Split selectable marker systems utilizing inteins facilitate gene stacking in plants

The study addresses the challenge of stacking multiple genes in plants, which is essential for metabolic engineering and improving complex, polygenic traits. Agrobacterium-mediated transformation is widely used but large multigene T-DNAs are unstable in Agrobacterium and subject to truncation before insertion, making simultaneous delivery of many genes difficult. The authors propose a solution using split inteins to create split selectable marker systems that enable co-transformation of two separate T-DNAs while requiring only a single selectable marker function to be reconstituted post-translationally. Building on prior applications of split inteins (notably NpuDnaE) in protein engineering and plant genome editing, the research aims to develop and validate split Kanamycin and Hygromycin resistance markers to facilitate efficient gene stacking in Arabidopsis and poplar.

The paper situates the work within the context of: (1) the need for multigene pathway engineering in plants to create new traits and produce high-value metabolites; (2) limitations of Agrobacterium-mediated delivery of large T-DNAs due to instability and truncation of high molecular weight DNA; and (3) the emergence of intein technology, particularly split inteins such as NpuDnaE, as tools for ligating polypeptides post-translationally. Previous studies demonstrated split inteins’ versatility in protein engineering and their application to split CRISPR/Cas9 systems for plant base editing. Traditional co-transformation approaches require multiple selectable markers and careful optimization of combined selection pressures, with variable performance across markers and genotypes (e.g., HygR often outperforming KanR in some poplar genotypes). A prior report introduced split selectable markers conceptually, but this work provides, for the first time in plants, a robust demonstration of split KanR and split HygR systems for both in planta and tissue culture co-transformation in herbaceous and woody species.

• Design of split reporters and selectable markers: The authors used the split intein NpuDnaE (codon-optimized for Arabidopsis) to mediate protein trans-splicing. As a proof-of-concept reporter, RUBY (encoded by CYP76AD1, DODA, and a glucosyltransferase, GT) was split by identifying a site in GT requiring a Cys at +1 of the C-extein (L231:C232), generating GTf1 and GTf2 fused to the N- and C-terminal intein fragments, respectively. For selectable markers, they identified split sites in nptII (KanR; T131:C132 and A192:C193) and hpt (HygR; S52:C53 and Y89:C90) constrained by the obligatory Cys at the C-extein +1 position. Each selectable marker was divided into N- and C-terminal fragments (F1/F2 for one split; F3/F4 for the other) and fused to N- (IntN) and C-intein (IntC), respectively.
• Vector construction: Split RUBY vectors PAXY0006 (CYP76AD1, DODA, GTf1-IntN) and PAXY0007 (IntC-GTf2) were created. Split Kan/Hyg vectors (pAXY0008–pAXY0015) carried one selectable marker fragment fused to an intein half and, importantly, each vector also included a distinct visual reporter (either eYGFPuv or RUBY) to enable easy scoring of co-transformation. All constructs were verified by Sanger sequencing.
• Transient assay in Nicotiana benthamiana: Agrobacterium-mediated leaf infiltration tested restoration of RUBY pigmentation upon co-infiltration of the two split RUBY vectors versus each alone.
• Arabidopsis co-transformation: Floral dip co-transformation used two Agrobacterium strains (GV3101), each carrying one split selectable marker vector (split KanR or split HygR) paired with either eYGFPuv or RUBY reporter. T1 seedlings were selected on MS medium containing Kanamycin or Hygromycin. Phenotypes (green fluorescence under 365 nm UV for eYGFPuv; red pigment for RUBY) were documented at seedling and mature stages. PCR genotyping confirmed the presence of eYGFPuv and RUBY transgenes. T2 generations were analyzed for segregation, antibiotic resistance, and reporter expression.
• Poplar co-transformation: Tissue-culture-based co-transformation of Populus tremula × alba clone INRA 717-1B4 using Agrobacterium strain EHA105 and standard leaf disk protocol. Shoots expressing eYGFPuv were transferred to Hygromycin-containing root induction medium to assess functional reconstitution of HygR. Reporter phenotypes were monitored; PCR genotyping assessed presence of eYGFPuv and RUBY.
• Western blot for protein trans-splicing: In HEK293T cells, N-terminal HygR fragments (F1 and F3) were tagged with 3xFLAG and C-terminal fragments (F2 and F4) with 3xHA. Lysates from single-fragment transfections and cognate co-transfections (F1+F2; F3+F4) were probed with anti-FLAG and anti-HA to detect full-length HygR resulting from trans-splicing and to confirm intein orthogonality.
• Experimental conditions and reproducibility: Arabidopsis and poplar transformations were each performed twice; at least eight Arabidopsis plants and ~150 poplar explants per transformation. At least five Arabidopsis and fifteen poplar transgenic events were analyzed by PCR and phenotyping. Details of growth conditions, transformation parameters, primers, and imaging were provided.

