Assembly of the 5 and 11-Stack GAANTRY Strains
The GAANTRY transgene stacking method was used to sequentially stack five or eleven sequences within the Agrobacterium virulence plasmid T-DNA (Fig. 1 and 2a). The selected cargo sequences (Table 1) were cloned into a P or B Donor plasmids, and multigene stacking within the GAANTRY ArPORT1 strain was achieved by alternating the use of B and P Donor vectors and gentamicin and kanamycin selection to stack the genes of interest into the recipient GAANTRY strain as previously described (Collier et al. 2018). The donors contained transcriptional units that were designed to confer functional phenotypes (i.e. antibiotic/herbicide resistance or reporter gene activity) and one cargo (which appears twice in the 11-stack T-DNA) was an enhancer-blocking insulator. The 5-stack GAANTRY strain contains the hptII, Renilla luciferase, TBS insulator, firefly luciferase and GUSPlus cargoes, while the bar, eGFP, TBS insulator, tdTomato, EPSPS and nptII cargo sequences were added to the 5-stack to generate the 11-stack GAANTRY strain (Table 1).
The Donors carrying each cargo sequence were iteratively stacked in either five or eleven independent steps to produce the 5-stack (16.9 kb) and 11-stack (37.4 kb) T-DNAs respectively (Fig. 1 and 2a). Although the 5- and 11-stack T-DNAs could have been constructed in fewer steps by designing donors that contained multiple cargos using Golden Gate-compatible and Gateway®-compatible auxiliary donor vectors (Collier et al. 2018), the 5 and 11 step pathways were chosen to demonstrate the repetitive and efficient nature of the GAANTRY assembly system. Thorough molecular characterization of the assemblies was performed on three randomly selected Agrobacterium colonies via genomic PCR using gene-specific primers that span each of the recombination junctions between the cargo sequences (Fig. 2). All of the junctions in all tested colonies during each round of assembly were validated by genomic PCR screening, demonstrating the efficiency and reliability of the GAANTRY-mediated cargo insertion process. Approximately equal number of colonies were obtained during each round of transformation from stack 1 to stack 11.
Stability of the Stacked Cargo Sequences
To determine the stability of the stacked cargo within the ArPORT1 GAANTRY strain pRi plasmid, the 5-stack and 11-stack strains were subcultured in nonselective LB broth medium every 24 h over a period of 6 days. Aliquots of the final culture were then plated on nonselective LB agar, and one hundred random colonies were screened using gentamicin selection. All one hundred colonies exhibited antibiotic resistance; ten of these were randomly selected for validation with genomic PCR and were confirmed to carry all expected cargo sequences. Representative PCR results are shown in Fig. 2b for an individual 11-stack clone.
Generation and Characterization of 5-Stack and 11-Stack Transgenic Rice Plants
To examine the ability of the GAANTRY strains to generate transgenic rice, Nipponbare wild-type embryogenic rice callus was co-cultured with either the 5 or 11-stack GAANTRY strains. Transgenic calli was identified using hygromycin antibiotic selection. Sixteen and thirty-seven independent transgenic (T0) events were recovered using the GAANTRY 5- and 11-stack strains respectively. The T0, T1 and T2 generations of each event were grown in the greenhouse to determine the functionality and stability of the introduced T-DNAs. Phenotypic and genotypic results indicate stable inheritance of the transgenes in each of the transgenic lines with no obvious differences observed in morphology or growth between the genetically engineered and wild-type Nipponbare rice plants.
The phenotypic expression of the introduced traits was analyzed in for each of the transgenic events. Cargo sequences 1 and 11 (Table 1) confer antibiotic resistance. The hptII gene is under the control of the rice ubiquitin 2 promoter and nptII expression is conferred by the maize ubiquitin 1 promoter; these genes confer resistance to hygromycin and paromomycin respectively (Collier et al. 2016). Germination assays were performed for T1 seed using antibiotic selection. As expected, all transgenic events produced progeny that were resistant to hygromycin, consistent with that antibiotic being used to select the original T0 transgenic events. A representative example of the hygromycin resistance observed is shown in Fig. 3a. The T1 seed from all 37 of the 11-stack events were germinated on paromomycin media and 73% had progeny that exhibited resistance (Supplemental Table 3; Fig. 3g). Cargo sequences 2 and 4 were expected to confer Renilla and firefly luciferase activity (Table 1). A dual-luciferase® activity assay was performed on leaf tissue extracts to assess which events had functional expression for each reporter gene. All 5-stack events exhibited Renilla luciferase activity and 81% had detectable firefly luciferase activity. Genomic PCR on the three events lacking activity confirmed that those plants also lacked at least a portion of the firefly luciferase expression cassette (Fig. 4, Supplemental Table 2). For the 11-stack events, 81% had detectable Renilla luciferase activity and 62% were positive for firefly luciferase activity (Fig. 3b-c, Supplemental Table 3). Similar to the 5-stack transgenic events, all of the 11-stack events that did not have luciferase activity also lacked that portion of the 11 stack T-DNA (Fig. 5, Supplemental Table 3).
