An Agrobacterium tumefaciens EHA105-Based GAANTRY Recipient Strain Generates High-Quality Transgenic Arabidopsis and Potato

Abstract

The GAANTRY (Gene Assembly in Agrobacterium by Nucleic acid Transfer using Recombinase technology) system enables efficient gene stacking within an Agrobacterium T-DNA. Using unidirectional site-specific recombinases and alternating selection markers, it allows precise, sequential assembly of multiple genes directly within an Agrobacterium virulence plasmid. Here, we modified Agrobacterium tumefaciens strain EHA105 to create JGT105 as a GAANTRY recipient and constructed a 15.8 kb T-DNA containing five cargo sequences. We compared the performance of the JGT105 5-stack strain against a conventional binary vector carrying the same cargo sequences in Arabidopsis and potato transformation. The transformation efficiencies were comparable for the GAANTRY strain and the binary vector (potato: 83% vs. 82%; Arabidopsis: 1.73% vs. 1.95%). Single T-DNA insertion frequencies were also similar between the two systems (17.6% for GAANTRY vs. 24.5% for the binary vector construct in potato; 10.3% vs. 18.2% in Arabidopsis, respectively). Notably, the GAANTRY construct had significantly reduced vector backbone transfer in potato (10.0% vs. 26.5%) for the binary vector, whereas rates were higher in Arabidopsis (37.5% vs. 48.9%). These results show that the JGT105 GAANTRY strain is an effective T-DNA delivery system, matching binary vector transformation efficiency while offering lower backbone integration frequency, facilitating the generation of high-quality, multi-gene transgenic plants.

1. Introduction

Efficiently assembling large, multi-gene constructs for stable plant transformation is crucial for advancing plant biotechnology. While conventional binary vectors often struggle with the stability and size limitations of complex T-DNAs [1,2,3,4], the GAANTRY (Gene Assembly in Agrobacterium by Nucleic acid Transfer using Recombinase technology) system provides a robust solution allowing for the precise, sequential assembly of multiple genes directly within the Agrobacterium T-DNA region. This technology has proven effective in transforming model and crop species like Arabidopsis, potato, and rice, achieving high transformation efficiencies while significantly reducing co-transfer of the vector backbone compared to binary vectors [5].

GAANTRY is built upon the use of site-specific recombinase enzymes, which mediate genetic recombination at defined recognition sites. The type and arrangement of these sites, along with the choice of recombinase, determine the outcome—whether by insertion, deletion, or inversion of the DNA segment. The system incorporates the ParA-MRS recombination module, in which ParA is a unidirectional serine recombinase, and MRS is its 106 bp recognition site. ParA catalyzes irreversible excision of DNA flanked by MRS sites in direct orientation, while regenerating the recognition site for repeated use [6,7]. For integration, GAANTRY utilizes the large serine recombinases A118 and TP901-1, which act on distinct 55–65 bp attachment sites (attB and attP) to produce non-functional hybrid sites (attPB and attBP), ensuring unidirectional integration [8,9]. These recombinases work sequentially to mediate the insertion of donor DNA into predefined loci on the Agrobacterium virulence plasmid, simultaneously introducing a bacterial selection marker. Following successful integration, a subsequent excision event—driven by the proper arrangement of recombination sites—removes the previous bacterial selection marker gene, enabling marker recycling and iterative construct assembly.

This coordinated recombination and selection strategy allows for the precise and efficient stacking of multiple genes into large and complex constructs [5,7,10]. Because the system relies on recombinase-specific recognition sequences rather than host-specific elements, GAANTRY is modular and transferable to any Agrobacterium strain that is sensitive to at least two antibiotics. The GAANTRY system was originally developed using ArPORT1 as the recipient strain. This strain was derived from Agrobacterium rhizogenes NCPPB 2659, and although this strain has been successfully used to transform Arabidopsis, potato, rice, and other crop species, it is not as frequently used as some other Agrobacterium strains for plant transformation.

To expand the utility of the GAANTRY system and its capacity for modular, site-specific assembly of large multigene constructs, we adapted the widely used Agrobacterium tumefaciens strain EHA105 [11], which is noted for its broad applicability across plant species [12,13], as a GAANTRY recipient. The resulting strain, JGT105, was generated, validated, and evaluated for its ability to accept hierarchically assembled GAANTRY cargo arrays and to stably deliver complex T-DNAs for plant transformation. To benchmark GAANTRY-mediated multigene delivery, we compared a GAANTRY strain (JGT5046) harboring five cargo sequences within a single 15.8-kb T-DNA with an EHA105 strain carrying the pBi5045 binary vector, which contains an equivalent five-cargo T-DNA. Transformation efficiency, T-DNA copy number, vector backbone integration frequency, and phenotypic function were systematically assessed across multiple independent transgenic Arabidopsis and potato lines.

