Optimisation of Agrobacterium ‐Mediated Seedling Leaf Transformation Across Multiple Sorghum Genotypes

Genetic transformation in sorghum faces major challenges, including limited explant availability, phenolic compound accumulation, and strong genotype dependence. Morphogenic transcription factors (MTFs), such as Baby boom (Bbm) and Wushel2 (Wus2), markedly improve transformation efficiency using seedling leaf whorls. The Cre/loxP system subsequently removes MTF genes after somatic embryogenesis, enabling fertile plant regeneration (Wang et al. 2023). This scalable strategy performs well in Tx430 (Butler et al. 2025), though phenolic interference continues to restrict broader genotype adoption. In this study, we present an Agrobacterium-mediated MTF-assisted protocol using seedling leaf whorl in 13 economically important sorghum genotypes. This work includes the first successful report for several of these genotypes. Additionally, we demonstrate CRISPR-Cas9-mediated editing of the glossy2 (gl2) gene using the same explant system. Finally, we evaluate alternate visual markers, RUBY (He et al. 2020) and free-use GFP (fuGFP; Coleman and Somerville 2019), for transformation tracking. Thirteen sorghum genotypes were selected to evaluate the Agrobacterium-mediated seedling leaf whorl transformation protocol (Supporting Information). Tx430 served as a positive control due to its established transformability and low phenolic compound production; Tx623 represents the reference genome; IA100RPS and IA101RPS are sister inbred lines developed for biomass breeding in the northern latitudes (Salas-Fernandez and Kemp 2022). The remaining nine genotypes are part of the Sorghum Association Panel (SAP), representing broad genetic diversity and agronomic relevance. Figure 1a summarises the transformation frequency (TF) and efficiency (TE) across the 13 genotypes. In this study, TF is the percentage of rooted regenerants per total number of infected leaf whorl explants. Using the standard eight-step procedure (Supporting Information; Figure S1), Tx430 consistently produced regenerants at a high frequency, and the transgene was stably inherited into the next generation (Figure S2). Across 10 independent experiments involving 115 explants, a total of 483 regenerants were obtained, yielding a TF range of 33% to 3620% with an average of 420% ± 230% (mean ± SE; Figure 1a). Similar trends were observed in IA101RPS, ICSV400 and SC51, with average TFs of 1269% ± 594%, 940% and 181% ± 89%, respectively (Figure 1a, easy). However, the remaining nine genotypes exhibited variable tissue culture responses, primarily due to genotype-dependent production of phenolic compounds (Figure 1b; Figure S3). While the colour and intensity of phenolic compound accumulation varied among genotypes, the effects were consistently detrimental, leading to tissue browning, necrosis, and reduced regeneration potential. As a result, protocol optimisation was necessary to improve regeneration outcomes in these lines. In CK60, SC329, SC673 and Tx623, explants produced moderate levels of phenolic compounds, causing partial tissue necrosis and reduced regeneration, while some explants, calli, or regenerating shoots progressed to rooting. To boost survival, two extra subcultures were added to reduce the buildup of phenolic compounds. The second maturation step was shortened from 14 days to 10 days, and a third 10-day subculture was then implemented, continuing until shoots formed. Likewise, in the rooting phase, a second subculture step was added after 10 days to mitigate phenolic-induced tissue necrosis. These modifications enhanced explant viability and regeneration, resulting in TFs ranging from 17% to 290% ± 90% across these four genotypes (Figure 1a, moderate). In P898012, IA100RPS, San Chi San, SC265 and SC1345, explants produced markedly high levels of phenolic compounds, leading to severe browning and necrosis (Figure 1b, images 5–8). Initial attempts using the standard eight-step subculture protocol failed to yield regenerants in IA100RPS and P898012 (Figure 1b, images 5 and 6). To address this, the protocol was extended to a minimum of 14 subcultures or more to improve explant survivability. Specifically, explants were transferred to fresh selection media every 10 days, instead of 21-day intervals. In the maturation stage, subculture was performed every 7–10 days to reduce phenolic compound buildup. This strategy was carried out to the rooting stage. This labour-intensive but effective modification greatly improved callus survival and regeneration, yielding TFs of 46% (IA100RPS), 532% ± 160% (P898012), 293% (San Chi San), 57% (SC1345) and 260% (SC265), respectively (Figure 1a, Difficult). This metric helps researchers evaluate the practical feasibility of working with specific genotypes. For instance, P898012 achieved a TE of 38%, indicating that ~40% more effort is needed compared to Tx430, which had a TE of 53% (53/38 = 1.4; Figure 1a). Thus, while TF may appear similar, TE provides a more realistic measure of transformation practicality. We next tested CRISPR-Cas9-mediated genome editing using the sorghum seedling leaf whorl as explants. A single gRNA construct, pKL2536 (Figures S4 and S5), targeting the second exon of the gl2 gene in Tx430, was introduced into three LBA4404 derivative strains (Figure S6): (1) LBA4404Thy- (= LTA), (2) LBA4404T1 (= LT1A) and (3) an engineered Agrobacterium LBA4404Thy- strain carrying an insertional mutation in the recA gene to improve plasmid stability (= LTR1A; Method S1). These Agrobacterium strains were each carrying the virulence gene helper plasmid, pKL2299A (Aliu et al. 2024). During the preliminary experiment, targeted indel mutations were detected from 26% to 58% of the randomly selected transgenic calli samples (Figure S6). Two independent T0 plants were later regenerated from the calli transformed with the LTR1A strain. Sanger sequencing and ICE analysis (See Method S1) showed that both plants had short indel mutations (−3 bp/−3 bp) at the target site (Figure 1c, image 2). Due to the continuous Cas9 activity, additional indel mutations were observed in the T1 seedlings such as a 16 bp deletion plus a 119 bp insertion (55 bp + 59 bp + 5 bp) at the gl2 target site (Figure 1c, image 2). gl2 encodes a key enzyme involved in waxy cuticle biosynthesis on the juvenile leaf surface. As expected, biallelic and homozygous loss-of-function mutant T1 plants exhibited the gl2 knockout phenotype, with water droplets adhering to leaf surfaces upon misting (Figure 1c, image 1). Lastly, we demonstrated the successful expression of two alternative visual markers, RUBY and free-use GFP (fuGFP) in Tx430, confirming their utility for monitoring transformation events in sorghum seedling leaf whorls (Figure S7). K.W. conceived the project idea. M.K.A., K.L. and K.W. designed the project. K.L. designed and constructed the vectors and Agrobacterium strains. M.K.A. performed the sorghum transformations. M.K.A. and K.L. conducted molecular analysis. M.K.A., K.W. and K.L. wrote the manuscript. The authors thank William Gordon-Kamm and Ning Wang for their support and guidance with the protocol and Agrobacterium strain, Jianming Yu, Maria Salas-Fernandez and Gregory Schoenbaum for sorghum seeds and technical assistance. This work was partially supported by the National Science Foundation (NSF) award 2121410 to K.W. and K.L., by NSF IOS-2341535 to K.L., and by the National Institute of Food and Agriculture of the United States Department of Agriculture Hatch project #IOW05768, by State of Iowa funds. K.W.'s contribution to this work is partially supported by (while serving at) the NSF. The data that support the findings of this study is available in the Supporting Information of this article. Data S1: pbi70438-sup-0001-Supinfo.pdf. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.