# Source paper archive - Source title: Developmental regulator-mediated transcriptional reprogramming reveals novel regulators of somatic embryogenesis in sorghum - Author(s): Dhiman, P., D'Agostino, L., Deshmukh, V., Herrera Estrella, L., Janga, M., Jiao, Y., Patil, G. B. - Source landing URL: https://www.biorxiv.org/content/10.64898/2026.11.07.764635v1 - PDF URL: https://www.biorxiv.org/content/10.64898/2026.11.07.764635v1.full.pdf - PDF SHA-256: add4b9fd9e5a7420e4833482a3040baaed72f38616a3e0f7d273c8eebeecbb6e - Converter: pymupdf4llm 0.0.27 - Limits: Automated PDF text extraction; reading order, equations, tables and formatting may be incomplete or incorrect. Images are omitted, and no OCR was performed. Markup and links are neutralized for safe downloading. Consult the original PDF for the authoritative version. --- [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1 **Developmental regulator-mediated transcriptional reprogramming reveals novel** 2 **regulators** **of somatic embryogenesis in sorghum** 3 Pallavi Dhiman, Leonidas D’Agostino, Vaishnavi Deshmukh, Luis Herrera-Estrella, 4 Madhusudhan Janga, Yinping Jiao [*], Gunvant B Patil [* ] 5 Institute of Genomics for Crop Abiotic Stress Tolerance, Department of Plant and Soil Science, 6 Texas Tech University, Lubbock, TX, 79409, USA 7 ***Corresponding author:** Yinping Jiao: yijiao@ttu.edu and Gunvant B. Patil: 8 gunvant.patil@ttu.edu 9 10 **Abstract:** 11 Recalcitrance to genetic transformation remains a fundamental bottleneck in crop improvement, 12 driven by genotype-specific barriers that include entrenched defense transcriptional states, 13 limited embryogenic competence, and inadequate hormonal reprogramming potential. Here, we 14 present a multi-layered transcriptomic analysis of somatic embryogenesis (SE) and regeneration 15 in two sorghum genotypes (forage Tx2910 and grain Tx2783), two explant types (immature 16 embryos and leaf bases), and three developmental regulator (DR) treatments (GRF-GIF, WUS, 17 and BBM+WUS). This design lets us separate genotype, explant, and DR-specific regulatory 18 programs. Using in-depth transcriptomics analysis (bulk RNAseq and single nucleus 19 transcriptomics), we identified the molecular logic by which DRs override genotype-specific 20 barriers and redirect somatic cells toward embryogenic and regeneration-competent states. 21 Compared to Tx2910, the transformation-recalcitrant sorghum line Tx2783 shows a stronger 22 defense response to both _Agrobacterium_ infection plus culture stress and depends more heavily 23 on DR-driven transcriptional reprogramming to exit this defense-locked state. WUS acts as an 24 early, broad suppressor of defense and stress signaling while activating fate-specification 25 programs, whereas GRF-GIF drives a delayed but pronounced transcriptional shift toward cell 26 proliferation and morphogenesis. Integrative gene network analyses further identified a 27 prioritized set of sorghum-specific transcription factors with potential to increase regeneration 28 efficiency. Among the top candidates, we identified that _SbEREB167_ and _SbEREB174_ 29 (AP2/EREBP family) efficiently induce SE and regeneration in sorghum. Collectively, this study 30 establishes the transcriptional architecture underlying DR-mediated transformation across 1 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 31 sorghum types and identifies novel regulators that hold potential to improve plant transformation 32 across recalcitrant cereal species. 33 **Keywords:** sorghum, regeneration, developmental regulators, gene-editing, transcriptomics 34 **Introduction** 35 Sorghum is a resilient and resource-efficient crop with diverse uses, including food, fuel, fodder, 36 and forage. The rich genetic resources in sorghum provide great potential for improvement, and 37 biotechnological advances have potential to accelerate this trait improvement process (Park _et_ 38 _al._, 2025; Scully _et al._, 2018). While traditional genetics have identified many genes linked to 39 traits such as seed nutrition, stress tolerance, and disease resistance, few have been functionally 40 characterized (Lin _et al._, 2022; Varoquaux _et al._, 2019; Wang _et al._, 2026). High-throughput 41 genotyping, mutant populations, and functional genomics have helped bridge this gap; however, 42 the overall process of trait discovery and deployment in sorghum remains slow. Genetic 43 transformation is an essential tool for both functional genomics and crop improvement, enabling 44 precise manipulation of genes to study their roles and enhance agronomic traits. Sorghum, like 45 many other monocots, has historically been considered recalcitrant to _Agrobacterium_ -mediated 46 transformation (Nigam _et al._, 2025).Over the years, plant transformation has advanced from 47 particle bombardment to optimized _Agrobacterium_ -mediated methods, improved tissue culture 48 protocols, and the use of DR genes. Recent studies leveraging these advances have led to notable 49 enhancements in sorghum transformation efficiency, paving the way for more effective genetic 50 engineering and trait improvement in this globally important crop (Aregawi _et al._, 2022; ‐ 51 Fontanet Manzaneque _et al._, 2024; Park _et al._, 2025; Wang _et al._, 2023b). 52 53 The synergistic expression of DR genes such as _Wuschel2_ ( _WUS2_ ), _BabyBoom_ ( _BBM_ ), and 54 _GROWTH-REGULATING FACTOR 4 (GRF4)_ and its cofactor _GRF-INTERACTING FACTOR 1_ 55 _(GIF1)_ has been demonstrated to enable efficient, genotype-independent transformation 56 in several monocots, including sorghum and recalcitrant maize inbred lines, and has ‐ 57 outperformed conventional methods(Debernardi _et al._, 2020; Lowe _et al._, 2016; Nelson 58 Vasilchik _et al._, 2018). Additionally, these DR-utilizing methods can be effectively implemented 59 in alternative explant sources, such as leaf segments, reducing reliance on immature embryos and 60 shortening the overall tissue culture timeline (Azanu _et al._, 2025; Butler _et al._, 2025; Wang _et al._, 2 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 61 2023b). Additionally **,** strategies have been developed to mitigate the pleiotropic developmental 62 and reproductive defects caused by constitutive expression of these stem cell developmental 63 regulators, such as Cre/loxP-mediated excision under heat shock or desiccation 64 inducible promoters. This advance enables the removal of DR genes post 65 transformation, yielding normal, fertile plants while simultaneously accelerating tissue culture 66 timelines and improving transformation efficiency (Mookkan _et al._, 2017). 67 Furthermore, advances in sorghum transformation have been largely driven using DR genes 68 combined with optimized vector systems. The ternary vector system, incorporating 69 the pVIR accessory plasmid, enabled improved transformation frequencies in the sorghum 70 reference line Tx430 (6% to 29%) and was successfully adapted to recalcitrant African 71 varieties, facilitating the first stable CRISPR/Cas9-mediated gene knockouts in Tx430 (Che _et_ 72 _al._, 2018; Che _et al._, 2022). Subsequent studies showed that _ZmWUS2_ can induce rapid SE 73 formation, enabling the production of high-quality, morphogenic gene-free, and selectable 74 marker-free plants, highlighting the genotype-independent potential of DR-mediated 75 transformation. Beyond Tx430, DRs combined with the ternary vector system markedly 76 improved transformation efficiencies in previously non-transformable sorghum genotypes such 77 as Tx623 and Tx2752 (Che _et al._, 2022). Building on maize leaf transformation strategies, Wang 78 et al. (2023 b) used a maize-optimized WUS2/BBM construct with novel promoter combinations 79 to generate embryogenic callus and plantlets from sorghum leaf explants (Wang _et al._, 2023b). 80 Further refinements, such as the GRF/GIF ternary vector system pVS1_VIR2, achieved high 81 efficiency transformation in Tx430 without pleiotropic effects or the need for excision (Li _et al._, 82 2024). Optimizations in _Agrobacterium_ strains, selection regimes, and binary vector design ‐ 83 further enhanced transformation efficiencies (Fontanet Manzaneque _et al._, 2024). Innovative 84 approaches, including viral-mediated, altruistic delivery of DRs, enabled highly efficient leaf 85 based transformation in Sorghum (Butler _et al._, 2025). Further, scalable seedling leaf whorl 86 transformation mitigated phenolic accumulation and overcame genotype dependence across 13 87 genetically diverse sorghum genotypes (Azanu _et al._, 2025). Collectively, these studies 88 underscore the transformative impact of DR genes, enabling broader genotype applicability and 89 paving the way for efficient functional genomics and genetic improvement in sorghum. 3 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 90 An underexplored question in this field is whether genotype-dependent recalcitrance reflects a 91 simple quantitative deficit in regenerative capacity or a qualitative transcriptional state 92 dominated by persistent _Agrobacterium_ -induced defense signaling and stress-response programs 93 that must be actively overridden before an embryogenic identity can be established. 94 Distinguishing between these possibilities has direct practical consequences (1) a quantitative 95 deficit could in principle be addressed simply by increasing DR dosage, whereas (2) a distinct, 96 competing transcriptional state would instead require regulators capable of actively suppressing 97 defense/stress programs while simultaneously activating fate-specification and pluripotency 98 programs. In this study, we test the latter hypothesis by comparing the transcriptional trajectories 99 of forage genotype (Tx2910) and grain genotype (Tx2783) under three DR treatments 100 ( _ZmBBM_, _ZmWUS2_, and GRF-GIF) and two explant types. By leveraging the ability of known 101 DR genes using multi-dimensional transcriptional comparison, we prioritized identifying novel, 102 sorghum-native developmental regulators associated with key developmental regulatory 103 networks. 104 To the best of our knowledge, these genotypes have not previously been transformed using either 105 conventional or DR-mediated approaches. Importantly, the molecular pathways that induce SE in 106 monocots remain largely uncharacterized, and genotype-specific responses to morphogenic 107 regulators suggest that the underlying regulatory networks differ among sorghum genotypes. 108 Through this experimental design, we identified novel transcriptional modules, key 109 regulators, and potential bottlenecks in sorghum transformation, enabling the rational design of 110 improved, more universal transformation systems. Ultimately, this study functionally validated 111 novel genes, provided a framework to enhance transformation efficiency in sorghum, and offered 112 insights that may apply to other monocot crops. 113 **Materials and methods** 114 _**Agrobacterium**_ _**tumefaciens**_ **strain and plasmid vectors:** 115 For all experiments, _Agrobacterium tumefaciens_ Thymidine-auxotrophic strain LBA4404 Thy 116 containing the helper plasmid ternary vector transformation system pPHP71539 (Anand _et al._, 117 2018; Zhang _et al._, 2019) was used. For immature-embryo transformation, two vectors were 118 used: (1) the T-DNA plasmid vector pPHP96021 carrying _ZmpltP:ZmWUS_ and _CmYLCV:Ruby_ 119 as a visible marker. Here, _CmYLCV:Ruby:HspT_ was amplified using attB4r and attB3 flanking 4 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 120 sites and subsequently cloned into pDonr221 with attP3 and attP4 sites via BP cloning (Roure _et_ 121 _al._, 2007), then cloned into pPHP96021 and pPHP69219 (containing attL4 and attL3 sites) via 122 LR cloning. (2) The GRF:GIF vector containing _ZmUbi:TaGRF4:GIF1:NosT_ was amplified 123 from JD633 (Debernardi _et al._, 2020) and cloned into module C. _ZmUbi:Cyan:PinII_ 124 (fluorescence marker) was amplified from pPHP81858 and cloned into module B; both modules 125 were combined via one-step Golden Gate assembly into the pTrans220d transformation 126 backbone containing the _neomycin phosphotransferase II_ ( _NptII_ ) selectable marker from the 127 Voytas toolkit. The empty control vector consisted of _PvUbi2:GFP_ (pModC_3013) and _NptII_ 128 (pTrans220d) ( Č ermák _et al._, 2017). For leaf transformation, pPHP81858 (containing _Nos:WUS2_ 129 and _Ubi:BBM_ along with _ZmUbi:Cyan_ as visible marker) and pPHP69219 (with _Nos:WUS2_ and 130 _Ubi:BBM_ along with _CmYLCV:Ruby_ as visible marker) were used. All vectors included the _NptII_ 131 selection marker except pPHP69219 (no selection used for leaf transformation). pPHP96021 132 employs a heat-shock-inducible Cre-loxP system for excision of _ZmWUS_ and the _NptII_ marker. 133 Similarly, pPHP81858 and pPHP69219 use a desiccation-inducible Cre-loxP system to excise 134 _Nos:WUS2_ and _Ubi:BBM_ . In contrast, the GRF-GIF construct was retained without an excision 135 system, as previous studies report that its expression did not cause pleiotropic effects (Butler _et_ 136 _al._, 2025; Li _et al._, 2024). Primer sequences used for construct design are given in File S1; vector 137 maps are provided in **Figure S1.** 138 139 **Plant material, explant preparation, and** _**Agrobacterium**_ **Infection:** 140 The study used sorghum inbred line Tx2910 (PI 598070), a Sudangrass-type genotype (Rooney 141 _et al._, 1998), and Tx2783 (PI 656001), a sorghum-aphid-resistant germplasm (Peterson _et al._, 142 1984). We adapted the immature-embryo (IE) transformation protocol from Che _et al._, 2022. 