2. Materials and Methods

2.1. Installation of the GAANTRY Components into Agrobacterium tumefaciens EHA105

Hood and colleagues described the generation of the pEHK2 plasmid that was used to disarm the pTiBo542 virulence plasmid to create pEHA101 [14]. The EHA105 strain is a derivative of EHA101, which was created by removing the introduced kanamycin resistance marker [11]. The two strains are expected to carry the same deletion of the native T-DNA region. To confirm that our EHA105 strain contained the expected deletion within the T-DNA region, multiple pairs of primers were designed to regions near the Left and Right borders of the pTiBo542 T-DNAs and used for PCR amplification with EHA105 genomic DNA as template. Results confirmed that ~57 kb of the pTiBo542 virulence plasmid including all of the T-DNA sequences plus about 16 kb downstream of the T-DNA Right Border (RB) region and 14 kb of the upstream of the Left Border (LB) region was deleted. The subsequent sequencing of the disarmed pEHA101 virulence plasmid by others (GenBank KY000035), confirmed our results.

Once confirmed, the primer pair EHA105 TAG LB4 F60/R60 983 bp amplicon was used to generate the Left Arm for and the EHA105 TAG RB4 F60/R60 1072 bp amplicon (Figure 1; Supplementary Table S3) was used to generate the Right Arm for the homologous recombination vector (pLAKanRA; Supplementary Figure S1) used to insert the GAANTRY technology into the EHA105 virulence plasmid. The triparental mating protocol and selection strategy for homologous recombination are as previously described [5].

Figure 1. (A) Diagram of JGT105 A. tumefaciens GAANTRY recipient strain target region for cargo sequence insertion. The modified Agrobacterium tumefaciens EHA105 strain carrying the C58 Left Border sequence (LB T-DNA repeat), an A118 attP recognition site, a bacterial kanamycin resistance marker and a ParA MRS site (ParA res) integrated within the Ti plasmid. (B) GAANTRY strain JGT5046 T-DNA diagram showing the modified potato ALS (mStALS), GUS, EGFP, GAox1 RNAi, and nptII sequences. (C) Diagram of the pBi5045 binary vector T-DNA also carrying the same order and configuration of cargo sequences. See Table 1 for a detailed description of the sequences shown in the (B,C) diagrams. Promoter (green arrows) include Potato Ubiquitin 7 (StUbi7p), FigWort Mosaic virus (FMVp); double enhanced Cauliflower Mosaic 35S (db35Sp), and Arabidopsis Ubiquitin 10 (AtUbi10p). Terminator (white boxes) include potato Ubiquitin 3 (Ubi3T) and Nopaline Synthase 3′UTR (Nos3′T). attPB and attBP indicate hybrid, non-functional recombinase recognition sites. LB and RB represent T-DNA border sequences and Multiple Cloning Site (gray boxes). See Table 1 for a detailed description of the coding sequences denoted by (orange arrows). Schematics are not to scale.

Figure 1. (A) Diagram of JGT105 A. tumefaciens GAANTRY recipient strain target region for cargo sequence insertion. The modified Agrobacterium tumefaciens EHA105 strain carrying the C58 Left Border sequence (LB T-DNA repeat), an A118 attP recognition site, a bacterial kanamycin resistance marker and a ParA MRS site (ParA res) integrated within the Ti plasmid. (B) GAANTRY strain JGT5046 T-DNA diagram showing the modified potato ALS (mStALS), GUS, EGFP, GAox1 RNAi, and nptII sequences. (C) Diagram of the pBi5045 binary vector T-DNA also carrying the same order and configuration of cargo sequences. See Table 1 for a detailed description of the sequences shown in the (B,C) diagrams. Promoter (green arrows) include Potato Ubiquitin 7 (StUbi7p), FigWort Mosaic virus (FMVp); double enhanced Cauliflower Mosaic 35S (db35Sp), and Arabidopsis Ubiquitin 10 (AtUbi10p). Terminator (white boxes) include potato Ubiquitin 3 (Ubi3T) and Nopaline Synthase 3′UTR (Nos3′T). attPB and attBP indicate hybrid, non-functional recombinase recognition sites. LB and RB represent T-DNA border sequences and Multiple Cloning Site (gray boxes). See Table 1 for a detailed description of the coding sequences denoted by (orange arrows). Schematics are not to scale.