• Split RUBY reporter validation: Co-infiltration of Nicotiana benthamiana leaves with the two split RUBY vectors restored strong red pigmentation, whereas either fragment alone did not, demonstrating successful intein-mediated reconstitution of a functional reporter.
• Identification of functional split sites: For nptII (KanR), split sites T131:C132 and A192:C193; for hpt (HygR), split sites S52:C53 and Y89:C90 supported intein-mediated reconstitution.
• Arabidopsis co-transformation: T1 seedlings exhibited robust antibiotic resistance on selection media (Kan or Hyg), indicating functional reconstitution of the selectable marker from two inactive fragments. Co-expression of eYGFPuv (UV-visible green fluorescence) and RUBY (red pigment) in resistant T1 plants confirmed successful co-transformation of both vectors. PCR genotyping detected both eYGFPuv and RUBY in all resistant plants tested. Traits (antibiotic resistance and reporter expression) were heritable and observed in many T2 plants, with expected Mendelian segregation of T-DNA insertions.
• Poplar co-transformation: More than 20 eYGFPuv-positive shoots were recovered; 15 were tested on Hygromycin root induction medium, with approximately 80% rooting success, indicating functional HygR reconstitution. eYGFPuv fluorescence persisted in rooted plants; typical strong RUBY pigmentation was not consistently observed, though PCR genotyping detected both eYGFPuv and RUBY in all rooted events.
• Protein-level confirmation: Western blot in HEK293T cells showed full-length HygR only upon co-transfection of cognate fragment pairs (F1+F2 and F3+F4), with no full-length product from single fragments, confirming protein trans-splicing and intein orthogonality.
• Overall: Split-KanR and split-HygR systems effectively enabled co-selection and co-transformation in both an herbaceous model (Arabidopsis) and a woody species (poplar), facilitating gene stacking without requiring multiple selectable markers.

The results demonstrate that split-intein-mediated selectable markers can reliably reconstitute functional antibiotic resistance in plants, enabling the use of a single selection regime to co-select two independent T-DNAs. This directly addresses the bottleneck of limited selectable markers and the difficulties of delivering large, multi-gene T-DNAs in one construct. By distributing genetic cargo across two vectors, the approach reduces the size and complexity of each T-DNA, alleviating issues of instability and truncation in Agrobacterium. The consistent co-occurrence of eYGFPuv and RUBY reporters in resistant Arabidopsis plants, confirmed by PCR, shows that selection for the reconstituted marker effectively enriches for co-transformation events. In poplar, efficient rooting on Hygromycin among eYGFPuv-positive shoots, together with PCR confirmation, supports applicability in tissue-culture-based transformation of woody plants, though reporter expression (RUBY) was less consistent phenotypically. Limitations include uncontrolled insertion sites for the two T-DNAs, leading to Mendelian segregation in sexually reproducing species. Nonetheless, this is analogous to classical genetic approaches where homozygosity for both inserts can be achieved by the next generation. The method also reduces repetitive promoter-driven recombination within single plasmids by splitting expression cassettes across two vectors. Collectively, the technique broadens the practical capacity for multigene stacking and pathway engineering in diverse plant systems.

This study establishes and validates split-intein-based selectable marker systems (split KanR and split HygR) that enable efficient co-transformation and gene stacking in plants. The approach was demonstrated in Arabidopsis (in planta floral dip) and poplar (tissue culture) using visual reporters (eYGFPuv and RUBY) and confirmed by phenotyping, PCR genotyping, and protein-level evidence of trans-splicing. These systems can reduce vector size and complexity, mitigate plasmid recombination caused by repetitive sequences, and potentially double the transformation capacity for multigene engineering. Future work could focus on: (1) strategies to coordinate or target co-insertion sites to minimize segregation; (2) expanding to additional selectable markers and intein pairs; (3) optimizing reporter systems (e.g., GUS, LUC) for more consistent phenotyping across species; and (4) applying the method to complex metabolic pathways and agronomic trait stacking in diverse crops.

• Independent T-DNA insertions are not co-localized, leading to Mendelian segregation of the two inserts in sexually reproducing species; homozygosity for both inserts requires additional generation(s).
• Reporter variability: RUBY pigmentation was inconsistent in poplar despite PCR detection, potentially influenced by stress responses during transformation; alternative reporters (GUS, LUC) may be more reliable in some species.
• Selectable marker performance can vary by genotype/species (e.g., HygR often outperforms KanR in some poplars), necessitating optimization of selection conditions.
• The approach does not control T-DNA copy number or insertion site, which may affect expression levels and stability across events.
• Demonstrated in two species (Arabidopsis and poplar); broader validation across diverse crops and transformation platforms remains to be performed.