143 Briefly, we collected immature seeds 13–15 days after pollination from greenhouse-grown plants 144 and stored them overnight at 4°C. We selected uniform-sized seeds, sterilized them in 30% 145 bleach with 0.1% Tween-20, and rinsed them 5–6 times with sterile water. IEs approximately 2 146 2.5 mm were excised using sterile tweezers and a scalpel; approximately 30–40 IEs were 147 transferred to a 2 mL tube containing infection media and washed 2–3 times to remove milky 148 endosperm. IEs were heat-treated (45°C, 3 min), then incubated in _Agrobacterium_ suspension 149 supplemented with 200 µM acetosyringone, 50 mg thymidine, and 10% Pluronic-F68 surfactant 150 (OD550 = 0.4) for 10 min in the dark. Treated embryos were plated onto co-cultivation medium 5 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 151 (scutellum side up), sealed with micropore tape, and incubated in the dark at 24°C overnight, 152 followed by 3 days at 28°C (Che _et al._, 2022). For leaf transformation, mature seeds were 153 surface-sterilized with 75% ethanol followed by 30% bleach with Tween-20, then transferred to 154 germination medium supplemented with 2 mg/mL ancymidol. _In vitro_ germination was carried 155 out at 28°C with a 16/8 h light/dark photoperiod. Before explant harvesting, seedlings were 156 placed at 45°C and 70% relative humidity for 3 h. Leaf-base segments (~3 cm) were harvested 157 and fragmented using a Cuisinart Mini-Prep Food Processor. For _Agrobacterium_ infection, 158 OD550 = 0.6 was prepared in infection medium supplemented with 200 µM acetosyringone, 50 159 mg thymidine, and 10% Pluronic-F68 surfactant (Thermo Fisher Scientific, Cat. No. 24040032). 160 Leaf explants were infected for 20 min at room temperature, blotted dry, and transferred to solid 161 co-cultivation medium; plates were incubated at 28°C in the dark for two days (Wang _et al._, 162 2023b).We based the media composition on published protocols for IE transformation (Che _et_ 163 _al._, 2022) and leaf transformation (Wang _et al._, 2023b) with minor modifications. G418 (250 164 mg/L) was used to increase selection stringency and minimize escapes. 165 166 **Transformation and analysis of transgenic events:** 167 To avoid ambiguity from ratios that can exceed 100% when multiple shoots regenerate from a 168 single explant, we used the following efficiency matrix (1) Transformation efficiency (%) = 169 (Number of IEs producing embryogenic structures with visible marker gene expression 7 DPI/ 170 Total number of starting explants) × 100; this value is bounded 0–100% and reflects the 171 proportion of explants that responded. (2) Regeneration index = total number of plantlets 172 recovered ÷ total number of starting explants, reported as a unitless ratio (plantlets per explant) 173 rather than a percentage; because a single embryogenic callus can give rise to more than one 174 shoot, this index is expected to routinely exceed 1 and should not be expressed as a bounded 175 percentage. (3) Transformation frequency or T ransgenic purity (%) = (Number of Ruby � or 176 AmCyan � plantlets with visible expression as well as PCR confirmation for transgene/Total 177 number of regenerated explants) × 100 ( **Table S2** ). We performed PCR-based genotyping using 178 _Ruby_ primers for the pPHP96021 and pPHP69219 constructs, _AmCyan_ primers for the 179 pPHP81858 construct, and _GIF_ -specific primers for GRF-GIF construct. The list of primers is 180 provided in **File S1** . 6 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 181 **Sample Collection for bulk and single nucleus RNA-seq:** 182 For transcriptome analysis, immature embryo samples were collected after _Agrobacterium-_ 183 mediated transformation with control or EV (No DR), WUS (pPHP96021+Ruby), and GRF-GIF 184 at 7 DAT (Days After Transformation) and 15 DAT for both genotypes. Leaf samples from the 185 control were collected at 7 DAT as later stages exhibited necrosis in the absence of DRs, while 186 WUS/BBM (pPHP81858) samples were collected at 7 DAT and 20 DAT. These time points 187 correspond to key stages of SE formation and the initiation of regeneration ( **Figure S2** ). A 188 summary of sample designations is given in **Table S1** . Each sample name corresponds to a 189 specific genotype, explant type, developmental regulator gene used, and timepoint **.** Samples 190 were flash-frozen upon collection and ground in liquid nitrogen. mRNA was extracted using the 191 Invitrogen PureLink RNA mini kit. We prepared libraries and performed Illumina sequencing 192 following the standard NovaSeq 6000 PE150 protocols. Three biological replicates were 193 collected for each sample. Single-nucleus RNA-seq was performed using sorghum genotype 194 Tx2783 to compare control (Empty vector with _AmCYAN_ marker) vs DR treatment 195 (BBM+WUS; pPHP81858) samples. Samples were collected at 7 DAT to compare bulk RNA 196 seq data. Samples were flash-frozen upon collection and ground in liquid nitrogen. Nuclei were 197 isolated using an adapted protocol from D’Agostino et al. 2025. Briefly, frozen tissue was 198 ground in a mortar and pestle in liquid nitrogen. Samples were transferred to nuclei isolation 199 buffer, passed through 100 µm, 70 µm, and 30 µm cell filters (Miltenyi Biotec), and the filtrate 200 was centrifuged to collect the nuclei pellet. The nuclei washed once with wash buffer and spun 201 again. The final washed pellet was resuspended, stained with propidium iodide, and counted with 202 the Luna-FL cell counter (Logos Biosystems). The nuclei were then loaded onto the chip per the 203 manufacturer’s recommendations to target 10,000 recovered nuclei (10X Genomics, Pleasanton, 204 CA, USA). Library construction for Illumina sequencing was performed with the Chromium™ 205 Single Cell 3’ Library & Gel Bead Kit v3.1 protocol (10X Genomics). The single-indexed 206 sequencing and paired-end libraries were carried out on an Illumina™ NovaSeq 6000 platform 207 according to the 10X Genomics manual (D’Agostino _et al._, 2025). 208 209 **RNA-seq data analysis:** 210 Raw fastq.gz files were processed with fastp (0.23.2) using "–detect_adapter_for_pe," "– 211 trim_poly_x," and "–trim_poly_g" to remove adapters and low-quality reads (Chen _et al._, 2018). 7 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 212 We used the _Sorghum bicolor_ BTx623 v3.1 genome as reference (McCormick _et al._, 2017). To 213 confirm transgene integration, transgene sequences ( _ZmWUS, ZmBBM, Ruby, GRF-GIF,_ 214 _AmCyan_ ) were added to the reference annotation. Filtered reads were mapped to the modified 215 reference using kallisto (0.48.0) (Bray _et al._, 2016). We used kallisto abundance files for 216 differential expression analysis in R with DESeq2 (1.44.0), applying a threshold of |log2FC| ≥ 2 217 and Benjamini–Hochberg-adjusted _P_ < 0.05 to define DEGs for each pairwise contrast (Love _et_ 218 _al._, 2014). Variance-stabilized transformation (VST) values were used for PCA visualization, 219 and Z-scaled VST values for k-means clustering, visualization, and hormone dot plots. The 220 optimal cluster number for k-means clustering was determined using the elbow and silhouette 221 methods implemented in fviz_nbclust (factoextra 1.0.7), followed by visual inspection of 222 heatmap coherence (Kassambara and Mundt, 2016). Heatmaps were generated using "pheatmap" 223 within ComplexHeatmap (2.20.0) (Gu, 2022). GO enrichment used the topGO package in R with 224 annotations from Phytozome (Goodstein _et al._, 2012; Gu, 2022). Gene regulatory network 225 (GRN) analysis used the "GENIE3" algorithm within the bionero package (1.17.3) (Almeida 226 Silva and Venancio, 2022); each sample was subset to its top 5,000 expressed genes prior to 227 GRN construction. Gene co-expression network (GCN) analysis used the top ~10,000 variable 228 genes per explant type (embryo or leaf samples), analyzed via the "WGCNA" algorithm in 229 bionero with _net_type_ = "signed hybrid" and a module-merging threshold of 0.7. GRNs and 230 GCNs were visualized in Cytoscape (3.10.2) (Shannon _et al._, 2003). Hub TFs were defined 231 separately for the two network types used in this study, since they capture conceptually different 232 properties. For GRN analysis, hub TFs were defined as the top 10% of regulatory nodes within 233 each expression module by weighted out-degree (predicted regulatory connections to target 234 genes). For GCN analysis, hub genes were defined using standard WGCNA criteria: top 10% of 235 genes based on module membership (kME) as well as those with a kME > 0.8. 236 237 **snRNA-seq data analysis:** 238 Sequencing data was processed using kallisto bustools to obtain cell feature counts for analysis 239 (Sullivan _et al._, 2025). To confirm transgene integration, transgene sequences ( _ZmWUS,_ 240 _ZmBBM, AmCyan_ ) were added to the reference annotation. Count data was further analyzed 241 within the R package Seurat v5. (Hao _et al._, 2024). Nuclei in the 95th percentile for unique gene 242 count and 90 [th] -99 [th] percentile for UMI were selected for further pre-processing. Putative doublets 8 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 243 were removed with scDblFinder, and background RNA was removed with SoupX (Germain _et_ 244 _al._, 2022). The data were then normalized by SCTransform and processed via runPCA. Samples 245 were aligned with align_cds and reducedDim, both from the Monocle 3 R package, and UMAP 246 plots were created; clusters were assigned via cluster_cells (Cao _et al._, 2019). Annotation of cell 247 types/states was performed by utilizing orthologs from published snRNA-seq datasets from 248 soybean seed, SAM and rice regenerative callus (Lyu _et al._, 2026) ( **File S15** ), highlighting select 249 marker genes within clusters (Seurat FindAllMarkers with min.pct = 0.3, logfc.threshold = 0.3, 250 min.diff.pct = 0.2) and GO term enrichment analysis. Expression of marker gene sets were 251 analyzed using the AddModuleScore function from Seurat and observed using the dotplot and 252 featureplot functions. GO term enrichment was performed using the topGO package in R (Alexa 253 and Rahnenfuhrer, 2010) with gene annotations from the _Sorghum bicolor_ BTx623 v3.1 genome 254 obtained from Phytozome. Trajectory and pseudo-time analysis were performed using Monocle3, 255 and other plots were visualized using ggplot2 (Gómez-Rubio, 2017). 256 257 **Validation of the novel DR genes:** 258 The coding sequences of the candidate genes were either PCR-amplified from the cDNA or 259 commercially synthesized (Twist Bioscience), including the attB1 and attB2 sites. We codon 260 optimized several genes to facilitate synthesis while preserving the native amino acid sequence. 261 The coding sequences were first cloned into the pDONR221 Zeocin entry vector using the 262 Gateway® BP recombination reaction. Subsequently, the genes were transferred into the 263 Gateway-compatible binary vector pANIC6A by an LR recombination reaction placing each 264 gene under the control of the maize ubiquitin ( _ZmUbi_ ) promoter for constitutive expression 265 (Mann _et al._, 2012). We performed _Agrobacterium_ -mediated transformation using the Thy 266 LBA4404 strain, along with a helper plasmid, and regenerated plants in Tx2783 genotypes, as 267 described above. For preliminary experiments, we used the pANIC6A overexpression vector 268 with candidate genes cloned for transformation. Candidate genes that showed a positive 269 embryogenic response in preliminary experiments were subsequently transferred to the 270 pMOD_A vector ( _ZmUBI:GENE:OcsT_ ) and cloned with the _CmYLCV:AmCYAN_ fluorescent 271 marker (pMOD_C) using golden gate assembly ( Č ermák _et al._, 2017; Devkar _et al._, 2025) to 272 facilitate rapid identification of transformed tissues. _ZmWUS_ + _Ruby_ (pPHP96021) or _ZmWUS_ + 9 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 273 _AmCYAN_ (pPHP96564), and ZmWUS+ZmBBM (pPHP81858) were included as positive 274 controls. 275 **Morphological imaging and histological analysis of somatic embryos:** 276 The progression of SE formation was monitored at different time points (7, 12, 15, 20 DAT) by 277 morphological imaging using Olympus SZX10 stereo microscope. Images were captured under 278 comparable magnification and illumination conditions. For scanning electron microscopy 279 (SEM), samples were freeze-dried and mounted on double-sided carbon tape. We sputter-coated 280 samples with iridium using a Safematic CCU-010 compact coating unit and imaged them with a 281 Hitachi S3400 SEM (Hitachi, Ltd., Tokyo, Japan) (Dhiman _et al._, 2025). Images were acquired 282 at a working distance of 15 mm and 15 kV beam acceleration at different magnifications. For 283 histological analysis, developing calli with SE were fixed in FAA solution containing 90 mL of 284 50% ethanol, 5 mL of 37–40% formaldehyde, and 5 mL of glacial acetic acid. Samples were 285 fixed for 12–24 h at room temperature, followed by dehydration through a graded ethanol series 286 and clearing with xylene. Samples were then infiltrated with paraffin and embedded for 287 sectioning. Paraffin-embedded samples were sectioned at 5–8 µm thickness and mounted on 288 glass slides. Sections were deparaffinized with xylene, rehydrated through a descending ethanol 289 series to water, and stained with Toluidine Blue O. Following staining, sections were rinsed, 290 dehydrated through an ethanol series, cleared with xylene, and used for microscopic examination 291 (Steinmacher _et al._, 2007). 