Table 1. List of B and P Donor vectors used to build the GAANTRY 5-stack and the transcriptional units that were used to assemble the binary vector design that mirrors the GAANTRY configuration. * CDS: gene coding sequence.

Stack #	Donor Vector	Plasmid ID	Cargo Size (kb)	Phenotype	Cargo Sequence
					Promoter	CDS *	Term
1	B	pB-5025	4.4	Imazamox resistance	StUbi7	mStALS	Ubi3′
2	P	pP-5028	3.0	β-glucuronidase activity	FMV	uidA (GUS)	Ubi3′
3	B	pB-5033	2.4	Green fluorescence protein	db35s	eGFP	Nos3′
4	P	pP-5032	3.7	Silencing of gibberellin (GA) 20-oxidase	StUbi7	sGA20ox1	Ubi3′
5	B	pB-5150	2.3	Kanamycin resistance	AtUbi10	nptII	Nos3′

Table 1. List of B and P Donor vectors used to build the GAANTRY 5-stack and the transcriptional units that were used to assemble the binary vector design that mirrors the GAANTRY configuration. * CDS: gene coding sequence.

Stack #	Donor Vector	Plasmid ID	Cargo Size (kb)	Phenotype	Cargo Sequence
					Promoter	CDS *	Term
1	B	pB-5025	4.4	Imazamox resistance	StUbi7	mStALS	Ubi3′
2	P	pP-5028	3.0	β-glucuronidase activity	FMV	uidA (GUS)	Ubi3′
3	B	pB-5033	2.4	Green fluorescence protein	db35s	eGFP	Nos3′
4	P	pP-5032	3.7	Silencing of gibberellin (GA) 20-oxidase	StUbi7	sGA20ox1	Ubi3′
5	B	pB-5150	2.3	Kanamycin resistance	AtUbi10	nptII	Nos3′

Using primer pair EHA105 LA1210 F60/NPTIII 1130 R59 a 1979 bp band was generated to confirm genome placement of the LB region and allowed sequence to verify the A118 attP site. The EHA105 LA1210 F60 primer was designed outside the homologous recombination arm for the LB, while NPTIII 1130 R59 is located within the GAANTRY construct (Supplementary Figures S1 and S2; Supplementary Table S3). Using primer pair NPTIII 1700 F60/EHA105 RA5510 R60, a 1836 bp band was generated to confirm genome placement of the RB region and allowed sequence to verify the ParA MRS site. The EHA105 RA5510 R60 primer was designed outside the homologous recombination arm for the RB, while NPTIII 1700 F60 is located within the GAANTRY construct (Supplementary Figures S1 and S2; Supplementary Table S3).

2.2. Building the Donor Plasmids

The genes of interest were inserted into the B or P Donor plasmids [5] using standard restriction-ligation cloning techniques. The ligated plasmids were transformed into E. coli DH5α and plasmid DNA was column purified using a kit following the manufacturer’s instruction (Zymo Research, Irvine, CA, USA). The Donors that were used to create the 5-stack JGT5046 GAANTRY strain T-DNA are listed in Supplementary Figure S3.

2.3. Construction of the 5-Stack JGT105 GAANTRY Strain

Transgene stacking was performed as previously described [5,15] using the JGT105 GAANTRY recipient strain to generate JGT5046. This involved a process of iterative recombinase-mediated integration and excision reactions using B Donor vectors (which confer bacterial gentamicin resistance) and P Donor vectors (which confer bacterial kanamycin resistance), each carrying the desired cargo. The 5-stack JGT5046 GAANTRY strain was created in five steps using the B and P Donor vectors specified in Table 1. To prevent potential issues with constitutive kanamycin resistance, the nptII expression cassette in the 5-stack construct was added as the last stack (stack 5). The strains generated from each stacking event were characterized and validated using genomic PCR reactions with sequence-specific primers that spanned the junctions between preexisting sequences and newly inserted cargo (Supplementary Figure S5). Supplementary Table S3 contains the primers used for the analysis, and the PCR amplification conditions included an initial denaturation step of 5 min at 95 °C, followed by 35 cycles of 30 s at 95 °C, 30 s at 58 °C, and 2 min at 72 °C, with a final extension at 72 °C for 5 min.

