide new tools for understanding Agrobacterium‐plant interactions (Otten, 2021; Weisberg et al., 2020).

Like other bacterial genome editing systems, INTEGRATE also possesses certain drawbacks. First, due to end sequence‐specific recognition by the INTEGRATE machinery, this system cannot be leveraged for applications involving point mutations and scar‐free insertions (Klompe et al., 2019; Vo et al., 2021). Second, guide RNA design tool is not yet available for targeted insertions other than the 5’‐CC‐3’ PAM preference (Klompe et al., 2019), thus designing efficient guide RNAs would require testing multiple gRNAs for successful targeted DNA insertions. This study observed that certain guide RNAs have higher efficiency than others. For instance, the two guide RNAs used for pTiBo542 were more efficient than those used for the pTiC58 (Table S4). Guide RNA design tools can be developed in the future when more in vitro and in vivo data are accumulated, as were the case with previous CRISPR/Cas systems. Third, insertional mutagenesis may not be used for multiple rounds of transformation as there is a chance of remobilization of the previous cargo when the same INTEGRATE system is used in the subsequent round of engineering (Vo et al., 2021). This can be avoided by deploying additional RNA‐guided DNA insertion systems such as the ShCAST that utilizes a type V‐K CRISPR effector (Strecker et al., 2019). As an emerging tool for bacterial genome engineering, RNA‐guided DNA insertion systems utilizing transposon‐associated CRISPR systems such as INTEGRATE and ShCAST would be further tuned for robust and precise DNA insertion‐mediated bacterial genome engineering applications.

In summary, we successfully implemented and applied the RNA‐guided DNA integration system, INTEGRATE, for insertional mutagenesis approaches as well as precise large DNA fragment deletion for Agrobacterium genome engineering. INTEGRATE systems with highly efficient insertional mutagenesis and multiplexing capability can accelerate gene functional analyses and the generation of useful bacterial strains for synthetic biology or plant genetic transformation.

Materials and methods

Bacterial strains and growth conditions

Agrobacterium tumefaciens strains, plasmids, and primer sequences used in this study are listed in Tables S1 and S2, respectively. Three A. tumefaciens strains EHA101 (Hood et al., 1986), EHA105 (Hood et al., 1993), and AGL1 (Lazo et al., 1991) were used to generate thymidine auxotrophs via insertional gene mutagenesis using the INTEGRATE system. EHA101Thy‐ and EHA105Thy‐ previously derived from EHA101 and EHA105, respectively, by homologous recombination‐mediated thyA knockout (Aliu et al., 2020), were used to generate recA mutants, while EHA105 was used for INTEGRATE multiplexing studies. All strains have the A. tumefaciens C58 chromosomal background.

Vector construction

All V. cholerae INTEGRATE plasmid constructs listed in Table S1 were generated from the pSPIN vectors pSL1142 and pSL1765, which were gifts from Dr. Samuel H. Sternberg (Addgene plasmid # 160730 and # 160734), using a combination of polymerase chain reaction (PCR), restriction enzyme digestion, ligation, and Gibson assembly (Gibson et al., 2009). All PCR fragments for cloning were generated using the Phusion high‐fidelity DNA polymerase (ThermoFisher Scientific, MA, USA) or Q5 DNA Polymerase (New England Biolabs, MA, USA). crRNAs (for targeting genomic sites were designed with 32‐nt spacers) and all spacer sequences used for this study are available in Table S2.

Construction of the single and multiplex targeting constructs

We first made pBBR1‐based INTEGRATE vector pEA179 (Figure S1a) using pSL1142 and pSL1765 (Table S1). pVS1‐based INTEGRATE vector pEA191 (Figure S1b) was made by transferring the INTEGRATE components from pSL1765 to the vector pTF505 (Lee et al., 2013). To utilize the BsaI‐mediated spacer cloning in the INTEGRATE vectors, an endogenous BsaI restriction site within the pTF505 backbone (Figure S2a) was cured by site‐directed mutagenesis using the primers (primer IDs 29 and 30) listed in Table S2. As a cargo, the mCherry cassette containing J23107 promoter and rrnB gene terminator (J23107‐mCherry‐TrrnB) was PCR amplified with overlapping sequences for Gibson assembly with pSL1142 (or pSL1765) digested with XhoI and PstI. For single targeted insertion, each oligonucleotide pair for a new spacer were annealed and ligated with the BsaI‐digested vector backbone. Triple‐spacer arrays for multiplexing were cloned by combining two or three oligonucleotide duplexes with compatible sticky ends in a single ligation reaction. crRNAs were designed with 32‐nt spacers targeting genomic sites with a 5′‐CC‐3’ PAM. For easy curing of the INTEGRATE plasmids after targeted DNA insertion, a negative selection marker gene sacB was inserted via Gibson assembly by utilizing a unique KpnI site (Figures S1a,b). Cloning reactions were used to transform DH5α E. coli competent cells, and plasmids were extracted using QIAprep Spin Miniprep kit (Qiagen, Hilden, Germany) and confirmed by Sanger sequencing. Transformed cells were cultured in liquid Terrific Broth (TB) (Tartof and Hobbs, 1987) media amended with 50 mg L−1 of kanamycin (pBBR1 backbone) or 100 mg L−1 spectinomycin (pVS1 backbone).