292 293 **RESULTS** 294 **DRs enabled efficient transformation and regeneration in sorghum genotypes:** 295 We selected the grain sorghum inbred line Tx2783 and the forage sorghum line Tx2910, neither 296 of which had previously been evaluated for tissue-culture response or regeneration capacity. IEs 297 of both genotypes were transformed with either an empty vector or a construct expressing DR 298 genes (WUS2 or GRF-GIF). In parallel, leaf segments were transformed with a WUS2 and BBM 299 combination. Transformation with the empty vector rarely to produce embryogenic calli and 300 failed to regenerate in either genotype across two independent experiments, confirming their 301 recalcitrant nature under conventional methods. In contrast, DR-mediated transformation 302 markedly enhanced regeneration efficiency in both genotypes ( **Figures 1–3** ). We first evaluated 10 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 303 the _WUS2_ construct pPHP96021 using _Ruby_ as a visual marker ( **Figure 1** ). In Tx2910, 70% of 304 explants produced embryogenic calli and we regenerated 108 plantlets in total, 86 of which 305 showed red shoots due to Ruby marker expression (regeneration index=2.7 plantlets/explant), 306 confirming successful transgene incorperation. In Tx2783, 37% of explants produced 307 embryogenic calli, generating 188 plantlets in total, of which 169 were Ruby-positive 308 (Regeneration index=0.94) ( **Figure 1 A–C, Table S2** ). No DR-mediated pleiotropic effects were 309 observed in the regenerating calli due to heat-shock-promoter-driven Cre-loxP-mediated excisio **n** 310 of the DR ( _WUS2_ ). Although _Ruby_ expression was highly observable in callus and regenerated 311 plantlets ( **Figure 1B, C** ), it’s presence gradually diminished in the vegetative tissues as the plant **s** 312 developed, but was consistently observed in the leaf base, reproductive tissues, and seeds 313 ( **Figure S3A–F** ). This expression pattern is consistent with earlier reports that the _CmYLCV_ 314 promoter is highly active in meristematic and reproductive tissues (Kshetry _et al._, 2025; 315 Stavolone _et al._, 2003). Overall, these results indicate that _WUS2_ can efficiently transform and 316 regenerate both genotypes. 317 318 **Figure 1.** Schematic representation and phenotypic outcomes of sorghum immature embryo 319 transformation using _**ZmWUS**_ _._ (A) T-DNA vector design for _Agrobacterium_ -mediated transformation. (B **)** 320 _ZmWUS_ -mediated transformation in forage sorghum (Tx2910). (C) _ZmWUS_ -mediated transformation in 321 grain sorghum (Tx2783), showing _Ruby_ expression in callus, regenerated plantlets, and panicles. Scale 322 bar = 3 mm. 1 **1** [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 323 To further assess an alternative DR, we tested the _TaGRF-GIF_ chimera driven by the _ZmUbi_ 324 promoter, which has been shown to drive efficient transformation in wheat (Debernardi _et al._, 325 2020), along with _AmCyan_ as a visual reporter ( **Figure 2A–C** ). In Tx2910, 38 of 75 IEs (51%) 326 formed embryogenic calli, yielding 110 plantlets (regeneration index = 1.47 plantlets/explant), 327 84 (76%) of which were AmCyan-positive. In Tx2783, 32 of 65 IEs (49%) formed embryogenic 328 calli, giving rise to 84 plantlets (regeneration index = 1.29 plantlets/explant), of which 68 were 329 AmCyan-positive ( **Table S2** ). Overall, the lesser embryogenic calli response in Tx2783 with 330 _WUS2_ and _GRF-GIF_ compared to Tx2910 was associated with pronounced phenolic exudation 331 ( **Figure S4** ). At the seed-setting stage, all T0 plants with constitutive _GRF-GIF_ chimera 332 expression under the _ZmUbi_ promoter showed a substantial reduction in seed set. The effect was 333 particularly severe in Tx2910, where seed number was reduced to as few as 0–5 seeds per 334 panicle ( **Figure S5A, B** ). 335 336 **Figure 2.** Schematic representation and phenotypic outcomes of sorghum immature embryo 337 transformation using _**TaGRF-GIF**_ chimera. (A) T-DNA vector design for _Agrobacterium_ -mediated 338 transformation. (B) GRF-GIF-mediated transformation in forage sorghum (Tx2910). (C) GRF-GIF339 mediated transformation in grain sorghum (Tx2783). Fluorescent somatic embryos expressing _AmCyan_ 340 were observed 5–7 DAT, followed by plant regeneration. Scale bar = 3 mm. 341 We further explored leaf segments as an explant source by transforming Tx2910 and Tx2783 342 _with Nos:WUS2_ and _Ubi:BBM_ co-expressing _AmCyan_ ( **Figure 3A** ). This promoter and DR 343 combination had been previously used in by Lowe et al. 2016 for maize transformation, which 344 produced embryogenic callus response (Lowe _et al._, 2016). The same DR combination, but with 345 enhanced promoters, was previously used for sorghum leaf-base transformation in Tx430, giving 1 **2** [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 346 rapid somatic-embryo formation (Wang _et al._, 2023b) and more recently for gene editing and 347 transformation across multiple sorghum genotypes (Azanu _et al._, 2025). Although SE from leaf 348 explants using _Nos:WUS2_ and _Ubi:BBM_ was comparatively slower than from IEs, we observed 349 embryogenesis within 7–10 days after transformation and recovered fully regenerated plants 350 within approximately 3–3.5 months ( **Figure 3** ). _AmCyan_ fluorescence and further molecular 351 analysis confirmed transgene integration in regenerated plantlets ( **Figure 3B, C; Figure S6,** 352 **S7C, D** ). 353 354 **Figure 3.** Schematic representation and phenotypic outcomes of sorghum leaf base transformation. **(A)** T **-** 355 DNA vector design used for transformation. **(B)** Somatic embryogenesis and plant regeneration in forage 356 (Tx2910) from leaf base explant ( **C** ) Somatic embryogenesis and plant regeneration in grain sorghum 357 (Tx2783) from leaf base explant, showing _AmCyan_ fluorescence in callus and regenerated plantlets. Scale 358 bar = 3 mm. 359 In a separate experiment using leaf-base explants with the same DR combination and _Ruby_ as a 360 visual marker under a selection-free strategy, we obtained reproducible transformation results 361 ( **Figure S8** ). Because constitutive _WUS and BBM_ expression causes pleiotropic effects, these 362 cassettes were flanked with a CRE-loxP excision system. As previously observed, whenever 363 _WUS/BBM_ failed to excise, pleiotropic effects appeared in Tx2783 ( **Figure S5C, D** ). We also 364 tested the _GRF-GIF_ chimera for leaf-based transformation, but it failed to induce regeneration in 365 either genotype. **Table S2** summarizes the transformation outcomes, including the number of 1 **3** [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 366 initial explants, transformation, and regeneration efficiency. **Figure S3G, H,** and **S6A-D** show 367 T1 segregating individuals from both genotypes with _Ruby_ and _AmCyan_ phenotypes, 368 respectively. Collectively, these results show that forage type Tx2910 has superior tissue culture 369 responsiveness with minimal phenolic interference, despite a higher escape frequency under 370 G418 selection. The elevated escape rate may reflect forage sorghum's intrinsic regenerative 371 vigor and tillering capacity and/or transient or non-cell-autonomous expression of DR genes in 372 certain cells, leading to shoot regeneration without transgene integration. 373 **DRs reprogram transcriptional networks and promote regeneration competence:** 374 Although transformation outcomes varied among genotypes, overall efficient regeneration by 375 DR overexpression suggested a role in reshaping developmental responses. To investigate the 376 underlying molecular basis, we compared transcriptomic profiles of DR-expressing and control 377 tissues during early embryogenesis and regeneration competence acquisition in the two sorghum 378 genotypes ( **Table S1** ). We selected embryogenic calli at 7 and 15 DAT from immature-embryo 379 explants across control, _GRF-GIF_, and _WUS_ overexpression treatments. For leaf explants, we 380 selected control samples at 7 DAT and _BBM+WUS2_ overexpression samples at 7 and 20 DAT. 381 The replicates showed a high correlation with an average R [2] >=0.9. Following the replicates, QC 382 samples with detectable transcripts from at least one of the transgenes from the vector used 383 ( _ZmBBM_, _ZmWUS2_, _GRF-GIF_, _Ruby_, _AmCyan_ ) were retained for downstream analysis. ( **File** 384 **S2** ). Differential expression analysis revealed genotype-specific responses as well as conserved 385 DR-responsive programs ( **Table S3; Figures S9, S10** ). Both _WUS_ and _GRF–GIF_ induced 386 progressively larger transcriptional changes from 7 to 15 DAT, with overlapping genotype 387 independent responses enriched for pathways associated with cellular development, metabolism, 388 photosynthesis, hormone regulation, sucrose metabolism, and cell-wall remodeling. 389 This conserved reprogramming is reflected in 387 and 604 overlapping differentially expressed 390 genes (DEGs) between genotypes at 7 and 15 DAT for _WUS_ ( **Figures S9** ), and from 20 shared 391 DEGs at 7 DAT to 339 at 15 DAT for _GRF-GIF_ ( **Figures S10** ). The shift from a comparatively 392 modest early response (7 DAT) to a larger response at 15 DAT indicates that these DRs act 393 primarily by reinforcing transcriptional programs required to establish regeneration competence. 394 Transcriptome analysis of leaf-derived embryogenic callus further demonstrated strong 395 genotype- and time-dependent responses following _WUS+BBM_ expression. At 7 DAT, FL_BW 14 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 396 (F= forage type, Tx2910 background) showed 8,151 upregulated genes relative to control. 397 Temporal comparison (7 vs. 20 DAT) showed 8,481 genes upregulated at 7 DAT versus 6,535 at 398 20 DAT. GL_BW (G=grain type, Tx2783 background) showed 4,895 upregulated genes at D7 399 relative to control, and 4,991 vs. 3,330 upregulated genes in the 7 vs. 20 DAT comparison 400 ( **Table S3** ). The strong WUS+BBM-induced upregulation in leaf-base callus reinforces the role 401 of combinatorial DR activity in driving embryogenic reprogramming, with genotype-dependent 402 magnitude differences. 403 404 **Figure 4.** K-means clustering-based identification of transcriptional modules responding to a 405 developmental regulator in forage sorghum (Tx2910). (A) K-means clustering heatmaps of differentially 406 expressed genes (log2FC >=2, an adjusted p-value (Benjamini–Hochberg corrected <0.05) across sample **s** 407 (control, GRF-GIF, and WUS) at 7 and 15 days after transformation (DAT) in Tx2910 (B) Representativ **e** 408 enriched Gene Ontology (GO) biological process terms (C) TF family frequency within each cluster; Tex **t** 409 size is proportional to the number of TF family members within each cluster, larger text indicates higher 410 counts. (D) Representative hub transcription factors (TFs) from each cluster/sample (with known roles in 411 development, hormone related, pathogen interaction, flavonoids, cell proliferation, etc.). 412 _**GRF-GIF**_ **and** _**WUS**_ **accelerate SE in immature embryos through distinct transcriptional** 413 **modulations:** 414 K-means clustering of DEGs coupled with GO (Gene Ontology) enrichment revealed 415 coordinated, regulator- and genotype-specific transcriptional shifts underlying SE ( **Figures 4, 5)** . 416 In forage sorghum Tx2910, six co-expression clusters captured this reprogramming ( **Figure 4A** ). 1 **5** [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 417 Cluster 1, associated with _WUS_ treatment was enriched in GO pathways promoting cell-fate 418 identity, and mobilizing metabolic reserves for dedifferentiation. Cluster 5 (all 7 DAT samples) 419 showed enrichment for cell-fate specification, dormancy, and catabolic regulation, consistent 420 with an early, transient suppression of differentiation. Clusters 2 and 3 (control_D15, GIF_D15) 421 were enriched for culture induced stress and growth-related GO terms; their downregulation 422 specifically in WUS-expressing tissue indicates that WUS actively suppresses these persistent 423 stress/cell-wall-remodeling programs. Cluster 6, induced in D15 samples, represents a 424 developmental-competence module (cell communication, signaling, meristem/reproductive 425 identity specification) that sustains fate-specification programs needed for regeneration ( **Figure** 426 **4B; Files S3, S4** ). Similarly, in grain sorghum, among four K-means clusters, Cluster 1 (abiotic 427 stress, secondary-metabolite/phenolic pathways) was high in control and GRF-GIF_D7 but 428 strongly downregulated in WUS_D7, WUS_D15, and GIF_D15, indicating that WUS helps 429 reprogram stress-dominant states as early as 7 DAT, while GRF-GIF's suppressive effect is 430 delayed until 15 DAT ( **Figure 5A, B** ). Cluster 2 (cell/organ morphogenesis) was induced in 431 WUS (D7, D15) and GRF-GIF_D15, and Cluster 4 contained genes expressed in all conditions 432 except control_D15 indicating that, without DRs, Tx2783 remains locked in a stress-dominant 433 state driven by high phenolic release. Overall, grain type Tx2783 showed a greater dependence 434 on DR-mediated activation of developmental pathways than forage type Tx2910. 435 1 **6** [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 436 **Figure 5.** K-means clustering-based identification of transcriptional modules responding to a 437 developmental regulator in grain sorghum (Tx2783). (A) K-means clustering heatmaps of differentially 438 expressed genes (log2FC >=2, an adjusted p-value (Benjamini–Hochberg corrected<0.05) across control, 439 GIF, and WUS treatments at 7 and 15 days after transformation (DAT) are shown in Tx2783 (B) 440 Representative enriched Gene Ontology (GO) biological process terms (C) TF family frequency within 441 each cluster; Text size is proportional to the number of TF family members within each cluster; larger text 442 indicates higher counts. (D) Representative hub transcription factors (TFs) from each cluster/sample (with 443 known roles in development, hormone related, pathogen interaction, flavonoids, cell proliferation, etc.). 444 Further **,** gene regulatory network analysis of cluster-specific hub TFs revealed both shared and 445 genotype-distinct regulatory architectures underlying developmental control, embryogenic 446 competence, and regeneration potential. In Tx2910, hub TFs from _SbWRKY_, _SbMYB_, _SbNAC_, 447 _SbEREBP/AP2_, and _SbbHLH_ families were present in multiple clusters, reflecting probable roles 448 in early morphogenetic reprogramming, hormone signaling, and embryogenesis ( **Figure 4 C, D** ). 449 For example, a homolog of Arabidopsis CUP-SHAPED COTYLEDON3 (CUC3), _SbNAC8_ a 450 hub in F_GIF_D15 is a NAC-domain transcription factor involved in developmental patterning, 451 organ boundary specification, and shoot meristem formation (Hibara _et al._, 2006; Takada _et al._, 452 2001; Takeda _et al._, 2011). Similarly, _SbbHLH111_, a hub in GRF-GIF samples and homolog of 453 _AtbHLH137_, functions in cytokinin signaling to promote cell proliferation and expansion (Park 454 _et al._, 2021) ( **Figure 4D, Files S5, S6** ). 