Main
Agrobacterium-mediated transformation (AMT) is an indispensable tool in both plant and fungal biotechnology for transgene insertion into target cells 1,2,3 . By replacing the native oncogenic genes used by the bacterial pathogen Agrobacterium tumefaciens with user-defined DNA sequences, the innate DNA transfer capacity of this bacterium can be exploited for untargeted DNA insertions into diverse plant, fungal and mammalian cell lines 4,5,6,7 . Initial improvements to AMT involved relocating two DNA sequences known as the left and right border (LB and RB) from the native ~200-kb tumor-inducing (Ti) plasmid to a smaller helper plasmid known as a binary vector 8 . Any DNA contained between the LB and RB is mobilized as a transfer DNA (T-DNA) and inserted into target cells with help from virulence (vir) genes contained within the Ti plasmid, enabling tractable engineering by cloning different sequences into the T-DNA region 1 . The simplicity of changing the transgene target by customizing the sequence between the LB and RB has made AMT an important tool for agricultural biotechnology, bioenergy crop engineering and synthetic biology.
Despite the genetic revolution that AMT initiated, efficient transformation is still a considerable bottleneck for genetic engineering in most plant species. Decades of optimization of AMT have identified induction conditions 9,10 , strains of A. tumefaciens 4 and enhancements to vir gene expression that have increased AMT efficiency in numerous plant species 11 . In addition to the transgene-harboring binary vector, some protocols have used a second introduced plasmid, termed a ternary vector, which overexpresses various genes involved in Agrobacterium virulence to improve AMT efficiencies in recalcitrant plants such as maize 11 . Such work has inspired more recent research to completely refactor the Ti plasmid that harbors the vir genes 12 , laying the groundwork for fine-tuned engineering of individual virulence components to further enhance AMT. Even with these improvements, transformation efficiencies remain low for most genetic backgrounds and limit the scale and throughput of genetic engineering projects, presenting a need for improved tools to better control and enhance the AMT process.
While many improvements to AMT have focused on either altering vir gene expression or sequences within the T-DNA 1,2,13,14 , relatively little work has been conducted to engineer the binary vector backbone itself. Binary vectors consist of a T-DNA region, a bacterial selectable marker and an origin of replication (ORI) to enable proper replication and maintenance of the plasmid in Agrobacterium. Within plasmid terminology, the term ORI refers to both the origin of vegetative replication (oriV, as cataloged by Dong et al. 15 ) where plasmid replication initiates and the coding sequences for specific proteins that bind to the oriV or host replication factors to regulate plasmid copy number, stability and partitioning such as Rep, Stb and Par proteins, respectively. In this manuscript, the term ORI refers to the entire DNA sequence that mediates plasmid replication, which includes the oriV and trans-acting factors such as Rep proteins 16 . For binary vectors, the ORI, which dictates both host range and copy number, is known to impact AMT efficiency in plants 16,17,18,19 . Zhi et al. postulated a direct relationship between binary vector copy number and maize transformation efficiency through a comparison of three binary vector ORIs, suggesting that higher copy numbers could be used to improve AMT 17 . A similar comparison of four ORIs conducted by Oltmanns et al. in Arabidopsis thaliana and a different cultivar of maize uncovered a more complicated relationship that implicated the strain of A. tumefaciens, the ORI and the plant genetic background as factors that independently influence transformation efficiencies 18 . Notably, however, neither study evaluated the impact of copy number variants within the same ORI, making it difficult to isolate the impact of copy number alone while controlling for intrinsic differences inherent to each ORI. While there is evidence that different ORI copy numbers influence AMT efficiency, there has never been an attempt to systematically modify this variable for the binary vector. The best attempt to date at direct manipulation of the binary vector ORI to alter AMT outcomes was by Vaghchhipawala et al., who evaluated a single higher-copy-number mutant of the pRi ORI and found no improvement in stable transformation 19 . Thus, a more comprehensive screen of ORI variants is needed to better understand the relationship among ORI identity, plasmid copy number and AMT transformation outcomes.