2.4. Construction of a 5-Stack pBi5045 Binary Vector Plasmid

A nearly identical T-DNA was constructed in a binary vector to compare with the GAANTRY assembled 5-stack T-DNA. The pBi5045 5-stack binary vector (Supplementary Figure S4) was constructed using restriction digest/ligation cloning techniques in a step-by-step procedure by PCR amplifying (Phusion polymerase; NEB, Ipswich, MA, USA) the cargo sequences from the previously built B/P Donors and ligating them into the binary vector until the 5-stack was obtained and termed pBi5045 (Supplementary Figure S4). The ligated plasmids were introduced into E. coli DH5α via transformation, and subsequent purification of the plasmid DNA was carried out using a kit according to the manufacturer’s instructions (Zymo Research, Irvine, CA, USA). The purified pBi5045 plasmid vector was then transformed into A. tumefaciens EHA105 cells to obtain the pBi5045 strain. The pBi4030 binary vector construct (pCambia0390 with the StALSm resistance gene; Supplementary Figure S4) was also created and used as a control.

2.5. Transformation and Generation of Transgenic Potato and Arabidopsis Plants

The 5-stack GAANTRY strain JGT5046, the EHA105 strain harboring the 5-stack pBi5045 binary vector, or the pBi4030 control binary vector were used for Ranger Russet (RR) potato (Solanum tuberosum) transformation and regeneration as previously described [6]. Briefly, explants were infected with Agrobacterium, co-cultivated, and transferred to shoot induction medium containing imazamox (0.3 mg L⁻¹) for selection of ALS-expressing transformants. This concentration was previously validated to completely suppress regeneration of non-transformed explants while allowing robust recovery of transformed shoots.

A total of 687 RR explants transformed with JGT5046 generated 570 imazamox-resistant shoots, corresponding to a transformation efficiency of 83%, calculated as the percentage of explants producing at least one imazamox-resistant shoot. Similarly, 465 RR explants transformed with pBi5045 generated 380 (82%) imazamox-resistant shoots. Shoots were confirmed by qPCR for presence of the ALS gene.

Eighty phenotypically normal imazamox-resistant shoots (independent events) from each construct were selected for visual assays (GUS, eGFP), antibiotic resistance rooting (StALSm, nptII) and molecular genotyping. Rooting was performed on Murashige and Skoog (MS) medium supplemented with imazamox (0.3 mg L⁻¹) and kanamycin (100 mg L⁻¹) to confirm stable transgene integration. Off-type shoots were discarded. Both the wildtype (WT) and plants transformed with pBi4030 control vector were included throughout regeneration, selection and molecular confirmation to ensure selection stringency and media performance. Fifty randomly selected lines from each construct that expressed the four reliably assayed traits were transferred to the greenhouse to generate plant material for backbone detection and transgene copy number analysis.

Arabidopsis thaliana (Columbia-0) was transformed using the floral dip method [16]. The T₁ seeds were collected, subjected to surface sterilization, and subsequently germinated on Murashige and Skoog (MS) medium with 1.0% sucrose, timentin (150 mg L⁻¹) to prevent contamination by Agrobacterium, and kanamycin (50 mg L⁻¹) as the selection agent. The transformation efficiency was calculated by estimating the number of seeds from seed mass (100 mg ≈ 4000 seeds) and counting the number of kanamycin-resistant seedlings recovered. For JGT5046, of the 4000 seeds plated, 69 were kanamycin resistant, giving a (1.73% transformation efficiency). For pBi5045, a binary vector construct, 78 resistant seedlings were recovered from the 4000 seeds tested, resulting in a (1.95% efficiency). Resistant seedlings were transferred to soil after two weeks and grown to maturity for molecular analysis and seed harvest.

2.6. Phenotypic Analysis of Transgenic Plants

2.6.1. Antibiotic and Herbicide Resistance

Resistance of transgenic potato plants to selection agents was assessed using sequential rooting assays. Briefly, transformed explants for each construct were regenerated on shoot induction medium containing imazamox (0.3 mg L⁻¹) for ALS selection. One plantlet per explant was collected and rooted on MS medium containing imazamox (0.3 mg L⁻¹) to confirm herbicide resistance during rooting. Successfully rooted plants were subsequently transferred to MS medium containing 100 mg L⁻¹ kanamycin to examine antibiotic resistance. After two weeks, the number of rooted shoots was recorded (Supplementary Figure S6). These plantlets were then transferred to the greenhouse for growth, tissue harvest, and further molecular analysis.