455 Similarly, the grain sorghum Tx2783 regulatory clusters (Clusters 1–4) were dominated by 456 _SbMYB, SbWRKY, SbHB, SbDOF, SbbZIP, SbGATA_, and _SbNAC_ families, reflecting genotype 457 dependent transcriptional control of regeneration ( **Figure 5 C, D; Files S5, S7** ). Notably 458 _SbEREB165_, the closest sorghum homolog of _ZmBBM_ ( _Zm00001d042492_ ), was identified as a 459 hub gene in G_GIF_D15, G_GIF_D7, and G_WUS_D15. _SbNAC97_, a hub in G_GIF_D15 460 (Cluster 2), is an ortholog of _AtCUC3_ and contributes to shoot apical meristem formation 461 (Hibara _et al._, 2006). Similarly, _SbHB75_ is a hub in G_GIF_D15, and _SbHB71_, a hub in Cluster 462 4, both belong to the WUSCHEL-RELATED HOMEOBOX (WOX) family and are associated 463 with meristematic development ( **Files S5, S7** ) ~~. T~~ he recurrence of WRKY, MYB, NAC, and 464 EREB hubs across both genotypes underscores their broadly conserved role in embryogenesis 465 and regeneration, while genotype-specific enrichment of DOF, HB, bZIP, and GATA families in 466 Tx2783 points to additional layers of regulatory control governing its more recalcitrant 467 regeneration response. We next compared the ten most highly expressed TFs from four major TF 468 families (AP2-EREBP, ARF, C2C2-Dof, and GRF) comprising 40 TFs in total. Notably, 36 of 17 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 469 the 40 top-ranked TFs were shared between genotypes, indicating a largely conserved regulatory 470 core with varying expression patterns. For instance, _SbDOF15, SbEREB174, SbARF26_, and 471 _SbARF13_ were uniquely enriched among top TFs in Tx2910, whereas _SbEREB29, SbARF11,_ 472 _SbDOF8_, and _SbDOF12_ were specific to Tx2783 ( **Figure S11A, B** ). These findings enriched our 473 understanding of genotype-specific transcriptional mechanisms and identify candidate TFs for 474 targeted manipulation to improve regeneration efficiency. 475 **Transcriptional networks driving BBM/WUS2-induced SE:** 476 K-means clustering of leaf-derived DEGs (control vs. _BBM+WUS_ at 7 and 20 DAT) identified 477 five transcriptional modules in Tx2910 ( **Figure S12** ) and six in Tx2783 ( **Figure S13** ). Overall, 478 K-means clustering and GO enrichment in both genotypes indicate that BBM/WUS drives a 479 coordinated transition from stress- and differentiated-tissue-associated programs toward 480 developmental reprogramming; early responses (7 DAT) were dominated by stress signaling and 481 extensive hormone-mediated reprogramming, while later stages were dominated by cell 482 proliferation, morphogenesis, and metabolic activation, reflecting establishment of regenerative 483 competence ( **Figures S12B, S13B** ). Cluster 3 shows a highly conserved, strongly upregulated 484 profile across both genotypes, with core biological functions related to cell proliferation and 485 structural tissue differentiation, including shared hub genes _SbGRF1_ and _SbGRAS74_ ( **Figures** 486 **S12C, S13C, File S8, S9** ). We next performed weighted gene co-expression network analysis 487 (WGCNA), identifying 18 shared and genotype-specific co-expression modules (41–2,760 488 genes; **File S10; Figure S14** ). Module-trait correlations identified three categories of co 489 expression programs: (1) BBM/WUS-activated modules, (2) sample-specific early-response 490 modules, and (3) control-state modules antagonistic to DR-driven reprogramming ( **File S11** ). For 491 instance, Tx2783 (GL_BW_D20) showed a strong association with module 17 (MEblue4), 492 enriched for auxin transport, meristem maintenance, and cell-cycle regulators, consistent with 493 sustained reprogramming activity. Similarly, the Tx2910 (FL_BW_D20) MEblue module 494 showed GO enrichment for functional categories associated primarily with root development and 495 morphogenesis, post-embryonic root morphogenesis ( **Figure S14, File S12** ). Tx2783 exhibited a 496 stronger transcriptional transition, with activation of DNA replication, RNA processing, 497 ribosome biogenesis, and chromosome organization by 20 DAT ( **Figure S14F)**, indicating that 498 the recalcitrant genotype requires more extensive molecular reprogramming to achieve 499 regeneration as evident by K-mean clustering GO analysis and GO ontology of WGCNA module 18 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 500 for D20 samples. Furthermore, among AP2-EREB TFs, _SbEREB174_ showed the highest 501 expression at 20 DAT in both genotypes. Similarly, _SbARF5, SbARF3, SbARF13_, and _SbARF22_ 502 showed elevated expression at 20 DAT DR samples, indicating activation of auxin-responsive 503 transcriptional programs during acquisition of regenerative potential. _SbGRF1_ and _SbGRF3_ 504 showed marked increases in FL_BW_D20 samples relative to control tissue ( **File S13** ). Across 505 both genotypes, suppressing defense- and promoting differentiation-associated modules is a 506 conserved prerequisite for regeneration. 507 **Single-nucleus transcriptomic profiling reveals regeneration-associated cell states:** 508 To test whether bulk RNA-seq-identified developmental trajectories and putative regeneration 509 genes can be resolved at the single-cell resolution, we performed single-nucleus RNA-seq 510 (snRNA-seq) on relatively recalcitrant Tx2783 callus at 7 DAT, comparing the empty vector 511 (expressing _AmCYAN_ only) and BBM+WUS overexpression ( **Figure 6A** ). We obtained 3,761 512 and 3,298 high-quality nuclei from control and transgenic callus, respectively (median 839 UMIs 513 and 606 genes per nucleus), detecting 26,599 of the 34,120 annotated protein-coding genes in 514 sorghum ( **Table S4** ). Unsupervised clustering identified 21 distinct clusters with strongly skewed 515 composition between samples; cluster 16 for example comprised entirely of cells from transgenic 516 calli ( **Figure 6B, Figure S15A, B** ). Pseudo time trajectory inference placed these clusters along 517 a continuum from a quiescent, defense-associated state to two proliferative, transgene-associated 518 endpoints. The trajectory originated in clusters 4, 7, and 15 ("Wound-Responsive Embryo 519 Axis"), marked by elevated general defense and cell-wall-remodeling transcripts and G1/G0 cell 520 cycle markers ( **Figure 6 B, C, S15 D, E** ). This cell population reflects the counterpart of the 521 defense-dominant transcriptional state of our bulk k-means analysis identified as the default, DR 522 suppressed baseline in Tx2783 ( **Figure 5 K-mean Cluster 1** ). This same signature appeared as a 523 down-weighted expression module in bulk tissue, and here, it appears as a discrete cell 524 population that DR-treated tissue actively moves away from, rather than a uniform tissue-wide 525 change. Clusters 1 and 5 ("Early Embryo-Axis Callus") expressed markers consistent with a 526 transition out of a reserve-mobilizing state, including the rice scutellar-epithelium marker 527 _alphaAmy3/RAmy3D_ (Toyofuku _et al._, 1998) ( **Figure S17** ). 19 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 528 529 **Figure 6. snRNA-seq analysis (A)** Representative images of samples collected for nuclei isolation (EV530 Empty vector with AmCYAN), BW (BBM + WUS over expression with AmCYAN). **(B)** Integrated 531 UMAP projection of 7,059 high-quality nuclei annotated to 21 differentiated cell states. The color of the 532 cluster represents a cell state as represented in the legend with a dotted circle highlighting the putative 533 proliferative calli. Black open box represents cluster 16 specific to BBM + WUS. **(C)** UMAP plot 534 depicting projected pseudo time of the samples (color represents pseudo time and line indicates pseudo 535 time trajectory). **(D)** UMAP plots depicting localized expression of candidate genes _SbEREB167_ 536 ( _Sobic.006G245500_ ) and _SbEREB174_ **(** _Sobic.010G215100)_ . The top two plots represent _SbEREB167,_ an **d** 537 the bottom represents _SbEREB174_ . Red box represents expression in putative proliferative calli. Cluster 538 16 (black box) specific to BBM + WUS. EV samples are present on the left of each row and BW on the 539 right. 540 SE from immature-embryo explants in cereals is well studied and it originates from the scutellu **m** 541 rather than from the embryo axis proper (Eudes _et al._, 2006; Vasil _et al._, 1985). The presence of 542 scutellar markers in the earliest founder-like population was notable and suggests the possibility 543 that this cluster is at least partly scutellum-derived. From this founder-like state, the trajectory 544 bifurcated into two proliferative branches marked by S-phase and G2/M cell-cycle activation, 545 wherein cluster 2 and 3 ("Regeneration-Associated Proliferative Callus") were identified using 2 **0** [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 546 converging markers for "initial/proliferating cells" and "shoot meristem niche" from rice callus 547 snRNA-seq and "SAM" identity from soybean ( **Figure S15-S17, File S14, S15** ). These 548 proliferative endpoints correspond closely to the cell-cycle module our bulk analysis identified as 549 the hallmark of DR-driven reprogramming in Tx2783 ( **Cluster 2, Figure 5: "cell-cycle G2/M** 550 **phase transition"** ), in which _SbEREB167_ and _SbEREB165_ were identified as network hubs. 551 Clusters 6 and 14 lacked proliferative/cell-cycle markers and were dominated by non-transgenic 552 nuclei ( **Figure S15 A-C** ). Cluster 14 additionally showed mesophyll/photosynthetic markers, 553 consistent with untransformed cells and defaulting to a differentiated "greening" callus rather 554 than regeneration ( **Figure S16, S17** ). Remaining clusters were annotated as "Cell-Wall 555 Remodeling Callus" (cluster 13; _SbEXPB2_ orthologs, _SbLRX4_, _SbCWINV2_ ), "Epidermal/Outer 556 Layer-like" (cluster 18), and "Phloem/Vascular-like" (cluster 20) ( **File S14, Figure S16** ). 557 _SbEREB167_ and _SbEREB174_ (selected for functional validation below) were detectable broadly 558 across cell types and treatments, including in control tissue, but showed markedly elevated 559 expression specifically within the proliferative/regeneration-associated population under 560 BBM+WUS treatment ( **Figure 6D** ). This pattern of broad baseline expression suggested that 561 these genes' activity is gated less by where they are expressed than by how highly they are 562 expressed within a cell already primed for proliferation. 563 **Novel developmental regulators to promote sorghum embryogenesis:** 564 Integrative analysis of DEGs and gene regulatory networks converged on a set of high 565 confidence candidate genes potentially orchestrating SE and regeneration in sorghum. These 566 candidates span several key functional categories, including AP2/EREBP, GRF, DOF, 567 HB/WOX, ARF, ARR, epigenetic/cell-cycle regulators, and explant-specific factors, and were 568 prioritized by hub connectivity across multiple modules, conservation with functionally 569 characterized orthologs in maize, rice, and _Arabidopsis_, and consistent enrichment in DR 570 induced samples across both genotypes and explant types ( **Table S5** ). These candidates were 571 grouped into three priority tiers (Tier 1–3) based on hub connectivity across multiple network 572 modules and/or genotypes, and the strength of supporting ortholog evidence in Arabidopsis, rice, 573 and maize ( **see Table S5 legend for full tier criteria** ). Because of its recalcitrance, we selected 574 Tx2783 for functional validation. Four genes ( _SbHB41, SbEREB157, SbEREB167_, and 575 _SbEREB174_ ), previously uncharacterized for their role in SE and regeneration in sorghum, were 576 evaluated, with WUS-mediated (pPHP96021 or pPHP96564) embryogenesis as a positive 21 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 577 control ( **Figure 7** ). In the absence of a new DR (control), explants rapidly underwent necrosis 578 following transformation and failed to produce regenerated plants ( **Figure 7D** ). 579 580 **Figure 7. Functional validation of novel sorghum developmental regulators in the recalcitrant** 581 **genotype Tx2783.** **(A–F)** Immature embryos (IEs) transformed with _ZmWUS_ (A, positive control), 582 _SbEREB167_ (B), _SbEREB174_ (C), an empty vector (D, No DR), _SbHB41_ (E) or _SbEREB157_ (F), imaged 583 at 12 and 20 days after transformation (DAT) and, for A–C, at 40 DAT during plantlet regeneration. Scal **e** 584 bars, 1 mm (12 and 20 DAT). D–F produced no plantlets. **(G)** Somatic embryo (SE) frequency (% of IEs 585 forming SE, blue) and regeneration index (plantlets per 100 IEs, orange), pooled across independent 2 **2** [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 586 experiments (total events / total IEs; _ZmWUS, SbEREB167 and SbEREB174_, three experiments; No DR, 587 two); dots show individual experiments. _SbHB41_ and _SbEREB157_ produced no SE or regeneration. 588 Source data are in Table S6. **(H)** _AmCYAN_ fluorescence in _ZmWUS_, _SbEREB167_ and _SbEREB174_ callus 589 at 12 days post infection (DPI); arrows mark fluorescent foci of morphogenic structures (bright-field 590 images in Fig. S19). **(I)** Histology of _SbEREB174_ callus at early, intermediate and late embryogenic 591 stages. EC, embryogenic cells; VS, vasculature; PC, parenchymatous callus cells; LP, leaf primordia. 592 SEM images of embryogenic structures are in Fig. S18. 593 In three independent experiments, _SbEREB174_ and _SbEREB167_ induced somatic embryogenesis 594 (SE) in 52.3% (138/264) and 56.4% (220/390) of immature embryos, respectively, comparable to 595 the ZmWUS positive control (52.0%, 195/375), whereas empty-vector explants rarely formed SE 596 (3.3%, 6/180; Figure 7A–D, G; Table S6). The three DRs yielded 46.2, 51.5 and 61.9 plantlets 597 per 100 IEs, respectively. AmCYAN fluorescence marked the embryogenic foci ( **Figure 7H** ), 598 and the induced structures resembled those formed with _WUS_ and _WUS+BBM_, including early to 599 late embryogenic stages with vascular tissue and leaf primordia ( **Figure 7I; Figures S18, S19** ). 