Cargo sequences 3 and 8 were the Transformation Booster Sequence (TBS) from Petunia (Hily et al. 2009). This sequence is expected to block interactions between the CaMV 35S enhancer and nearby transgenes within the T-DNAs, thus maintaining the fidelity of their organ- or tissue-specific expression. The fifth cargo sequence was the GUSPlus expression cassette (Table 1) where expression of the reporter was controlled by the rice organ-specific Leaf Panicle 2 promoter (Thilmony et al. 2009). For the 5-stack events, 56% exhibited the expected green tissue-specific β-glucuronidase activity via histochemical staining (Supplemental Table 2). Approximately 54% of the 11-stack events had detectable GUSPlus reporter activity only in leaves and other photosynthetic tissues (Fig. 3d, Supplemental Table 3). Most of the events lacking activity also lacked the GUSPlus portion of the T-DNA based on genomic PCR and ddPCR screening, although two of the 5-stack events and two of the 11-stack events carried at least a part of the GUSPlus construct sequence but failed to produce detectable histochemical staining (Figs. 4, 5, Supplemental Tables 2, 3).
The sixth cargo sequence in the 11-stack T-DNA was the bar gene under control of the constitutive switchgrass Ubiquitin 1 promoter (PvUbi1) (Mann et al. 2011). Finale® herbicide was applied to individual leaves and 57% of the events showed tolerance (Fig. 3e). The events that were as sensitive to Finale® herbicide as the wildtype Nipponbare plants lacked the region of the T-DNA with the bar cargo sequence (Fig. 5, Supplemental Table 3). The seventh cargo sequence contained the eGFP (enhanced Green Fluorescence Protein) gene with its expression controlled by the rice Root6 promoter which controls root-tip specific expression (Xing et al., unpublished). A total of 22 of the events (60%) exhibited root-tip specific green fluorescence and carried that portion of the 11-stack T-DNA (Fig. 3f, Supplemental Table 3). Cargo sequence 9 carried the red fluorescent tdTomato reporter gene under control of the rice PS2 (also known as OsLPS3) pollen-specific promoter sequence (Oo et al. 2014). Even though 60% of the transgenic events carried the complete tdTomato expression cassette based on genomic PCR, none of the events exhibited detectable red fluorescence in pollen, or any other tissues (Fig. 5, Supplemental Table 3). Since the tdTomato transgene failed to generate a detectable phenotype in any of the 11-stack events, this sequence was considered a nonfunctional cargo sequence. The tenth cargo sequence was the 5-enol-pyruvylshikimate-3-phospate synthase (EPSPS) gene under the control of the rice GOS2 promoter (de Pater et al. 1992). Seedling germination assays determined that 68% of the events were tolerant to Round-up herbicide and genomic PCR screening confirmed that the 12 events that lacked herbicide tolerance, also lacked the EPSPS portion of the 11-stack T-DNA (Supplemental Table 3). Similarly, T1 seeds from each of the 11-stack events was germinated on paromomycin and 73% of the events exhibited antibiotic resistance and carried the 11th cargo sequence (Supplemental Table 3). Additionally, to confirm that the 11-stack plants can be both antibiotic and herbicide tolerant, a germination assay was performed with media that simultaneously contained hygromycin, paromomycin, glufosinate and glyphosate. An image of a sensitive wildtype Nipponbare seedling and a resistant/tolerant 11-stack seedling is shown in Fig. 3g.
Overall, for the GAANTRY 5-stack events, 56% contained all four expected phenotypes (Fig. 4). Of the remaining events, 43% were lacking β-glucuronidase and 13% were missing firefly luciferase and β-glucuronidase activity. In the case of the GAANTRY 11-stack events, 51% contained all eight functional phenotypes and appeared to have at least one copy of the entire 37 kb T-DNA. The remaining 49% of the events were missing one or more of the functional phenotypes and two or more of the cargo sequences (Supplemental Table 3). The phenotypes near the center of the T-DNA were the most frequently lacking in this set of events. The observed phenotypic and genotypic results from the 5- and 11-stack events remained stable in the T2 generation, indicating stable inheritance and expression of the integrated transgenes in each event.
Quantification of Transgene Copy Number and Detection of Sequences from outside the T-DNA
The transgene copy number was measured using droplet digital PCR (ddPCR) in T0 transgenic 5-stack and 11-stack events. A summary of the results is shown in Fig. 6. For the 5-stack events, 38% were shown to likely have one copy of a complete T-DNA (i.e. they had 1 copy of both the hptII and GUSPlus transgenes and exhibited all four of the expected phenotypes) (Supplemental Table 2). For the GAANTRY 11-stack events, 19% were high-quality/desirable lines carrying one copy of the hptII, GUSPlus and nptII cargo sequences, and exhibited all eight of the functional phenotypes (Figs. 3 and 5, Supplemental Table 3).
Overall, the copy number for the hptII transgene was quite low, with 75% and 81% of the 5-stack and 11-stack events having a single copy, respectively (Fig. 6, Supplemental Tables 2, 3). However, only 38% of the 11-stack events had a single copy of the nptII cargo sequence. Furthermore, ddPCR results confirmed the lack of portions of the T-DNA that correlated with the missing phenotypes. The transgenic 5-stack and 11-stack GAANTRY rice plants were also screened by genomic PCR for the presence of introduced ‘backbone’ sequences. One 5-stack event (transgenic event 7; Fig. 4, Supplemental Table 2) had DNA from outside T-DNA left border, while the remaining 52 events were free of backbone sequences (Supplemental Tables 2 and 3).