Leaf size was characterized using fully expanded, mature terminal leaves collected from the mid-canopy of russet potato (Solanum tuberosum L.) plants at the vegetative–early tuber bulking stage. The terminal leaflet, which represents the largest leaflet, was used as a standard morphological reference. All measurements were taken from leaves free of visible damage or disease.

The assessment of Arabidopsis resistance to selective chemicals involved germinating T1 seedlings on solid MS medium with kanamycin (50 mg L⁻¹). After two weeks, the number of healthy, resistant seedlings was recorded, and plants were transferred to the greenhouse for further analysis. Under our experimental conditions, Arabidopsis seedlings failed to show resistance to imazamox (0.3 mg L⁻¹); therefore, herbicide resistance was excluded from phenotypic analysis in this species. Consequently, the presence of the StALS transgene was confirmed via genomic PCR amplification (Supplementary Figure S5) using the primers shown in Supplementary Table S3.

2.6.2. β-Glucuronidase (GUS) Staining

To assess GUS activity, potato or Arabidopsis leaflets were incubated with histochemical staining solution (0.1 M sodium phosphate, pH 7.0, 0.5 mM potassium ferrocyanide, 1.5 g/L X-gluc (5-bromo-4-chloro-3-indolyl-β-D-glucuronic acid), and 0.5% v/v Triton X-10, Sigma-Aldrich, St. Louis, MO, USA) for 18 h at 37 °C [17].

2.6.3. Detection of Green Fluorescence

To score green fluorescence, potato or Arabidopsis leaves were viewed under a Leica Microsystems-MZ16F stereomicroscope (Bannockburn, IL, USA) using an excitation wavelength of 488 nm with an emission filter of 510–530 nm. This bandpass filter set eliminates chlorophyll autofluorescence.

2.6.4. Analysis of GA20 Oxidase Silencing

The plants were visually observed, and the height of each plantlet was compared to the wild-type control plantlets of the same age. Plants significantly shorter were scored positive for GA20 oxidase silencing (Supplementary Figure S6). However, scoring of the GAox1 phenotype was considered unreliable, and thus the presence of the GAox1 RNAi cassette was evaluated via genomic PCR amplification using the primers shown in Supplementary Table S3.

2.7. Genotypic Analysis of Transgenic Plants

Potato and Arabidopsis gDNA were extracted from leaf tissues using the PureGene DNA isolation kit (Qiagen, Valencia, CA, USA). To verify the presence or absence of vector backbone sequences, either PCR, qPCR, and or ddPCR was performed in Arabidopsis and or potato using the primers shown in Supplementary Tables S3–S5.

Transgene copy number was measured using droplet digital PCR (ddPCR). The potato vacuolar invertase gene INV-2 (aka. Pain-1) (GenBank HQ110080) and the AAP1 gene (At1g58360) were used as a single copy reference genes for potato and Arabidopsis, respectively, while primers for the StALSm and nptII transgenes were used to quantify the transgene copy number. The primers and probes used are shown in Supplementary Table S4. The genomic DNA template was digested with NcoI prior to use in ddPCR amplification.

The presence of the vector backbone in potato was assessed by qPCR using the SsoAdvanced Universal Probes SuperMix (Bio-Rad, Hercules, CA, USA) and following manufactures’ suggested protocol. The primers and probes used are shown in Supplementary Table S5.

3. Results

3.1. Development of the JGT105 GAANTRY Recipient Strain and Assembly of a 5-Stack T-DNA

To broaden the utility of the original GAANTRY gene stacking system [5], A. tumefaciens strain EHA105 was modified to make it a recipient for GAANTRY stacked cargos. A. tumefaciens EHA105 strain had the GAANTRY components, including the C58 Left Border (LB), an A118 attP recombinase recognition site, a bacterial kanamycin resistance gene, and a ParA MRS site (Figure 1A and Figures S1 and S2) inserted within its disarmed pTiBo542 virulence plasmid (NC_010929). This insertion was achieved through homologous recombination using flanking sequences derived from the NCBI reference pTiBo542 (NC_010929) and earlier reports [11,14], as previously described [5].