600 In contrast, _SbHB41_ (Homeobox family) and _SbEREB157_ (AP2/EREB family) did not induce 601 visible embryogenic structures, showing a response comparable to control ( **Figure 7E-G** ). These 602 results indicate that somatic-embryogenesis induction was gene-specific, and not all candidate 603 DRs can induce embryogenic reprogramming. Among the tested genes, we emphasized on 604 _SbEREB167_ and _SbEREB174_ that promoted rapid SE, with embryogenic structures observed as 605 early as 5 DAT and well-developed embryogenic calli by 12 DAT ( **Figure 7** ). Overall, two 606 independently selected, previously uncharacterized genes were sufficient to drive robust SE and 607 suggested that our network-based prioritization strategy was a constructive approach for 608 discovering native DR genes in recalcitrant crops. Regenerated T � plants grew normally after 609 transfer to soil and showed no obvious pleiotropic phenotypes at this stage, comparable to the 610 _ZmWUS_ control **(Figure S21)** . The remaining prioritized candidates in **Table S5** ( _GRF, DOF,_ 611 _HB/WOX, ARF/ARR_, and epigenetic/cell-cycle regulators) represent a resource for future 612 functional validation in sorghum and other cereals. Chromatin and cell-cycle regulators (e.g., 613 SbPHD20, SbSNF2.27, SbE2F5/7) and hormone-signaling hubs (SbARF2, SbARR5/9) represent 614 complementary, non-AP2/EREBP entry points that may act synergistically with the validated 615 regulators. Collectively, we identified and characterized novel sorghum genes that improve SE 616 and genetic transformation in a recalcitrant genotype. 617 **DISCUSSION** 23 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 618 This work establishes an effective transformation framework for two previously recalcitrant 619 sorghum genotypes using developmental regulators, significantly expanding the genetic 620 accessibility of sorghum improvement. Additionally, omics analysis supports the hypothesis that 621 sorghum reflects a distinct, defense/stress-dominated transcriptional state rather than a simple 622 quantitative shortfall in regenerative capacity. _WUS, BBM_ and _GRF–GIF_ actively suppress this 623 competing state while engaging fate-specification and proliferative programs, and the more 624 recalcitrant genotype (Tx2783) required broader and more sustained transcriptional 625 reprogramming. In summary: (1) early timepoints (7 DAT) are dominated by stress and cell-wall 626 remodeling, while later stages (15 DAT) shift toward developmental signaling and regeneration 627 competence; (2) WUS drives fate specification, whereas GRF-GIF preferentially modulates 628 growth, hormone metabolism, and structural remodeling; and (3) both DRs accelerate the 629 transition from a stress-dominated to a developmental transcriptional state. However, it remains 630 unclear whether these differences are driven primarily by genetic variation, the epigenetic 631 landscape, hormone responsiveness, cell wall properties, or interactions among these factors. 632 Future studies integrating comparative genomics, genome-wide association analyses using 633 diverse sorghum association panels, and chromatin accessibility profiling will be instrumental in 634 identifying the causal genetic and regulatory factors that determine transformability across 635 sorghum germplasm. Although _SbEREB167_ and _SbEREB174_ efficiently promote somatic 636 embryogenesis and regeneration, their direct downstream targets and molecular mode of action 637 remain determined. 638 639 **Figure 8:** **Proposed model of DR-mediated transcriptional reprogramming during SE in sorghum.** 640 **(A)** Effect of DR versus no-DR treatment on cellular reprogramming and SE induction. Green dots mark 641 DR/reporter-expressing cells; arrows indicate up- or down-regulation of the indicated gene categories. In 642 Empty vector defense/stress signaling (ROS, PP - phenylpropanoid pathway) remains active and explants 643 progress to necrosis rather than embryogenic callus. In immature embryo explants with WUS2 suppresse **s** 2 **4** [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 644 defense signaling while activating pluripotency ( _WOX11-LBD16_, _PLT_ ) and fate-specification genes 645 ( _ABI3_, _FUS3_, _HB62_ ) within 7 days of transformation. Immature seed explants with GIF-GRF first 646 suppress defense signaling (7d) and then activate fate genes ( _CYCB_, _CDKB-1_ ) and embryo-identity genes 647 ( _GRF1_, _GRF2_, _GRF7_ ). Leaf explants transformed with WUS+BBM produces synergistic suppression of 648 defense signaling together with activation of pluripotency and fate genes, yielding the robust SE 649 induction. **(B)** Downstream regulatory layers coordinated by DRs, based on GRN/GCN hub-gene analysis 650 showed genes related meristem organizers ( _WOX/HB, PLT, STM, CUC, BBM, ABI, YAB, GRAS, LBD_ ), 651 hormonal signaling ( _ARF, ARR_ ), and epigenetic/cell-cycle regulation ( _E2F, CYC, PHD, SNF_ ). 652 Convergence of these layers prioritized novel DR candidates, including SbEREB167 and SbEREB174 653 (Figure 7). 654 **Transcriptomics insights reveal mechanisms underlying DR-mediated regeneration:** 655 Ectopic expression of WUS and BBM is sufficient to induce embryogenic competence in 656 somatic tissue by activating fate-specification and stem-cell-identity genes (Lowe _et al._, 2016; 657 McFarland _et al._, 2023). The GRF-GIF module functions through a complementary route that 658 promotes cell proliferation and meristem regeneration (Debernardi _et al._, 2020; Lee _et al._, 2009; 659 Li _et al._, 2024). The genotype- and time-dependent suppression of defense/secondary-metabolite 660 pathways by both regulators align with the broader concept that developmental reprogramming. 661 The stress due to culture and _Agrobacterium_ -induced defense can block establishment of the 662 embryogenic identity unless actively overridden by strong developmental regulators (Duan _et al._, 663 2018; Gelvin, 2003; Schmidt _et al._, 2025) ( **Figure 8** ). The Tx2783 (grain) genotype reflects the 664 more extensive transcriptional reprogramming required to overcome its stronger defense and 665 secondary-metabolite states. By comparison, the less pronounced stress response in Tx2910 666 (forage) reflects its comparatively higher transformation and regeneration efficiency. Together, 667 our data indicate that somatic-cell reprogramming proceeds through an integrated sequence of 668 signal perception, activation of core cell-cycle and division modules, stress and maturation 669 adaptation, and engagement of embryogenic pathways. Cell-cycle regulators, _CDKs_, and _E2Fs_ 670 were elevated in DR-treated samples across both explant types ( **File S17** ), indicating conserved 671 basal cell-cycle regulation during SE. 672 snRNA-seq provides cellular-resolution support for these bulk-derived conclusions. The defense 673 dominant transcriptional state that bulk analysis identified as the default, DR-suppressed baseline 674 in Tx2783 corresponds to a discrete, non-dividing, wound-responsive cell population that DR 675 treated tissue moves away from, highlighting transitionary states masked by bulk RNA-seq 676 limitations. Likewise, the proliferative module in which we identified _SbEREB167_ and 677 _SbEREB174_ as bulk network hubs where both genes showed the strongest treatment-induced 25 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 678 upregulation. The two datasets therefore converge from complementary angles. Bulk RNA-seq 679 identified genes and pathways change with treatment across the tissue, while snRNA-seq 680 pinpoint that these changes are concentrated in, and plausibly define, the specific proliferative 681 cell population from which regenerated plants likely arise. 682 Somatic-embryogenesis regulators _ABI3_ ( _Sobic.007G000400, Sobic.003G398200_ ), _FUS3_ 683 ( _Sobic.003G275400_ ), and _NF-YA1_ ( _Sobic.001G127800_ ) were elevated in DR samples in both 684 genotypes ( **File S13** ). _ABI3_, a B3-family regulator of embryo maturation, promotes embryogenic 685 competence and is characteristically upregulated in dedifferentiated and embryogenic cells 686 (Freitas _et al._, 2019; Kang _et al._, 2021), while _FUS3_ cooperates with _LEC_ and _ABI3_ pathways to 687 maintain embryonic cell identity in _Arabidopsis_ (Horstman _et al._, 2017). Similarly, recent work 688 highlights the potential _ABI/FUS2_ family genes as positive regulators of regeneration, though 689 equivalent crosstalk in monocots remains poorly characterized (Ramasamy _et al._, 2026). Both 690 genotypes exhibited high expression of transcription factor (TF) families known to regulate plant 691 development and stress responses, including AP2/EREBP, ARF, C2C2-Dof, MYB, NAC, 692 WRKY, MADS-box, bHLH, Homeobox, and GRF, suggesting their involvement in coordinating 693 developmental and embryogenic processes during regeneration ( **File S13** ). Among the identified 694 hub genes, MYB TFs such as _SbMYB132_ ( _AtMYB105_ ) and _SbMYB18_ ( _AtMYB43_ ), which regulate 695 cell division in the Arabidopsis shoot meristem (Haga _et al._, 2011; Wang _et al._, 2023a), were 696 detected in both genotypes. AP2/EREBP TFs, including _SbEREB156_ ( _ZmEREB11_ ) 697 and _SbEREB49_ ( _OsERF3_ ), were identified as hub genes in F_WUS_D15 and in 698 G_WUS_D15/G_WUS_D7, respectively, further supporting their involvement in regeneration. 699 Homeobox/WUS-related TFs, such as _SbHB62_ and _SbHB44_, were predominantly active in 700 F_GIF_D15, G_WUS_D7, and G_WUS_D15, consistent with their established roles in 701 maintaining stem-cell identity and promoting embryogenic competence. Additionally, _SbNAC87_, 702 identified as a hub gene in Tx2783 (WUS 7 and 15 DAT), is homologous to _AtNAC030_ (VND7), 703 a key regulator of vascular differentiation and xylem formation (Yamaguchi _et al._, 2008) 704 indicating a potential role in vascular tissue development during regeneration. These highly 705 expressed candidate genes along their conserved roles in regeneration demonstrated in dicots 706 species, represent promising regulators for improving sorghum transformation ( **File S16** ). 707 **Pleiotropic effects associated with developmental regulators:** 26 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 708 Constitutive expression of DR can enhance transformation efficiency but may also cause 709 undesirable developmental phenotypes. In wheat, the same _GRF4–GIF1_ chimera used here 710 caused only a modest (~23.9%) reduction in grain number per spike, showing normal fertility 711 (Debernardi _et al._, 2020). In later study, Kronos showed the greatest sensitivity, while Paragon 712 and Cadenza were comparatively less affected in terms of fertility and spikelet number indicating 713 a genotype dependent response to GRF4-GIF over expression (Hayta _et al._, 2025). Similarly, in 714 maize, reduced fertility was associated with _TaGRF4–GIF1/ZmGRF1–GIF1_ under the _ZmUbi_ 715 promoter (Vandeputte _et al._, 2024), whereas the same chimera expressed under the _Arabidopsis_ 716 _ubiquitin10_ promoter did not affect fertility in sorghum cv. Tx430 (Li _et al._, 2024). Consistent 717 with these reports, we observed a significant, consistent reduction in seed number per plant 718 across all independent GRF–GIF events in both Tx2910 and Tx2783 under the strong _ZmUbi_ 719 promoter, despite an otherwise normal overall plant phenotype ( **Figure S5A, B** ). These 720 observations collectively indicate that _GRF–GIF_ pleiotropy is strongly influenced by genetic 721 background, promoter choice, and transgene dosage, and support the use of controlled strategies 722 such as tissue-specific promoters (e.g., _ZmPLTP_ ) or Cre-loxP excision to limit DR expression 723 once its developmental function has been achieved (Hayta _et al._, 2025). 724 **Potential of Novel DRs for overcoming genotype dependence:** 725 DR genes have substantially improved transformation and regeneration in several recalcitrant 726 crops. However, as discussed earlier, their effectiveness varies by genotype, explant type, and the 727 molecular state of the target tissue. Our transcriptomic analysis demonstrates that regeneration 728 competence requires coordinated activation of several regulatory networks at once, such as cell 729 cycle progression, embryogenic identity, hormone signaling, and suppression of stress responses. 730 Supporting this idea, combinatorial deployment of _PLT5_ with _GRF–GIF_ recently enabled 731 transformation of several previously recalcitrant plant species (Huo _et al._, 2026), showing that 732 rationally designed DR combinations can extend genotype-independence further than any single 733 regulator alone. Beyond _WUS_ and _GRF–GIF_, our integrative GRN/GCN analysis identified a 734 broader set of candidate regulators. Notably, the AP2/EREBP family was disproportionately 735 represented among high-priority, high-connectivity hubs across both genotypes and multiple 736 treatment modules ( **Table S5** ). This pattern supported our functional results, in which we 737 validated both genes ( _SbEREB167_ and _SbEREB174_ ) as efficient inducers of SE. Notably, plants 27 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 738 expressing _SbEREB167_ or _SbEREB174_ appeared phenotypically normal at the seedling stage 739 (Figure S21), although fertility and seed set in the T � and T � generations remain to be 740 evaluated. _SbEREB174_ (ortholog of _ZmEREB152/OsPLT2_ ) and _SbEREB167_ (ortholog of 741 _AtPLT5/OsPLT1_ ) each integrate auxin–cytokinin signaling and PLT-like pluripotency functions 742 in embryogenic tissue (Renema _et al._, 2025). This is consistent with their strong induction in 743 _WUS_ and _GRF–GIF_ modules, and with their phylogenetic placement alongside 744 ZmBBM/AtBBM/AtPLT4. _SbEREB167_ clustered closely with _ZmBBM_ and _SbEREB174_ 745 grouped within the broader AtPLT-related clade ( **Figure S20A, B** ). Because BBM and related 746 PLT proteins are well-established regulators of SE and regeneration, this phylogenetic placement 747 further supports a potential role for these genes in promoting regeneration competence. Despite 748 their shared placement, _SbEREB167_ and _SbEREB174_ shared only a 65% and 66.1% amino acid 749 identity with _ZmBBM_, respectively. This divergence indicates that _SbEREB167_ and _SbEREB174_ 750 are not simply orthologs of _ZmBBM_, but distinct, sorghum-regulators, that achieve a similar 751 developmental outcome through non-identical sequences, underscoring their value as novel 752 transformation tools ( **Figure S20 B, C** ). 