Transposition assays in Agrobacterium

All transposition assays for A. tumefaciens were performed by introducing the INTEGRATE vectors via electroporation. Forty μL of electrocompetent cells were electroporated with 50–100 ng of plasmid DNA, in 2 mm electroporation cuvettes using a Bio‐Rad Gene Pulser (Bio‐Rad, CA, USA). Cells were recovered in 0.5 mL of SOC (or LB) media for 2 h in a shaking incubator at 28 °C and plated on yeast extract peptone (YEP) agar media with appropriate antibiotics and incubated at 28 °C for 48 h to yield the parent colonies.

Isolation of clonally integrated A. tumefaciens colonies

Based on previous report that single colonies might be genetically heterogeneous when integration occurs simultaneously with colony expansion (Vo et al., 2021), isolation of colonies was preceded by a ‘re‐plating’ step where the parent colonies were scraped and pooled together, then resuspended in YEP broth, and plated at an appropriate dilution to obtain a fresh set of colonies. A. tumefaciens cells transformed with INTEGRATE vectors were scraped from YEP agar plates and resuspended in liquid YEP. After 24 h growth in a shaking incubator at 28 °C with 200 rpm, cell density was measured using a spectrophotometer at 600 nm. About 50 μL of cell suspension after 106‐fold dilution prepared from the overnight culture was plated on YEP agar with appropriate antibiotics and incubated at 28 °C for 48 h to yield the first‐generation colonies that were screened by PCR using suitable primers (Table S2) to detect the integration of the transposon into the genome of A. tumefaciens.

PCR analysis of transposition

For detection of genomic integration, two genome‐specific primers flanking the anticipated integration region were used, and to determine the transposon orientation (T‐RL or T‐LR), one genome‐specific and one cargo‐specific primers were used. Thermocycling conditions were as follows: initial denaturation at 98 °C for 30 s, followed by 30 cycles of 10 s at 98 °C, 30 s at 63 °C, 30 s at 72 °C, with a final extension of at 72 °C for 5 min. PCR amplicons were resolved by electrophoresis on 1–1.3% agarose gels stained with Red Safe (Bulldog Bio, NH, USA). Colonies yielding the expected PCR products were cleaned up using the QIAquick PCR purification kit (Qiagen) and further subjected to Sanger sequencing to verify the targeted insertion site.

Thymidine‐dependent growth of auxotrophs

Seed cultures of thymidine auxotrophs generated by homologous recombination (HR) (Aliu et al., 2020) and INTEGRATE (EHA101T, EHA101TR, EHA105T, EHA105TR, and AGL1T) and their prototrophs were inoculated in 5 mL of YEP medium (supplemented with 50 mgL−1 of thymidine for the auxotrophs) in 50 mL falcon tubes and grown in a shaking incubator for 15 h at 28 °C with 200 rpm. A batch culture was prepared by transferring a calculated amount of overnight culture to a 50 mL of YEP medium in a 250 mL flask to a cell density of OD600 of 0.02. Batch cultures were grown in a shaking incubator (28 °C, 200 rpm) and 0.5 mL of culture was sampled every 4 h for 24 h to measure cell density using a spectrophotometer at 600 nm.