To demonstrate the capabilities of the GAANTRY JGT105 recipient strain, a 15.8 kb 5-gene stack was assembled using a stepwise recombination approach. Each cargo sequence was cloned into either the P or B donor vector (Table 1; Supplementary Figures S3 and S4), and gene stacking was performed by alternating between these vectors and selecting gentamicin and kanamycin. This five-step assembly, though it could be condensed, was intentionally structured to illustrate the iterative and reproducible nature of the GAANTRY assembly process, resulting in the final JGT5046 5-stack strain (Figure 1B).

Each inserted cargo sequence functioned as an independent transcriptional unit, eliciting a distinct phenotype, including: mStALS (mutant potato Acetolactate synthase) for imazamox resistance, uidA (β-glucuronidase with potato intron IV2), eGFP (enhanced green fluorescent protein with GBSS transit peptide), an RNAi construct targeting gibberellin 20-oxidase (GAox1) for short plant stature, and nptII (neomycin phosphotransferase II), conferring kanamycin resistance (Figure 1B,C; Table 1; Supplementary Figure S3).

In parallel, a binary vector construct (pBi5045) T-DNA was created by assembling the same five transcriptional units via restriction enzyme cloning, using the P and B donor plasmids as PCR templates. This plasmid construct was also assembled in five steps, resulting in a 14.8 kb T-DNA region (Figure 1C; Supplementary Figure S4). The difference in size between JGT5046 and pBi5045 T-DNAs is due to the presence of inactive recombinase recognition site sequences and donor plasmid multicloning site sequences located between the individually stacked cargo sequences within the GAANTRY T-DNA assembly.

Molecular validation of the JGT5046 GAANTRY assembly was conducted via genomic PCR amplification across stacked cargo junctions after each stacking step, confirming accurate and complete integration and excision (Figure 1B; Supplementary Figure S5; Table S3). The binary vector pBi5045 was fully sequence-verified prior to transformation (Supplementary Figure S4).

3.2. Generation and Phenotypic Analysis of Transgenic Plants

To evaluate transformation capability and compare performance between the two systems, the 5-gene stacks were introduced into Solanum tuberosum (cv. Ranger Russet) and Arabidopsis thaliana. More than 450 potato explants were transformed using either the JGT5046 strain or EHA105 carrying the binary vector pBi5045. Transformation efficiency, measured by rooting in the presence of imazamox and verified by qPCR (Supplementary Table S1), before being transferred to soil. Both systems showed similar efficiency at 83% for GAANTRY and 82% for the binary vector.

Eighty independent imazamox-resistant potato lines from each transformation were randomly selected for phenotypic screening, including kanamycin resistance. Wild-type (WT) and pBi4030 binary vector (Supplementary Figure S4) controls confirmed the validity of the assay conditions (Figure 2). Plants were grown to maturity and screened for introduced phenotypes (Figure 2, Supplementary Figure S5, and Supplementary Table S1). Leaf size was characterized using fully expanded, mature terminal leaves collected from the mid-canopy of plants at the vegetative–early tuber bulking stage. The pBi5045 control plant leaves ranged from 9 to 13 mm in length and 6–8.5 mm in width. The JGT5046 produced leaves 12–14 mm in length and 8–11 mm in width, while pBi5045 showed leaves sized 11–15 mm in length and 8–10 in width. Scoring of the GAox1 phenotype was determined to be unreliable through visual screening (Supplementary Figure S6) and was therefore evaluated via genomic PCR. Ultimately, 86.3% of GAANTRY and 96.3% of binary transgenic lines contained complete 5-stack cargo T-DNAs and exhibited 4 of the introduced phenotypes (Supplementary Table S1).

Figure 2. Representative phenotypes of 5-stack, mStALS control, and wild-type (WT) potato plants. (A) Explants on imazamox-containing media six weeks post transformation. (B) β-glucuronidase activity detected in histochemically stained leaves; (C) green fluorescence observed in leaves under blue light; (D) the same leaves as in (C), observed under white light. Leaves were determined to be phenotypically similar in size and shape.

Arabidopsis transformation yielded ~70 independent T₁ events per construct following kanamycin selection, with transformation efficiencies of 1.73% for the GAANTRY JGT5046 strain and 1.95% for the pBi5045 binary vector (Supplementary Table S2). Imazamox selection was only partially effective in Arabidopsis, and thus dual selection was not performed. T₁ plants were grown to maturity and screened for phenotypes (Figure 3), with GAox1 and mStALS cargo sequences being confirmed by PCR. Among the screened events, 58% (GAANTRY JGT5046) and 55.1% (pBi5045 binary vector) of lines expressed the three functional phenotypes and carried all of the introduced T-DNA sequences (Supplementary Table S2).