753 **CONCLUSIONS** 754 This study provides a transcriptomic dissection of DR-mediated SE and regeneration in sorghum, 755 revealing that the molecular barrier to transformation in recalcitrant genotypes is not simply a 756 quantitative deficit in regenerative capacity, but a qualitatively distinct transcriptional state. The 757 state is one dominated by persistent stress and must be actively overridden before embryogenic 758 identity can be established. Several mechanistic observations emerged from this work with broad 759 implications for monocot transformation. First, DRs do not operate through a single conserved 760 transcriptional logic but deploy genotype-tailored strategies to reach a common developmental 761 endpoint. In Tx2910, WUS and GRF-GIF acted primarily by suppressing competing 762 stress/defense programs and engaging a relatively accessible embryogenic network, allowing 763 rapid progression through the reprogramming window. In the relatively recalcitrant Tx2783, the 764 same DRs must perform more extensive molecular work, where they actively induced hormone 765 signaling, engaging a broader TF repertoire including DOF, HB, bZIP, and GATA family 766 members, and sustaining a distinct reprogramming transcriptional state. This genotype-tailored 767 mechanism explains why a single DR dosage or timepoint rarely achieves equivalent 28 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 768 transformation efficiency across diverse genetic constitutions. Second, the convergence of both 769 genotypes on a conserved late-stage co-expression module (anchored by auxin transport, 770 CLV/WUS-feedback meristem signaling, and CDK-driven cell-cycle progression) defines, a 771 molecular checkpoint that appears to be requirement for sorghum regeneration competence 772 regardless of genotype or explant type. This module represents a tractable target, rather than 773 empirically optimizing entire transformation protocols for each new genotype. Targeted 774 activation of core genes (MEblue4) may offer a more direct, generalizable route to regeneration 775 competence. Third, and most significantly for the broader goal of crop transformation, the 776 integrative GRN analysis identified a set of sorghum-native transcription factors with strong 777 ortholog support and high hub connectivity across multiple independent network analyses. 778 Among several priority genes ( **Table S5** ), _SbEREB167_ and _SbEREB174_ (AP2 family) were 779 highly effective and efficient at inducing SE in Tx2783. By identifying and understanding the 780 key barriers, this study lays a mechanistic foundation for developing transformation technologies 781 for precision crop improvement within reach for the full diversity of sorghum and, by extension, 782 other unruly staple crops. 783 784 **Data Availability:** 785 The raw snRNA-seq sequencing data generated in this study have been deposited in the NCBI 786 Sequence Read Archive (SRA) under bio-project number PRJNA1556803. Analysis scripts used 787 for snRNA-seq processing, clustering, trajectory inference, and co-expression network 788 construction are available from the corresponding author upon reasonable request. The request 789 for LBA4404 Thy- and all pPHP vectors can be made to Nagesh 790 Sardesai; nagesh.sardesai@vylor.com. The vector GRF-GIF could be obtained from addgene 791 (https://www.addgene.org/160393/). The request for sorghum specific DRs can be made to 792 Gunvant Patil; gunvant.patil@ttu.edu . 793 794 **Author Contributions:** 795 PD: methodology, experimental design, formal analysis, data interpretation, and writing – 796 original draft; LD: conducted data analysis, writing, and editing of the manuscript, VD: assisted 797 with sorghum transformation, MJ and LHE: writing, review, and editing of the manuscript; YJ 798 and GP: conceptualization, formal analysis, funding acquisition, supervision, and writing, review 799 and editing. 800 **Conflict of interest:** 801 G.B.P., Y.J, and P.D are co-inventors on a patent application submitted in relation to the findings 802 reported in this study. 29 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 803 804 **Acknowledgements** 805 We are grateful to the State of Texas’ Governor’s University Research (GURI) for the research 806 funding and partial funding for bulk RNA seq from Sorghum Checkoff. P.D. is grateful to the 807 Texas Tech University Graduate School for Distinguished Graduate Student Assistantship and 808 Texas Tech University Davis College for Helen DeVitt Jones fellowship. The authors thank Dr. 809 Bill Gordon-Kamm and Dr. Nagesh Sardesai (Vylor) for providing PHP vectors and 810 _Agrobacterium_ LBA4404-Thy auxotrophic strain for sorghum transformation. 811 812 **Figure legends:** 813 **Figure 1.** Schematic representation and phenotypic outcomes of sorghum immature embryo 814 transformation using _**ZmWUS**_ _._ (A) T-DNA vector design for _Agrobacterium_ -mediated transformation. (B) 815 _ZmWUS_ -mediated transformation in forage sorghum (Tx2910). (C) _ZmWUS_ -mediated transformation in 816 grain sorghum (Tx2783), showing _Ruby_ expression in callus, regenerated plantlets, and panicles. Scale 817 bar = 3 mm. 818 819 **Figure 2.** Schematic representation and phenotypic outcomes of sorghum immature embryo 820 transformation using _**TaGRF-GIF**_ chimera. (A) T-DNA vector design for _Agrobacterium_ -mediated 821 transformation. (B) GRF-GIF-mediated transformation in forage sorghum (Tx2910). (C) GRF-GIF822 mediated transformation in grain sorghum (Tx2783). Fluorescent somatic embryos expressing _AmCyan_ 823 were observed 5–7 DAT, followed by plant regeneration. Scale bar = 3 mm. 824 825 **Figure 3.** Schematic representation and phenotypic outcomes of sorghum leaf base transformation. **(A)** T826 DNA vector design used for transformation. **(B)** Somatic embryogenesis and plant regeneration in forage 827 (Tx2910) from leaf base explant ( **C** ) Somatic embryogenesis and plant regeneration in grain sorghum 828 (Tx2783) from leaf base explant, showing _AmCyan_ fluorescence in callus and regenerated plantlets. Scale 829 bar = 3 mm. 830 831 **Figure 4.** K-means clustering-based identification of transcriptional modules responding to a 832 developmental regulator in forage sorghum (Tx2910). (A) K-means clustering heatmaps of differentially 833 expressed genes (log2FC >=2, an adjusted p-value (Benjamini–Hochberg corrected <0.05) across samples 834 (control, GRF-GIF, and WUS) at 7 and 15 days after transformation (DAT) in Tx2910 (B) Representative 835 enriched Gene Ontology (GO) biological process terms (C) TF family frequency within each cluster; Text 836 size is proportional to the number of TF family members within each cluster, larger text indicates higher 837 counts. (D) Representative hub transcription factors (TFs) from each cluster/sample (with known roles in 838 development, hormone related, pathogen interaction, flavonoids, cell proliferation, etc.). 839 840 **Figure 5.** K-means clustering-based identification of transcriptional modules responding to a 841 developmental regulator in grain sorghum (Tx2783). (A) K-means clustering heatmaps of differentially 842 expressed genes (log2FC >=2, an adjusted p-value (Benjamini–Hochberg corrected<0.05) across control, 843 GIF, and WUS treatments at 7 and 15 days after transformation (DAT) are shown in Tx2783 (B) 844 Representative enriched Gene Ontology (GO) biological process terms (C) TF family frequency within 845 each cluster; Text size is proportional to the number of TF family members within each cluster; larger text 846 indicates higher counts. (D) Representative hub transcription factors (TFs) from each cluster/sample (with 847 known roles in development, hormone related, pathogen interaction, flavonoids, cell proliferation, etc.). 848 30 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 849 **Figure 6. snRNA-seq analysis (A)** Representative images of samples collected for nuclei isolation (EV850 Empty vector with AmCYAN), BW (BBM + WUS over expression with AmCYAN). **(B)** Integrated 851 UMAP projection of 7,059 high-quality nuclei annotated to 21 differentiated cell states. The color of the 852 cluster represents a cell state as represented in the legend with a dotted circle highlighting the putative 853 proliferative calli. Black open box represents cluster 16 specific to BBM + WUS. **(C)** UMAP plot 854 depicting projected pseudo time of the samples (color represents pseudo time and line indicates pseudo 855 time trajectory). **(D)** UMAP plots depicting localized expression of candidate genes _SbEREB167_ 856 ( _Sobic.006G245500_ ) and _SbEREB174_ **(** _Sobic.010G215100)_ . The top two plots represent _SbEREB167,_ and 857 the bottom represents _SbEREB174_ . Red box represents expression in putative proliferative calli. Cluster 858 16 (black box) specific to BBM + WUS. EV samples are present on the left of each row and BW on the 859 right. 860 861 **Figure 7. Functional validation of novel sorghum developmental regulators in the recalcitrant** 862 **genotype Tx2783.** **(A–F)** Immature embryos (IEs) transformed with _ZmWUS_ (A, positive control), 863 _SbEREB167_ (B), _SbEREB174_ (C), an empty vector (D, No DR), _SbHB41_ (E) or _SbEREB157_ (F), imaged 864 at 12 and 20 days after transformation (DAT) and, for A–C, at 40 DAT during plantlet regeneration. Scale 865 bars, 1 mm (12 and 20 DAT). D–F produced no plantlets. **(G)** Somatic embryo (SE) frequency (% of IEs 866 forming SE, blue) and regeneration index (plantlets per 100 IEs, orange), pooled across independent 867 experiments (total events / total IEs; _ZmWUS, SbEREB167 and SbEREB174_, three experiments; No DR, 868 two); dots show individual experiments. _SbHB41_ and _SbEREB157_ produced no SE or regeneration. 869 Source data are in Table S6. **(H)** _AmCYAN_ fluorescence in _ZmWUS_, _SbEREB167_ and _SbEREB174_ callus 870 at 12 days post infection (DPI); arrows mark fluorescent foci of morphogenic structures (bright-field 871 images in Fig. S19). **(I)** Histology of _SbEREB174_ callus at early, intermediate and late embryogenic 872 stages. EC, embryogenic cells; VS, vasculature; PC, parenchymatous callus cells; LP, leaf primordia. 873 SEM images of embryogenic structures are in Fig. S18. 874 875 **Figure 8:** **Proposed model of DR-mediated transcriptional reprogramming during SE in sorghum.** 876 **(A)** Effect of DR versus no-DR treatment on cellular reprogramming and SE induction. Green dots mark 877 DR/reporter-expressing cells; arrows indicate up- or down-regulation of the indicated gene categories. In 878 Empty vector defense/stress signaling (ROS, PP - phenylpropanoid pathway) remains active and explants 879 progress to necrosis rather than embryogenic callus. In immature embryo explants with WUS2 suppresses 880 defense signaling while activating pluripotency ( _WOX11-LBD16_, _PLT_ ) and fate-specification genes 881 ( _ABI3_, _FUS3_, _HB62_ ) within 7 days of transformation. Immature seed explants with GIF-GRF first 882 suppress defense signaling (7d) and then activate fate genes ( _CYCB_, _CDKB-1_ ) and embryo-identity genes 883 ( _GRF1_, _GRF2_, _GRF7_ ). Leaf explants transformed with WUS+BBM produces synergistic suppression of 884 defense signaling together with activation of pluripotency and fate genes, yielding the robust SE 885 induction. **(B)** Downstream regulatory layers coordinated by DRs, based on GRN/GCN hub-gene analysis 886 showed genes related meristem organizers ( _WOX/HB, PLT, STM, CUC, BBM, ABI, YAB, GRAS, LBD_ ), 887 hormonal signaling ( _ARF, ARR_ ), and epigenetic/cell-cycle regulation ( _E2F, CYC, PHD, SNF_ ). 888 Convergence of these layers prioritized novel DR candidates, including SbEREB167 and SbEREB174 889 (Figure 7). 890 891 **Supplementary Data:** 892 893 **Figure S1. T-DNA regions of the constructs used in this study.** Schematic maps (not to scale) of the T894 DNA of each binary vector (A–K; listed below). Abbreviations: _ZmWUS_, _Zea mays_ WUSCHEL; 895 _ZmBBM_, _Z. mays_ BABY BOOM; _NptII_, neomycin phosphotransferase II selectable marker; _HptII_, 896 hygromycin phosphotransferase II selectable marker; CmYLCV, Cestrum yellow leaf curling virus 897 promoter; AmCYAN, cyan fluorescent protein from _Anemonia majano_ ; _rab17_, stress- and ABA 31 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 898 responsive maize promoter; _cre_, Cre recombinase; LTP2, lipid transfer protein 2 promoter; SbBMR, 899 guide RNA targeting _Brown midrib 6_ ; PINII, proteinase inhibitor II terminator; IN2, maize _In2-1_ ( _Saf1_ ) 900 terminator; OCS, octopine synthase terminator; E9, pea _RBCS E9_ terminator. RUBY, betalain reporter. 901 **Figure S2. Tissue sampling scheme for RNA-seq.** Representative explants at each sampling time for 902 IEs and leaf explants transformed with the empty vector (EV), _ZmWUS_ + RUBY, _GRF4:GIF1_ + 903 AmCyan, or _ZmBBM_ / _ZmWUS_ . IEs were sampled at 7 and 15 DAT and leaf explants at 7 and 20 DAT 904 (sample codes in Table S1). Images are from Tx2910; Tx2783 was sampled at the same stages. 905 **Figure S3. RUBY reporter expression in transgenic Tx2910 and Tx2783 plants and T** � **segregation.** 906 **A–C**, Tx2910 plants expressing _CmYLCV:RUBY_ (PHP96021): (A) leaf base/ligule and midrib of mature 907 vegetative plants; (B) spikelets; (C) seeds. Scale bars, 1 cm. **D–F**, as in A–C for Tx2783, showing a 908 similar tissue distribution of betalain pigmentation. **G, H**, T � seedlings segregating for RUBY in Tx2910 909 (G) and Tx2783 (H) events (PHP96021); NTC, non-transgenic control; Ruby-T1, T � progeny. 910 **Figure S4. Genotype-dependent phenolic release in cultured sorghum explants.** Tx2783 IEs (A) and 911 leaf explants (B) release phenolics into the co-cultivation medium, whereas Tx2910 IEs (C) and leaf 912 explants (D) do not. Tx2783 also releases phenolics into rooting medium (E), unlike Tx2910 (F). 