In prokaryotic synthetic biology, it has long been known that the copy number of a plasmid can dramatically impact engineered metabolic pathways and the functionality of synthetic circuits 20,21,22,23,24 . As plasmid copy number influences transgene expression magnitude and metabolic load, previous work has been conducted to engineer plasmid copy number, particularly in narrow-host-range ORIs commonly used in Escherichia coli such as pMB1 and pSC101 derivatives 21,25,26 . For broad-host-range ORIs that are used within A. tumefaciens, engineering efforts have been extremely limited with only a few mutants isolated 20,27,28,29,30 . To evaluate the impact of plasmid copy number across multiple broad-host-range ORIs used for AMT, a high-throughput screen is needed to identify and characterize numerous copy number variants. To date, however, no generally applicable method exists that can be applied to diverse ORIs to systematically screen for copy number diversity.
To specifically evaluate the impact of binary vector copy on AMT, we sought to generate copy number variants in four broad-host-range origins (RK2, pVS1, pSa and BBR1) that have been widely used in Agrobacterium and prokaryotic biotechnology more broadly. To accomplish this, we leveraged a high-throughput growth-coupled selection assay to rapidly identify ORI mutations that influence copy number. Across all four origins, we were able to use a transient expression assay in Nicotiana benthamiana to rapidly screen for mutants that could improve AMT efficiency. Using top candidate plasmid variants from this screen, we were able to notably improve stable transformation in both A. thaliana and the oleaginous yeast Rhodosporidium toruloides. Thus, our work demonstrates the impact of binary vector backbone engineering on AMT across kingdoms, creating an easily deployable strategy to improve transformation.
Results Directed evolution pipeline to diversify plasmid copy number
As previous work suggested a relationship between binary vector copy number and AMT transformation efficiency 17 , we sought to develop a method to systematically select for higher-copy-number mutants across diverse ORIs. Despite numerous intrinsic differences, many ORIs use analogous but nonhomologous replication initiation proteins to control plasmid replication and copy number. These nonhomologous RepA proteins share a general function of binding in cis to a motif within the oriV of the plasmid ORI to recruit various endogenous bacterial factors to enable plasmid replication 31 . Previous studies demonstrated that mutations impacting RepA proteins can alter plasmid copy number for diverse ORIs including pSC101 (ref. 21 ), BBR1 (ref. 20 ), RK2 (refs. 29,30 ) and pVS1 (ref. 32 ). To leverage this general property, the entire repA open reading frame (ORF) was randomly mutagenized using error-prone PCR (epPCR) for the pVS1, pSa, RK2 and BBR1 ORIs. These ORIs are derived from unique incompatibility groups and their respective RepA proteins share no sequence homology (Supplementary Fig. 1). For each ORI, these mutagenized ORFs were then built into a selection vector and ~100,000 colonies were pooled per ORI to create four mutant libraries within the C58C1 strain of A. tumefaciens.
To select for higher-copy-number variants within these libraries, we use a directed evolution assay that coupled plasmid copy number to bacterial antibiotic tolerance (Fig. 1). This survival-coupled selection identified wild type (WT)-lethal conditions that were permissible for growth of the mutant population, enriching higher-copy-number mutants within this population (Supplementary Fig. 2). For each selected library, the entire population’s repA ORFs were sequenced using an Illumina MiSeq alongside unselected library controls to evaluate the selective pressure of each RepA residue on survival.
Fig. 1: Directed evolution pipeline to generate plasmid copy number variants using next-generation sequencing.
a, Schematic of the repA mutagenesis method for the four ORIs. A selection vector was designed with a gentamicin resistance gene driven by a salicylic-acid-inducible promoter to enable selection in a checkerboard assay. The repA ORF was mutagenized with epPCR and ~100,000 mutants per ORI were pooled to create a mutant library. b, Example checkerboard data from pVS1 (other ORI data in Supplementary Fig. 2) along with a depiction of the population sequencing strategy. Selection conditions within the checkerboard assay that were only permissible for growth of the mutant library are shown in purple; conditions that permitted growth for both the WT and the mutant populations are shown in brown.