Figure 3. Copy number variation in the GAANTRY and binary vector 5-stack Arabidopsis and potato events. (A) The nptII, mtStALS, and complete T-DNA insertion (based on the comparison of the simultaneous presence of both nptII and mtStALS) transgene copy number was measured using droplet digital PCR in 55 or more kanamycin-resistant Arabidopsis events (see Supplementary Table S2). (B) The nptII or mtStALS transgene copy number was also measured in 49 or more potato events identified with imazamox selection (see Supplementary Table S1).

3.3. Transgene Copy Number and Backbone Transfer Analysis

Transgenic potato and Arabidopsis lines were further analyzed to determine T-DNA copy number and the presence of vector backbone sequences. Copy number analysis was performed using droplet digital PCR (ddPCR). Single-copy insertions based on the presence of mStALS at the LB were found in 54.9–63.3% of potato events and 29.3–30.9% of Arabidopsis events for GAANTRY and binary, respectively. Analysis of the nptII gene at the RB indicates single copy events at 23.5–26.5% for potato and 20.7–25.5% for Arabidopsis from GAANTRY and binary vector constructs (Figure 3; Supplementary Tables S1 and S2).

Backbone transfer—defined as the integration of DNA outside the T-DNA borders—was assessed by PCR and ddPCR in Arabidopsis and qPCR in potato. In potato, the GAANTRY strain JGT5046 showed a lower backbone transfer rate of 10.0% compared to the pBi4045 binary vector with 26.5% of events carrying plasmid sequences from outside the T-DNA. In Arabidopsis, backbone integration occurred more frequently, and the difference was less pronounced, with backbone integration detected in 37.5% of the JGT5046 lines and 48.9% of the pBi4045 binary vector-derived lines (Supplementary Tables S1 and S2).

4. Discussion

Efficient and precise delivery of multiple transgenes is increasingly critical to advancing plant synthetic biology, particularly as researchers work to engineer complex traits using gene-editing tools, morphogenic regulators, and multigene biosynthetic pathways. The GAANTRY system [5] was developed to address this need through a modular, sequential, and recombinase-based approach to gene stacking that eliminates the reliance on binary vectors. In this study, we expanded the GAANTRY platform by adapting it for use in the widely utilized Agrobacterium tumefaciens strain EHA105, creating the new JGT105 strain. This modification extends the utility of the GAANTRY assembly to a broader group of researchers who currently use the EHA105 strain and already have transformation protocols in place for genetic engineering of their plant species of interest.

The JGT105 strain harbors the GAANTRY components within the region where the native T-DNA was removed from its disarmed virulence plasmid and thus is a low-copy and an inherently stable part of the host bacterial strain’s genome. These features are known to improve the fidelity of T-DNA delivery and integration in plant cells, a persistent challenge in binary vector-based systems, where the plasmid vector can be lost or rearranged in Agrobacterium in the absence of antibiotic selection [1,2,3]. Importantly, despite the recA⁺ background of EHA105, our data demonstrate that the 15.8 kb five-gene stack in JGT5046 remained stable throughout all recombinase-mediated stacking and propagation steps, indicating that no observable DNA rearrangements or losses occurred. This addresses a critical barrier to delivering large or complex constructs, particularly in co-cultivation conditions lacking selection pressure.

The five-gene stack included transcriptional units encoding imazamox and kanamycin resistance, fluorescent and enzymatic reporters, and a GA20-oxidase RNAi element, allowing for a thorough evaluation of transformation efficiency, phenotypic expression, T-DNA copy number, and stability. Comparative transformation experiments in Solanum tuberosum and Arabidopsis thaliana revealed that the JGT5046 GAANTRY strain performs similarly to the binary vector pBi5045 in transformation efficiency and phenotype expression. Notably, the GAANTRY system led to a lower frequency of undesired backbone transfer in potato, a significant quality improvement in transgenic event generation.