913 **Figure S5. Pleiotropic effects of constitutive DR expression. A, B**, Plants constitutively expressing 914 _ZmUbi:GRF4:GIF1_ look phenotypically normal but set few seeds in Tx2783 (A) and none in Tx2910 915 (B). Scale bar, 1 cm. **C**, Constitutive _ZmBBM_ / _ZmWUS_ expression (PHP81858) causes thin, curled leaves 916 in Tx2783. **D**, Watery, callus-like structures with deformed roots and shoots 917 **Figure S6. T** � **segregation of the AmCyan marker in transgenic sorghum. A, B**, Tx2910 T � progeny 918 of PHP81858 events under white light (A) and cyan fluorescence (B). **C, D**, Tx2783 T � progeny of 919 PHP81858 events under white light (C) and cyan fluorescence (D). WT, wild-type (non-transformed) 920 control; AmCyan-T1, T � progeny of a transgenic event. 921 **Figure S7. PCR genotyping of randomly selected marker-positive transgenic plants.** PCR with 922 transgene-specific primers on plants selected at random from visible-marker-positive events: _RUBY_ in 923 Tx2783 (A) and Tx2910 (B) (PHP96021); _AmCyan_ in leaf-derived Tx2783 (C) and Tx2910 (D) plants 924 (PHP81858); and _GIF_ in Tx2783 (E) and Tx2910 (F) plants ( _GRF4:GIF1_, pPD71). WT, wild-type 925 negative control; P, plasmid positive control; remaining lanes, independent T � plants. 926 **Figure S8. Selection-free transformation mediated by ZmWUS and ZmBBM using the RUBY** 927 **reporter. A**, T-DNA design of the selection-free vector. **B, C**, RUBY-expressing tissues during 928 embryogenesis and regeneration in Tx2910 (B) and Tx2783 (C). The Tx2783 callus releases abundant 929 phenolics (C). Scale bar, 3 mm. 930 **Figure S9. Genes upregulated by ZmWUS** _**in both genotypes.**_ **A**, Venn diagram of genes upregulated 931 relative to the matching no-DR control in _ZmWUS_ -expressing Tx2910 and Tx2783 IEs at 7 and 15 DAT, 932 showing genotype-specific and shared DEGs. **B**, Genes upregulated in both genotypes at 7 DAT (387 933 genes, including _SbYAB7_ and _SbEREB174_ ). **C**, Gene Ontology (GO) biological-process enrichment of the 934 387 shared genes. **D**, Genes upregulated in both genotypes at 15 DAT. **E**, GO biological-process 935 enrichment of the genes shared at 15 DAT. 936 **Figure S10. Genes upregulated by GRF4–GIF1 in both genotypes.** Panels A–E as in Fig. S9 for 937 _GRF4–GIF1_ -expressing Tx2910 and Tx2783 IEs: (A) Venn diagram of genes upregulated relative to the 938 matching no-DR control at 7 and 15 DAT; (B) genes shared between genotypes at 7 DAT; (C) GO 939 biological-process enrichment of the genes in B; (D) genes shared at 15 DAT; (E) GO enrichment of the 940 genes in D. 32 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 941 **Figure S11. Expression patterns of transcription factor (TF) families during somatic** 942 **embryogenesis.** Mean Z-scaled expression of TF families across the RNA-seq samples of Tx2910 (A) 943 and Tx2783 (B), with heatmaps of the top 10 members of selected families. _SbEREB174_ (OsPLT2 944 ortholog) and _SbEREB167_ (OsPLT1 ortholog) are labelled. 945 **Figure S12. K-means transcriptional modules responding to ZmBBM + ZmWUS in forage sorghum** 946 **(Tx2910) leaf explants.** Transcriptional modules identified by k-means clustering in forage sorghum 947 (Tx2910) leaf explants. **A**, Heatmaps of DEGs (log2 fold change ≥ 2; Benjamini–Hochberg-adjusted P < 948 0.05) clustered into five clusters across the control at 7 DAT (FL_C_D7) and _ZmBBM_ + _ZmWUS_ at 7 and 949 20 DAT (FL_BW_D7, FL_BW_D20). **B**, Representative enriched GO biological-process terms for each 950 cluster. **C**, Representative hub TFs for each cluster. 951 **Figure S13. K-means transcriptional modules responding to ZmBBM + ZmWUS in grain sorghum** 952 **(Tx2783) leaf explants.** Transcriptional modules identified by k-means clustering in grain sorghum 953 (Tx2783) leaf explants. **A**, Heatmaps of DEGs (log2 fold change ≥ 2; Benjamini–Hochberg-adjusted P < 954 0.05) clustered into six clusters across the control at 7 DAT (GL_C_D7) and _ZmBBM_ + _ZmWUS_ at 7 and 955 20 DAT (GL_BW_D7, GL_BW_D20). **B**, Representative enriched GO biological-process terms for each 956 cluster. **C**, Representative hub TFs for each cluster. 957 **Figure S14. Weighted gene co-expression network analysis (WGCNA) of leaf explants. A**, Module– 958 trait correlation heatmap for the 18 co-expression modules and the six leaf samples (FL_C_D7, 959 FL_BW_D7, FL_BW_D20, GL_C_D7, GL_BW_D7, GL_BW_D20); the modules fall into three 960 categories of sample-associated expression programs. **B**, Number of genes in each module. **C–F**, GO 961 biological-process enrichment of the modules associated with FL_BW_D7 (C), FL_BW_D20 (D), 962 GL_BW_D7 (E) and GL_BW_D20 (F). 963 **Figure S15. Single-nucleus RNA-seq of EV and ZmBBM + ZmWUS (BW) tissue. A**, Fraction of 964 nuclei from EV and BW samples in each of the 21 clusters. **B**, UMAP of the integrated dataset split by 965 sample, coloured by cluster; the red circle marks a BW-specific cluster. **C**, Feature plots of transgene 966 expression ( _ZmBBM_ and _AmCyan_ ) split by sample; darker colour indicates higher expression. **D**, GO 967 biological-process terms enriched among cluster marker genes (FDR < 0.05). **E**, Feature plots of cell968 cycle phase scores (G1/G0, S, G2/M); darker colour indicates higher scores. 969 **Figure S16. Expression of soybean cell-type marker orthologs in the sorghum snRNA-seq dataset.** 970 UMAP feature plots showing expression of marker genes for cell types defined in published soybean seed 971 and shoot apical meristem (SAM) snRNA-seq datasets: early-maturity (EM), cotyledonary-stage (CS) and 972 middle-maturity (MM) seeds, and SAM. Source datasets:doi:10.1016/j.cell.2024.10.050 and bioRxiv, 973 doi:10.1101/2024.07.08.602332. Marker genes are listed in Supplementary Data. 974 **Figure S17. Expression of rice cell-type marker orthologs in the sorghum snRNA-seq dataset.** 975 UMAP feature plots showing expression of marker genes for rice cell types: mesophyll, phloem, 976 epidermis, callus founder cell/scutellar epithelium, initial cells, proliferating cells and SAM niche. Source 977 dataset: doi:10.1186/s12870-026-08596-6. Marker genes are listed in Supplementary Data. 978 **Figure S18. Morphology and histology of somatic embryogenesis induced by DRs.** 979 **A**, Tx2783 IEs expressing _ZmWUS_, _ZmWUS_ + _ZmBBM_, _SbEREB174_ or _SbEREB167_ at 7 DAT, with 980 scanning electron microscopy (SEM) images of the corresponding embryogenic structures. Scale bar, 1 981 mm. **B–D**, Histological sections of _SbEREB174_ -overexpressing calli at early (B), intermediate (C) and 982 late (D) embryogenic stages. EC, embryogenic cells; PC, parenchymatous callus cells; VS, vasculature; 983 LP, leaf primordia. 984 33 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 985 **Figure S19. Early detection of embryogenic foci with the AmCYAN marker.** Tx2783 IEs expressing 986 _ZmWUS_ + AmCYAN (PHP96564), _SbEREB174_ + AmCYAN or _SbEREB167_ + AmCYAN at 12 DAT. 987 AmCYAN fluorescence marks transgenic callus and enables early detection of embryo formation; arrows 988 indicate foci of morphogenic structure development. 989 990 **Figure S20. Phylogeny and sequence alignment of the 15 candidate regulators. A**, Maximum991 likelihood tree of the 15 sorghum candidates (blue) with genes associated with pluripotency in 992 Arabidopsis, rice and maize. **B**, Tree of AP2/EREBP-family genes and related developmental regulators. 993 **C**, Sequence alignment of the sorghum AP2/EREBP candidates and related regulators to the reference 994 sequence (ZmBBM2/ZmEREB187) is indicated. Protein sequences were aligned with MUSCLE and trees 995 were inferred by maximum likelihood in R. 996 **Figure S21: Morphology of representative T** � **transgenic plants after transfer to soil.** Tx2783 plants 997 regenerated from immature embryos expressing _ZmWUS_ (positive control; label “W”), _SbEREB167_ 998 (pPD137; label “137”) or _SbEREB174_ (pPD136; label “136”). Three independent events per construct are 999 shown. Plants were photographed after hardening in soil (2 weeks after transfer). All plants show normal 1000 leaf and shoot morphology without or deformed shoots. 1001 1002 **Supplementary Tables:** 1003 1004 **Table S1: Sample designations for bulk RNA-seq.** Each sample name encodes the explant and 1005 genotype, the developmental regulator (DR) treatment and the sampling time in the format 1006 explant/genotype_DR_D[DAT]. Prefixes: F, forage sorghum Tx2910 immature embryos (IEs); G, grain 1007 sorghum Tx2783 IEs; FL, Tx2910 leaf explants; GL, Tx2783 leaf explants. DR codes: C, control (empty 1008 vector, no DR); GIF, GRF4–GIF1 chimera; WUS, ZmWUS; BW, ZmBBM + ZmWUS. D7, D15 and D20 1009 denote 7, 15 and 20 days after transformation (DAT). 1010 1011 **Table S2: Transformation efficiency, regeneration index, transgenic purity and escape frequency** 1012 **by genotype, explant type and DR treatment:** Immature embryos (IEs) and leaf explants of Tx2910 and 1013 Tx2783 were transformed with the constructs indicated (EV, empty vector without a DR; WUS, _ZmWUS_ ; 1014 GRF-GIF, _GRF4–GIF1_ ; BBM_WUS, _ZmBBM_ + _ZmWUS_ ). Expt, independent experiment. “No. of IEs” is 1015 the number of IEs transformed; for leaf explants the denominator is the number of starting seedlings 1016 [confirm]. Values are given as number (percentage) where a percentage is shown. 1017 1018 **Table S3: Numbers of differentially expressed genes (DEGs) in pairwise comparisons between** 1019 **sampling times, DR treatments and genotypes.** Each comparison is shown as a pair of sample names 1020 (defined in Table S1) with the number of DEGs beneath each name. The number under a sample is the 1021 count of genes with significantly higher expression in that sample than in its comparison partner 1022 (equivalently, downregulated in the partner). Comparisons cover (i) developmental progression (D15 vs 1023 D7), (ii) DR treatment vs control at the same DAT, (iii) genotype (Tx2910 vs Tx2783) and (iv) leaf1024 explant time courses. 1025 1026 **Table S4. Quality metrics for the snRNA-seq dataset:** Minimum, median, mean and maximum number 1027 of unique molecular identifiers (UMIs) and of detected genes per nucleus after quality filtering, for the 1028 empty-vector control (EV; no DR) and the ZmBBM + ZmWUS (BW) samples, and for both samples 1029 combined. A total of 7,059 nuclei were retained (3,761 EV; 3,298 BW), and 26,599 distinct genes were 1030 detected across the combined dataset. 1031 **Table S5: Prioritized candidate regulators of somatic embryogenesis and regeneration:** Candidates 1032 were selected from co-expression and gene regulatory network analyses on the basis of hub connectivity, 1033 differential regulation across tissues and time points, and functional evidence from orthologs. Genes are 34 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1034 grouped by family or functional category. Gene ID, sorghum locus with gene name; Ortholog, closest 1035 characterized ortholog in Arabidopsis (At), rice (Os) or maize (Zm); Key modules, RNA-seq sample1036 specific modules (sample names as in Table S1) or leaf co-expression network modules (named by 1037 colour) in which the gene is a hub; Functional relevance, supporting evidence with references. Priority 1038 tiers. Tier 1, hub in ≥ 3 modules and/or in both genotypes, with strong ortholog evidence. Tier 2, 1039 significant hub in ≥ 2 modules, supported by ortholog function. Tier 3, explant- or genotype-specific, with 1040 weaker cross-system support. SbEREB174, SbEREB167, SbEREB157 and SbHB41 were tested 1041 functionally in this study (Fig. 7; Table S6); the remaining candidates have yet to be validated by 1042 transformation. 1043 1044 **Table S6: Somatic embryogenesis and plant regeneration from Tx2783 immature embryos** 1045 **expressing novel DRs.** Tx2783 IEs were transformed with ZmWUS (positive control), SbEREB167, 1046 SbEREB174 or no DR (negative control). Each row is one independent experiment (Expt). No. of IEs, 1047 explants transformed. Somatic embryo frequency (%) = (IEs forming somatic embryos ÷ IEs transformed) 1048 × 100, scored from AmCyan-positive embryogenic structures (marked “AmCyan positive”) or, where 1049 AmCyan was not used, from embryogenic calli identified by morphology (marked “Embryogenic calli”) 1050 at [DAT]. Total regenerated plants = number of plantlets recovered per experiment. The regeneration 1051 index plotted in Fig. 7G is the pooled number of plantlets per 100 IEs (total plantlets/total IEs x 100). “No 1052 regeneration”: no plantlets were recovered. SbHB41 and SbEREB157 produced no somatic embryos or 1053 plants. 1054 1055 **References** 1056 **Alexa A, Rahnenfuhrer J** . 2010. topGO: enrichment analysis for gene ontology. R package version **2**, 1057 2010. 1058 **Almeida-Silva F, Venancio TM** . 2022. BioNERO: an all-in-one R/Bioconductor package for 1059 comprehensive and easy biological network reconstruction. Funct Integr Genomics **22**, 131-136. 1060 **Anand A, Bass SH, Wu E, Wang N, McBride KE, Annaluru N, Miller M, Hua M, Jones TJ** . 2018. 1061 An improved ternary vector system for Agrobacterium-mediated rapid maize transformation. Plant 1062 molecular biology **97**, 187-200. 1063 **Aregawi K, Shen J, Pierroz G, Sharma MK, Dahlberg J, Owiti J, Lemaux PG** . 2022. 1064 Morphogene � assisted transformation of Sorghum bicolor allows more efficient genome editing. 1065 Plant Biotechnology Journal **20**, 748-760. 1066 **Azanu MK, Lee K, Wang K** . 2025. Optimisation of Agrobacterium � Mediated Seedling Leaf 1067 Transformation Across Multiple Sorghum Genotypes. Plant Biotechnology Journal. 1068 **Bray NL, Pimentel H, Melsted P, Pachter L** . 2016. Near-optimal probabilistic RNA-seq quantification. 1069 Nature biotechnology **34**, 525-527. 1070 **Butler NM, Carlson AT, Starker C, Voytas DF** . 2025. Viral � mediated delivery of morphogenic 1071 regulators enables leaf transformation in Sorghum bicolor (L.). Plant Biotechnology Journal **23**, 44911072 4499. 