Full size image
This whole-population sequencing approach enabled the quantification of mutant enrichment at every residue of diverse RepA proteins (Fig. 2 and Supplementary Fig. 3). By comparing mutant distribution frequencies between the unselected and selected population, residues that contributed to the survival of the mutants in WT-lethal conditions can be identified. For the RK2 and BBR1 ORIs, large fold-change enrichments for specific nucleotide positions within the selected population were found that reached 63.8-fold and 26.7-fold over the unselected population. The pVS1 and pSa ORIs had notably lower fold-change enrichments reaching 7.2-fold and 4.4-fold, respectively. Across the four ORIs, we observed>2-fold enrichment for 38, 118, 126 and 150 nucleotide positions for pSa, pVS1, BBR1 and RK2. Mutations that were significantly enriched in the selected population compared to the unselected control were putatively associated with a higher-copy-number phenotype (Supplementary Table 1).
Fig. 2: Enriched residue sites after copy number variant selection.
Left, AlphaFold structures of the RepA proteins from each of the four ORIs screened: pVS1 (a), pSa (b), BBR1 (c) and RK2 (d). Right, primary structure of each protein with each box corresponding to a single residue. The shading of these boxes corresponds to the fold-change enrichment of a mutation impacting this residue within the selected population compared to the unselected control. Any residue that had an enrichment above a cutoff threshold for yielding the top ~20 residues is marked with an ‘X’, representing the sites with the strongest selective pressure. These residues are depicted as a purple orb on the AlphaFold models. These models focus on the structured dimerization interface, and unstructured N-terminal and C-terminal tails were trimmed for compaction. These tails, along with a quality of model analysis, are depicted in Supplementary Fig. 5.
Full size image
Mapping selected residues onto a RepA AlphaFold-generated model showed a significant enrichment of mutation sites on the predicted dimerization interfaces for all four ORIs (Supplementary Fig. 4). Some RepA proteins are postulated to regulate plasmid copy number through a ‘handcuffing’ process in which monomeric RepA promotes plasmid replication while the dimerized form inhibits further polymerase activity, finalizing plasmid copy number 33 . It has been hypothesized that mutations that weaken the affinity of RepA to itself may reduce dimerization and enable additional replication, resulting in a higher final plasmid copy number 21 ; the enrichment of selected residues on the dimerization interface for the four nonhomologous RepA proteins screened supports this notion. From this dataset, we chose ~20 residues per ORI that were found to be highly enriched for a mutation corresponding to an amino acid substitution to further characterize copy number and AMT performance in a N. benthamiana transient expression assay.
ORI copy mutants enhance plant transient transformation
Residues that were found to be significantly enriched in our selection were cloned into uniform plant expression binary vectors. For a given ORI, identical binary vectors consisting of a constitutive plant promoter driving green fluorescent protein (GFP) were created, varying by one single-nucleotide polymorphism (SNP) within the repA ORF corresponding to one of the selected mutations. These vectors were transformed into the EHA105 strain of A. tumefaciens and used to screen the impact of single repA SNPs in a transient expression assay in N. benthamiana (Fig. 3a). Because the binary vectors for a given ORI were identical except for a single SNP in repA, differences in plant GFP output were attributed to the impact of this SNP on AMT efficiency. A total of 71 candidates across all origins were screened, and mutants for all four ORIs were identified that significantly increased GFP output relative to their WT forms (Fig. 3b and Supplementary Table 1).
Fig. 3: Screening RepA mutants for improved N. benthamiana transient transformation.
a, Schematic of the transient expression assay in N. benthamiana used to screen the 71 mutants b, Box plots depicting the measured GFP output in N. benthamiana leaf discs for each construct (n = 48). Each point on the plot is the measured GFP fluorescence intensity of a single leaf disc. Boxes shaded in green were determined to have significantly higher GFP expression than the WT origin construct by a two-sided Tukey’s honestly significant difference (HSD) test (P
Information contained on this page is provided by an independent third-party content provider. This website makes no warranties or representations in connection therewith. If you are affiliated with this page and would like it removed please contact editor @cedarcity.business