Evaluation of T‐DNA delivery capability

Evaluation of T‐DNA delivery capability of thymidine auxotrophic strains was done by AGROBEST assay (Wu et al., 2014). The efr‐1 (SALK 044334) T‐DNA insertional mutant Arabidopsis thaliana line was obtained from the Arabidopsis Biological Resource Center (Columbus, Ohio). For surface sterilization of seeds, 50% bleach (3% sodium hypochlorite) and 0.1% SDS solution was used for about 200–400 efr‐1 seeds in a 2 mL tube. After incubating for 15 min on a rotary mixer, seeds were rinsed four times with sterile water and 1 mL of half strength Murashige and Skoog medium (½ MS) containing 0.5% sucrose was added to the tube before being transferred to a 60 mm petri dish using a pipette and a wide‐bore tip. The efr‐1 seeds were grown for 7 days in a growth chamber at 22 °C under a 16 h/8 h light/dark cycle after 48 h cold treatment at 4 °C for uniform seed germination. The binary vector pTF102 (Frame et al., 2002) harbouring a GUS reporter gene (ß‐glucuronidase) driven by a cauliflower mosaic virus 35S promoter was transformed by electroporation into the thymidine auxotrophs and their corresponding WT strains serving as positive controls. For negative control, EHA105 strain without pTF102 was used. The strains were grown for 16 to 20 h in 5 mL of YEP medium supplemented with 50 mg L−1 thymidine (for thymidine auxotrophs) and appropriate antibiotics (50 mg L−1 kanamycin and 100 mg L−1 spectinomycin for EHA101T; 100 mg L−1 spectinomycin for EHA105T and AGL1T) in 50 mL tubes at 28 °C with 200 rpm. Agrobacterium cells were pelleted by centrifugation prior to infection and re‐suspended in AB induction medium (Gelvin, 2006) to a density of OD600 of 0.04. Ten to fifteen efr‐1 seedlings were transferred to each well of a 12‐well plate using sterile inoculation loops and 0.5 mL of ½ MS medium supplemented with 50 mg L−1 thymidine was pipetted into each well followed by the addition of 0.5 mL of Agrobacterium cell suspensions. The 12‐well plates were sealed with 3 M micropore tape and incubated in a growth chamber for two days at 22 °C under a 16 h/8 h light/dark cycle. Agrobacterium cells were removed by pipetting after a two‐day co‐cultivation, and 1 mL of fresh ½ MS medium amended with 100 mg L−1 cefotaxime and 100 mg L−1 timentin were added. Arabidopsis seedlings were grown in the growth chamber for two more days before GUS staining. With slight modifications, GUS staining was conducted as previously described (Cervera, 2005) to visualize transient transgene expression. For GUS staining, the remaining MS medium was removed before adding 1 mL of GUS staining solution to the efr‐1 seedlings. After overnight incubation at 37 °C, GUS staining solution was removed by pipetting and 1 mL of 75% ethanol was added to remove chlorophyll from the efr‐1 seedlings and incubated at room temperature for 12 to 16 h. Images were taken on a white background to aid T‐DNA delivery comparison between the auxotrophs and their corresponding prototrophs.

Virulence gene induction assay of mutant strain

virG mutant strain and its corresponding WT were transformed with a reporter plasmid with VirG‐inducible PvirB promoter driving a green fluorescent protein gene (EGFP). The Agrobacterium strains were grown for 20 h in 5 mL of YEP medium supplemented with 100 mg L−1 spectinomycin in 50 mL tubes at 28 °C with 200 rpm. Agrobacterium cells were then pelleted by centrifugation and re‐suspended in AB induction medium (Gelvin, 2006) to a density of OD600 of ~0.2 with or without 200 μM AS, and grown at 25 °C with 150 rpm for 21 h. EGFP induction was quantified by a fluorometer using a SYBR filter (492 nm for excitation, 516 nm for emission).

Ultraviolet (UV) light sensitivity assay of mutants

UV‐B light was used to test the EHA101 and EHA105 recA mutant strains. An overnight culture of each strain was prepared by inoculating a single colony into 50 mL conical tube containing 5 mL YEP medium and incubated at 28 °C with 200 rpm for 20 h. Cells were then diluted at a 1:100 ratio in a fresh YEP medium and grown to a cell density of OD600 of 0.7–0.9. A serially diluted bacterial suspension (10–106‐fold) was prepared in a six‐well plate and exposed to UV‐B irradiation for 2 h in the dark using a 6 W handheld UV lamp (Model UVM; Analytik Jena US LLC, Upland, CA, USA) placed at 20 cm from the top of the six‐well plate. After 2 h, 4 μL of bacterial suspensions were pipetted on YEP agar plates, which were covered with aluminium foil and incubated at 28 °C for 48 h. As a control, Agrobacterium cell suspensions from a duplicate six‐well plate without UV irradiation were also pipetted on YEP agar plates and incubated for 48 h. The recA‐deficient Agrobacterium AGLI strain was used as a positive control, while EHA105 WT strain was used as a negative control.