1073 **Cao J, Spielmann M, Qiu X, Huang X, Ibrahim DM, Hill AJ, Zhang F, Mundlos S, Christiansen L,** 1074 **Steemers FJ** . 2019. The single-cell transcriptional landscape of mammalian organogenesis. Nature 1075 **566**, 496-502. 1076 Č **ermák T, Curtin SJ, Gil-Humanes J,** Č **egan R, Kono TJ, Kone** č **ná E, Belanto JJ, Starker CG,** 1077 **Mathre JW, Greenstein RL** . 2017. A multipurpose toolkit to enable advanced genome engineering 1078 in plants. The Plant Cell **29**, 1196-1217. 1079 **Che P, Anand A, Wu E, Sander JD, Simon MK, Zhu W, Sigmund AL, Zastrow** � **Hayes G, Miller** 1080 **M, Liu D** . 2018. Developing a flexible, high � efficiency Agrobacterium � mediated sorghum 1081 transformation system with broad application. Plant Biotechnology Journal **16**, 1388-1395. 35 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1082 **Che P, Wu E, Simon MK, Anand A, Lowe K, Gao H, Sigmund AL, Yang M, Albertsen MC,** 1083 **Gordon-Kamm W** . 2022. Wuschel2 enables highly efficient CRISPR/Cas-targeted genome editing 1084 during rapid de novo shoot regeneration in sorghum. Communications Biology **5**, 344. 1085 **Chen S, Zhou Y, Chen Y, Gu J** . 2018. fastp: an ultra-fast all-in-one FASTQ preprocessor. 1086 Bioinformatics **34**, i884-i890. 1087 **D’Agostino L, Devkar V, Lenin Y-V, Jagadish SK, Deshmukh R, Herrera-Estrella L, Patil GB** . 1088 2025. Single-cell transcriptomic landscape of sorghum leaf reveals cell-specific mineral nutrient 1089 regulation. Plant Physiology Reports **30**, 959-964. 1090 **Debernardi JM, Tricoli DM, Ercoli MF, Hayta S, Ronald P, Palatnik JF, Dubcovsky J** . 2020. A 1091 GRF–GIF chimeric protein improves the regeneration efficiency of transgenic plants. Nature 1092 biotechnology **38**, 1274-1279. 1093 **Devkar V, Ghose K, D’Agostino L, Patil GB** . 2025. Exonuclease-fused CRISPR-cas system enhances 1094 targeted genome editing for functional genomics in soybean. BMC Plant Biology **25**, 1283. 1095 **Dhiman P, Soumen S, Nigam D, Bean SR, Wu X, Patil GB, Xin Z, Jiao Y** . 2025. Effects of a novel 1096 loss-of-function waxy endosperm allele on sorghum seed development and grain quality. G3: Genes, 1097 Genomes, Genetics **15**, jkaf177. 1098 **Duan K, Willig CJ, De Tar JR, Spollen WG, Zhang ZJ** . 2018. Transcriptomic analysis of Arabidopsis 1099 seedlings in response to an Agrobacterium-mediated transformation process. Molecular plant-microbe 1100 interactions **31**, 445-459. 1101 **Eudes FAG, Laroche AJ, Acharya SN** . 2006. Process for inducing direct somatic embryogenesis in 1102 immature scutella cells of pooideae, and rapidly regenerating fertile plants. Google Patents. 1103 **Fontanet** � **Manzaneque JB, Haeghebaert J, Aesaert S, Coussens G, Pauwels L, Caño** � **Delgado AI** . 1104 2024. Efficient sorghum and maize transformation using a ternary vector system combined with 1105 morphogenic regulators. The Plant Journal **120**, 2076-2088. 1106 **Freitas NC, Barreto HG, Torres LF, Freire LL, Rodrigues LAZ, Diniz LEC, Beijo LA, Paiva LV** . 1107 2019. In silico and in vivo analysis of ABI3 and VAL2 genes during somatic embryogenesis of 1108 Coffea arabica: competence acquisition and developmental marker genes. Plant Cell, Tissue and 1109 Organ Culture (PCTOC) **137**, 599-611. 1110 **Gelvin SB** . 2003. Agrobacterium-mediated plant transformation: the biology behind the “gene-jockeying” 1111 tool. Microbiology and molecular biology reviews **67**, 16-37. 1112 **Germain P-L, Lun A, Meixide CG, Macnair W, Robinson MD** . 2022. Doublet identification in single1113 cell sequencing data using scDblFinder. f1000research **10**, 979. 1114 **Gómez-Rubio V** . 2017. ggplot2-elegant graphics for data analysis. Journal of statistical software **77**, 1-3. 1115 **Goodstein DM, Shu S, Howson R, Neupane R, Hayes RD, Fazo J, Mitros T, Dirks W, Hellsten U,** 1116 **Putnam N, Rokhsar DS** . 2012. Phytozome: a comparative platform for green plant genomics. 1117 Nucleic Acids Res **40**, D1178-1186. 1118 **Gu Z** . 2022. Complex heatmap visualization. iMeta **1**, e43. 1119 **Haga N, Kobayashi K, Suzuki T, Maeo K, Kubo M, Ohtani M, Mitsuda N, Demura T, Nakamura** 1120 **K, Jürgens G** . 2011. Mutations in MYB3R1 and MYB3R4 cause pleiotropic developmental defects 1121 and preferential down-regulation of multiple G2/M-specific genes in Arabidopsis. Plant Physiology 1122 **157**, 706-717. 1123 **Hao Y, Stuart T, Kowalski MH, Choudhary S, Hoffman P, Hartman A, Srivastava A, Molla G,** 1124 **Madad S, Fernandez-Granda C** . 2024. Dictionary learning for integrative, multimodal and scalable 1125 single-cell analysis. Nature biotechnology **42**, 293-304. 1126 **Hayta S, Smedley MA, Bayraktar M, Forner M, Backhaus A, Lister C, Clarke M, Uauy C,** 1127 **Griffiths S** . 2025. Controlling GRF4-GIF1 expression for efficient, genotype-independent 1128 transformation across wheat cultivars. bioRxiv, 2025.2008. 2029.673042. 1129 **Horstman A, Li M, Heidmann I, Weemen M, Chen B, Muino JM, Angenent GC, Boutilier K** . 2017. 1130 The BABY BOOM transcription factor activates the LEC1-ABI3-FUS3-LEC2 network to induce 1131 somatic embryogenesis. Plant Physiology **175**, 848-857. 36 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1132 **Huo H, Jiang T, Lian Z, Rodriguez K, Liu F, Karn A, Handoyo W, Kyum M, Tanwir SE, Van** 1133 **Deynze A** . 2026. Synergistic developmental regulators enable efficient plant regeneration and 1134 transformation across species. 1135 **Kang H-I, Lee C-B, Kwon S-H, Park J-M, Kang K-S, Shim D** . 2021. Comparative transcriptome 1136 analysis during developmental stages of direct somatic embryogenesis in Tilia amurensis Rupr. 1137 Scientific reports **11**, 6359. 1138 **Kassambara A, Mundt F** . 2016. Factoextra: extract and visualize the results of multivariate data 1139 analyses. CRAN: Contributed Packages. 1140 **Kshetry AO, Ghose K, Alok A, Devkar V, Raman V, Stupar RM, Herrera-Estrella L, Zhang F,** 1141 **Patil GB** . 2025. A synthetic transcription cascade enables direct in planta shoot regeneration for 1142 transgenesis and gene editing in multiple plants. Molecular plant **18**, 2066-2081. 1143 **Lee BH, Ko J-H, Lee S, Lee Y, Pak J-H, Kim JH** . 2009. The Arabidopsis GRF-INTERACTING 1144 FACTOR gene family performs an overlapping function in determining organ size as well as multiple 1145 developmental properties. Plant Physiology **151**, 655-668. 1146 **Li J, Pan W, Zhang S, Ma G, Li A, Zhang H, Liu L** . 2024. A rapid and highly efficient sorghum 1147 transformation strategy using GRF4 � GIF1/ternary vector system. The Plant Journal **117**, 1604-1613. 1148 **Lin C-Y, Tian Y, Nelson-Vasilchik K, Hague J, Kakumanu R, Lee MY, Pidatala VR, Trinh J, De** 1149 **Ben CM, Dalton J** . 2022. Engineering sorghum for higher 4-hydroxybenzoic acid content. Metabolic 1150 Engineering Communications **15**, e00207. 1151 **Love MI, Huber W, Anders S** . 2014. Moderated estimation of fold change and dispersion for RNA-seq 1152 data with DESeq2. Genome biology **15**, 550. 1153 **Lowe K, Wu E, Wang N, Hoerster G, Hastings C, Cho M-J, Scelonge C, Lenderts B, Chamberlin** 1154 **M, Cushatt J** . 2016. Morphogenic regulators Baby boom and Wuschel improve monocot 1155 transformation. The Plant Cell **28**, 1998-2015. 1156 **Lyu L, Lou W, Yan W, Yu H, Gong Z, Wu Y** . 2026. A single-cell transcriptome atlas reveals the 1157 transcriptionally earliest state in rice callus. BMC Plant Biology. 1158 **Mann DG, LaFayette PR, Abercrombie LL, King ZR, Mazarei M, Halter MC, Poovaiah CR,** 1159 **Baxter H, Shen H, Dixon RA** . 2012. Gateway � compatible vectors for high � throughput gene 1160 functional analysis in switchgrass (Panicum virgatum L.) and other monocot species. Plant 1161 Biotechnology Journal **10**, 226-236. 1162 **McCormick RF, Truong SK, Sreedasyam A, Jenkins J, Shu S, Sims D, Kennedy M, Amirebrahimi** 1163 **M, Weers BD, McKinley B** . 2017. The Sorghum bicolor reference genome: improved assembly, 1164 gene annotations, a transcriptome atlas, and signatures of genome organization. The Plant Journal **93** . 1165 **McFarland FL, Collier R, Walter N, Martinell B, Kaeppler SM, Kaeppler HF** . 2023. A key to 1166 totipotency: Wuschel � like homeobox 2a unlocks embryogenic culture response in maize (Zea mays 1167 L.). Plant Biotechnology Journal **21**, 1860-1872. 1168 **Mookkan M, Nelson-Vasilchik K, Hague J, Zhang ZJ, Kausch AP** . 2017. Selectable marker 1169 independent transformation of recalcitrant maize inbred B73 and sorghum P898012 mediated by 1170 morphogenic regulators BABY BOOM and WUSCHEL2. Plant Cell Reports **36**, 1477-1491. 1171 **Nelson** � **Vasilchik K, Hague J, Mookkan M, Zhang ZJ, Kausch A** . 2018. Transformation of 1172 recalcitrant sorghum varieties facilitated by Baby Boom and Wuschel2. Current Protocols in Plant 1173 Biology **3**, e20076. 1174 **Nigam D, Devkar V, Dhiman P, Shakoor S, Liu D, Patil GB, Jiao Y** . 2025. Emerging frontiers in 1175 sorghum genetic engineering. The Plant Journal **121**, e17244. 1176 **Park K, Quach T, Clark TJ, Kim H, Zhang T, Wang M, Guo M, Sato S, Nazarenus TJ, Blume R** . 1177 2025. Development of vegetative oil sorghum: from lab � to � field. Plant Biotechnology Journal **23**, 1178 660-673. 1179 **Peterson GC, Johnson J, Teetes G, Rosenow D** . 1984. Registration of Tx2783 greenbug resistant 1180 sorghum germplasm line. 37 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1181 **Ramasamy M, Li G, Guo L, Fang H, Eid A, Mandlik R, Cheng Y, Niedz RP, Culver JN, Coleman** 1182 **GD** . 2026. Boosting genome editing in perennial plants by CRISPR-Combo mediated morphogenic 1183 gene activation. Nature Communications. 1184 **Renema J, Lukicheva S, Verwaerde I, Aesaert S, Coussens G, De Block J, Grones C, Eekhout T, De** 1185 **Rybel B, Brew-Appiah RA** . 2025. Gene regulatory network analysis of somatic embryogenesis 1186 identifies morphogenic genes that increase maize transformation frequency. bioRxiv, 2025.2006. 1187 2022.659756. 1188 **Rooney W, Hussey M, Sanderson M** . 1998. Registration of Tx2909 and Tx2910 sorghum germplasm 1189 (sudangrass type). 1190 **Schmidt JJ, Brandenburg VB, Elders H, Shahzad S, Schäkermann S, Fiedler R, Knoke LR,** 1191 **Pfänder Y, Dietze P, Bille H** . 2025. Two redox-responsive LysR-type transcription factors control 1192 the oxidative stress response of Agrobacterium tumefaciens. Nucleic acids research **53**, gkaf267. 1193 **Scully ED, Gries T, Palmer NA, Sarath G, Funnell** � **Harris DL, Baird L, Twigg P, Seravalli J,** 1194 **Clemente TE, Sattler SE** . 2018. Overexpression of SbMyb60 in Sorghum bicolor impacts both 1195 primary and secondary metabolism. New Phytologist **217**, 82-104. 1196 **Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, Amin N, Schwikowski B, Ideker** 1197 **T** . 2003. Cytoscape: a software environment for integrated models of biomolecular interaction 1198 networks. Genome Res **13**, 2498-2504. 1199 **Stavolone L, Kononova M, Pauli S, Ragozzino A, De Haan P, Milligan S, Lawton K, Hohn T** . 2003. 1200 Cestrum yellow leaf curling virus (CmYLCV) promoter: a new strong constitutive promoter for 1201 heterologous gene expression in a wide variety of crops. Plant molecular biology **53**, 703-713. 1202 **Steinmacher DA, Krohn N, Dantas A, Stefenon VM, Clement CR, Guerra MP** . 2007. Somatic 1203 embryogenesis in peach palm using the thin cell layer technique: induction, morpho-histological 1204 aspects and AFLP analysis of somaclonal variation. Annals of Botany **100**, 699-709. 1205 **Sullivan DK, Min KH, Hjörleifsson KE, Luebbert L, Holley G, Moses L, Gustafsson J, Bray NL,** 1206 **Pimentel H, Booeshaghi AS** . 2025. kallisto, bustools and kb-python for quantifying bulk, single-cell 1207 and single-nucleus RNA-seq. Nature Protocols **20**, 587-607. 1208 **Toyofuku K, Umemura T-a, Yamaguchi J** . 1998. Promoter elements required for sugar � repression of 1209 the Ramy3D gene for α� amylase in rice. FEBS letters **428**, 275-280. 1210 **Vandeputte W, Coussens G, Aesaert S, Haeghebaert J, Impens L, Karimi M, Debernardi JM,** 1211 **Pauwels L** . 2024. Use of GRF � GIF chimeras and a ternary vector system to improve maize (Zea 1212 mays L.) transformation frequency. The Plant Journal **119**, 2116-2132. 1213 **Varoquaux N, Cole B, Gao C, Pierroz G, Baker CR, Patel D, Madera M, Jeffers T, Hollingsworth** 1214 **J, Sievert J** . 2019. Transcriptomic analysis of field-droughted sorghum from seedling to maturity 1215 reveals biotic and metabolic responses. Proceedings of the National Academy of Sciences **116**, 1216 27124-27132. 1217 **Vasil V, Lu C-Y, Vasil IK** . 1985. Histology of somatic embryogenesis in cultured immature embryos of 1218 maize (Zea mays L.). Protoplasma **127**, 1-8. 1219 **Wang F, Wang W, Liu Z** . 2023a. Cyclin A participates in the TSO1-MYB3R1 regulatory module to 1220 maintain shoot meristem size and fertility in Arabidopsis. Development **150**, dev201405. 1221 **Wang N, Ryan L, Sardesai N, Wu E, Lenderts B, Lowe K, Che P, Anand A, Worden A, Van Dyk D** . 1222 2023b. Leaf transformation for efficient random integration and targeted genome modification in 1223 maize and sorghum. Nature plants **9**, 255-270. 1224 **Wang Z, Wang J, Cheng M, Du Y, Godwin I, Wang L, Lv P, Liu G** . 2026. Biotechnological and 1225 genetic innovations to enhance sorghum adaptation under climate change. Frontiers in plant science 1226 **17**, 1757792. 1227 **Yamaguchi M, Kubo M, Fukuda H, Demura T** . 2008. Vascular � related NAC � DOMAIN7 is involved 1228 in the differentiation of all types of xylem vessels in Arabidopsis roots and shoots. The Plant Journal 1229 **55**, 652-664. 38 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.07.764635; this version posted November 11, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.07.764635) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1230 **Zhang Q, Zhang Y, Lu M-H, Chai Y-P, Jiang Y-Y, Zhou Y, Wang X-C, Chen Q-J** . 2019. A novel 1231 ternary vector system united with morphogenic genes enhances CRISPR/Cas delivery in maize. Plant 1232 Physiology **181**, 1441-1448. 39