Results indicate that potato transformation efficiency was similar across both systems, with 83% for GAANTRY and 82% for the binary vector. In potato, single-copy insertion frequencies, as indicated by the left border marker (mStALS), were 54.9% for JGT5046 GAANTRY and 63.3% for the pBi5045 binary vector. While analysis of the right border marker (nptII) demonstrated single insertion frequencies that ranged from 23.5% for JGT5046 GAANTRY to 26.5% for the pBi5045 binary vector (Figure 3; Supplementary Table S1). Simultaneous comparison for the presence of both marker genes (mStALS and nptII), and positive validation for the internal T-DNA cargos (GUS, EGFP, GAox-1 RNAi) indicate that complete single copy T-DNA insertion occurred in 17.6 % of the JGT5046 GAANTRY events and 24.5% for the binary pBi5045 events (Figure 3; Supplementary Table S1). Backbone integration rates were determined to be 10% for JGT5046 GAANTRY and 26.5% for the pBi5045 binary vector. These results confirm that the GAANTRY system is comparable to the more traditional binary vector system in transformation efficiency and single-copy T-DNA insertion, while offering the advantages of construct stability, the ability to handle large, complex designs, and providing lower integration of backbone sequences [5].

For Arabidopsis, the efficiency of transformation ranged from 1.73% for JGT5046 GAANTRY to1.95% for the pBi5045 binary vector (Supplementary Table S2). Single copy insertions based on the presence of the left border marker mStALS range from 29.3% for JGT5046 GAANTRY to 30.9% for the pBi5045 binary vector. Rates of the right border marker nptII ranged from 20.7% to 25.5%, similar to the potato results (Figure 3; Supplementary Table S2). Rates of backbone integration were higher than in potato at 37.5% for JGT5046 GAANTRY and 48.9% for the pBi5045 binary vector (Supplementary Table S2). Using the right border located marker for kanamycin resistance in Arabidopsis, only 58% (JGT5046 GAANTRY) to 55% (pBi5045 binary vector) of the transgenic Arabidopsis events respectively carried integration of a complete 5 cargo T-DNA (Supplementary Table S2), while imazamox selection in potato based on the left border located selection marker generated 88.8% (JGT5046 GAANTRY) and 97.5% (pBi5045 binary vector) of the events that contained complete copies of the entire T-DNA construct (Supplementary Table S1). Similar results in position-based selection marker T-DNA transfer rates were previously noted by [5].

With the ongoing expansion of plant biotechnology into precision genome editing and metabolic pathway engineering, there is a growing demand for platforms capable of delivering complex, modular, and high-fidelity transgene assemblies. The GAANTRY system is particularly well-suited to meet these emerging needs, as modern plant engineering increasingly involves large constructs incorporating CRISPR/Cas systems, multiple guide RNAs, and regulatory elements for tissue-specific or inducible expression. Additionally, the co-expression of developmental regulators such as BABY BOOM (BBM), WUSCHEL (WUS), GROWTH REGULATING FACTORS (GRFs), and GRF-INTERACTING FACTORS (GIFs) has become a standard approach to enhance transformation and regeneration in recalcitrant plant species [18]. These sophisticated applications require assembly platforms that support the stable integration of multigene cassettes with minimal risk of rearrangement or truncation. GAANTRY’s sequential, site-specific stacking mechanism provides a reliable and scalable solution for constructing such T-DNAs, offering a superior alternative to traditional binary vector systems, which often struggle with insert-size limitations and structural instability.

To further enhance GAANTRY’s modularity and integration with widely used synthetic biology platforms, we have developed a modified set of Level 1 donor vectors compatible with Golden Gate assembly using Type IIS restriction enzymes (Supplementary Figure S7), as described in the MoClo system [19]. These new GAANTRY donor vectors are fully compatible with the Phytobrick standard and the Joint Modular Cloning (JMC) Toolkit [20], enabling streamlined cloning and seamless reuse of standardized genetic parts across research groups. These Golden Gate-compatible GAANTRY donor plasmids are available upon request, providing a direct interface between GAANTRY’s high-fidelity gene stacking and the broader plant synthetic biology ecosystem.

In summary, the development of the JGT105 GAANTRY recipient strain and successful assembly and transformation of a complex five-gene stack reinforce GAANTRY’s utility as a scalable and precise alternative to traditional binary vector-based plant transformation. Its compatibility with the commonly used EHA105 Agrobacterium strain, high-fidelity T-DNA assembly and stability, and integration with standardized modular cloning toolkits make the GAANTRY system well-positioned to address the evolving needs of plant biotechnology and synthetic biology.