Temperature sensitivity assay of agp1 mutant

Seed cultures of EHA105 WT and the triple mutant EHA105VAF were inoculated in 5 mL of LB medium in a 50 mL falcon tube and grown in a shaking incubator for 15 h at 28 °C with 200 rpm. A batch culture was prepared by transferring a calculated volume of overnight culture to a 50 mL of LB medium in a 250 mL flask to a cell density of 0.02 at 600 nm. Batch cultures were grown in a shaking incubator (25 °C and 110 rpm) and 0.5 mL of culture was sampled every 7 h for 49 h to measure cell density using a spectrophotometer.

Flagellar motility assay

Swimming phenotypes of the EHA105 WT and the triple mutant EHA105VAF strains were assessed on YEP swim plates containing 0.25% Bacto agar (Xue et al., 2021). Single colonies were inoculated in 5 mL of liquid YEP medium in 50 mL conical tubes and the overnight cultures were then diluted to a density of OD600 of 0.4. Inoculation on swim plates was done by pipetting 1 μL of diluted culture onto the agar at the centre of the plate. Plates were incubated at 28 °C and images were captured at 24 h and 48 h to observe the phenotypic difference.

Cre‐‐mediated large DNA fragment deletion experiments

Cre‐loxP‐mediated targeted DNA deletion was performed with two‐step procedures. The first step was to insert the loxP sites into the two target sites using the INTEGRATE system. The loxP site was inserted into the cargo and two crRNAs targeting the flanking sequences of the T‐DNAs of pTiBo542 and pTiC58 were cloned into the CRISPR array of the INTEGRATE vector with pVS1 backbone (pKL2310) using the oligonucleotides listed in the Table S2. A. tumefaciens strain Bo542 and C58 were transformed with the pKL2308 and pKL2313, respectively, by electroporation as described above and the resulting parental colonies were pooled, diluted, and spread on fresh solid YEP medium (re‐plated). After two days, resulting colonies were screened by PCR using the target‐specific primer pairs listed in Table S2. PCR was performed using the Q5 2X master mix (New England Biolabs, MA, USA) and the thermocycling conditions defined above. For samples with only targeted insertion event (i.e., larger bands about 800 bp), Sanger sequencing analysis was followed to confirm the insertion of loxP sites and their orientation. The second step was to introduce another pVS1‐based vector expressing Cre recombinase (pKL2315; Figure S4c) into the Bo542 or C58 cells that have two loxP site insertions with same orientation (T‐RL). Because the Cre recombinase vector carries a different antibiotic resistance gene (kanamycin), the second step eliminates the INTEGRATE vector (spectinomycin) from Agrobacterium cells. PCR screening using the forward primer of target 1 and reverse primer of target 2 was carried out to identify Agrobacterium colonies with 42.6 kb or 25 kb deletions in Bo542 or C58 strains, respectively. Control PCRs using the target‐specific primers were also performed to detect the presence of undeleted T‐DNAs. The re‐plating step was implemented if the control PCRs were positive, indicating that the colonies were heterogeneous. Sanger sequencing analysis was used to validate the precise recombination and deletion of the T‐DNAs.

Tumorigenesis assay on Kalanchoe leaves

Tumorigenesis assay on Kalanchoe leaves was used to confirm the deletion of oncogenes encoded on the T‐DNAs. Agrobacterium strains were grown for 20 h in 5 mL liquid YEP medium in 50 mL conical tubes in a shaking incubator at 28 °C with 200 rpm. Cells were collection by centrifugation and resuspended in AB induction medium (Gelvin, 2006). After adjusting the cell density to OD600 of 0.2, Agrobacterium cells were grown under vir gene induction condition with 200 μM AS for 20 h at 25 °C with 200 rpm. Six to eight weeks old Kalanchoe leaves were inoculated as described previously (Guo et al., 2007). Hypodermic needle was first used to wound the leaves, then T‐DNA deletion mutants and their WT strain were inoculated onto the same leaves. Three microliters of bacterial inoculum were pipetted onto each wound and tumour formation was monitored after 2–3 weeks.

Statistical analysis

Differences in targeted DNA insertion frequencies were compared using two‐tailed paired sample t‐test (Hsu and Lachenbruch, 2014).

Author contributions

K.W. conceived the research idea, K.L. and E.A. designed and performed experiments. E.A. performed single and multiplex insertion assays and analysed loss of function mutants. K.L. performed programmed genomic deletion assays and analysed deletion mutants. K.L. and E.A. analysed the data. K.L., E.A. and K.W. wrote the paper.

Competing interests

The authors declare no competing interest.

Table S1‐S5 Supplementary Tables.

Figure S1‐S4 Supplementary Figures.

Click here for additional data file.