Wuschel2 helps improve precise gene editing during fast new shoot growth in sorghum

Pleiotropic effects have been reported in transgenic maize plants carrying Bbm and Wus2 using integrated morphogenic gene-enabled transformation^3,6. To evaluate the impact of morphogenic gene expression on plant growth and fertility in Tx430, Tx623, and Tx2752, transgenic T0 plants transformed with pPHP79066 were sent to the greenhouse and grown to fertility. While subtle leaf curling occurred in Tx430, no severe pleiotropic effects were observed for all single copy T0 plants and all were fertile (Fig. 1m). There were, however, considerable pleiotropic effects observed for Tx623 and Tx2752 T0 plants (Fig. 1n, o). All the T0 Tx623 and Tx2752 plants were dwarf with strong curled leaves and reduced fertility, even for single-copy events. Because of the genotype-specific pleiotropic effects of morphogenic gene expression in sorghum, we decided to examine various transformation strategies that eliminate or preclude the morphogenic genes from the regenerating T0 plant.

WUS is a bifunctional transcription factor that mediates stem cell homeostasis by regulating stem cell number and patterns of cell division and differentiation^10,17. It acts as a transcriptional repressor in stem cell regulation and as a transcriptional activator in floral patterning¹⁰. In addition to causing pleiotropic effects on mature plant development as described above, it has been reported that morphogenic gene expression has a negative impact on transgenic plant regeneration through tissue culture^3,4,6. It was reported that higher transformation efficiency was correlated with improved plant recovery after Wus2/Bbm cassette excision indicating the morphogenic genes may act differently at different tissue culture stages^4,5. To maximize the positive, stimulatory effects of morphogenic genes on somatic embryo formation during tissue culture initiation, but to mitigate the repressive impacts on plant regeneration and rooting during the late tissue culture stage, an morphogenic gene excision-induced, selection-activation system was designed and successfully applied for maize transformation to generate high-quality, single-copy transgenic plants without pleiotropic impacts¹⁸. In brief, binary vector pPHP81814^12,18 (Supplementary Fig. 1b) was constructed with a multi-gene cassette containing Wus2 (Axig1_pro:Wus2), Bbm (Pltp_pro:Bbm), moCRE (Glb1_pro:moCRE) and Zs-GREEN1 (Sb-Ubi_pro:Zs-GREEN1) inserted into the first intron of the Hra gene, in which the Wus2/Bbm/moCRE/GREEN1 cassette was flanked by two directly repeated LoxP sites. This design interrupts the Hra gene and ensures that the full length Hra gene cannot be expressed to confer herbicide resistance until the multi-gene cassette between the two LoxP sites is removed. In other words, only those transgenic plants carrying at least one copy of the T-DNA with the Wus2/Bbm/moCRE/GREEN1 cassettes effectively excised can be regenerated under selective pressure. Because the Glb1 promoter driving moCRE is activated by ABA¹⁹ in the maturation medium (Supplementary Table 5) through embryo development^12,18,20, moCRE/loxP-mediated Wus2/Bbm/moCRE/GREEN1 cassette excision occurs during somatic embryo maturation, activating functional Hra expression before further regeneration¹⁸.

Most plant transformation technologies involve the random stable insertion of a transgene into the genome by coupling a trait gene with a selectable marker gene for antibiotic or herbicide resistance²¹. The presence of marker genes in transgenic plants, however, often provides no advantage after transformation and may raise regulatory concerns. It is desirable, therefore, to develop a highly efficient marker-free transformation technology for supporting the commercialization of genetically engineered crops²¹.

It was reported that Wus2 expression alone, driven by the maize Pltp promoter (hereby referred to as integrated Wus2-enabled transformation system that follows morphogenic gene-enabled transformation protocol as illustrated in Supplementary Fig. 2), was sufficient to rapidly stimulate somatic embryo formation and produce T0 maize plants¹². As an example in sorghum using pPHP96564 (Supplementary Fig. 1g), an integrated Wus2-enabled transformation vector carrying Pltp_pro:Wus2 and Zm-Ubi1_pro:NPTII as the selectable marker, we demonstrated transformation efficiencies of 58.7 and 15.8% in Tx430 and Tx623, respectively, of which 52.7% of the Tx430 events and 48.7% of the Tx623 events produced were single copy QEs (Table 1). Corroborating the results initially reported in maize, these results clearly demonstrate the efficacy of Pltp_pro:Wus2 for stimulating somatic embryo formation in sorghum, in the absence of Bbm, and suggests that this may be genotype-independent. Taking advantage of this observation, it was recently reported that a heat shock-induced moCRE/LoxP-mediated excision system was developed to efficiently eliminate both Wus2 and selectable marker gene by heat-shock treatment before the maturation stage in maize²² (hereby referred to as Wus2/CRE-enabled marker-free transformation system).

We tested this design in sorghum using binary vectors pPHP94632 and pPHP94292. As illustrated in Supplementary Fig. 1c and d, both binary vectors pPHP94632 and pPHP94292 contain Wus2 (Pltp_pro:Wus2), a heat shock inducible moCRE gene (Hsp17.7_pro:moCRE) and a selectable marker (Zm-Ubi_pro:NPTII) flanked by directly repeated loxP sites. pPHP94632 and pPHP94292 were identical plasmids, except for the proprietary trait gene expression cassette (Supplementary Fig. 1c, d). Following Agrobacterium infection and co-cultivation of Tx430 immature embryos, the immature embryos were then sub-cultured on multi-purpose medium with selection (250 mg l⁻¹ G418) for three weeks to induce somatic embryo formation (Supplementary Fig. 2). The intent of this tissue culture modification compared with the morphogenic gene-enabled transformation protocol (Supplementary Fig. 2) was to maximize transgenic somatic embryo formation, but at the same time minimize the proliferation of non-transgenic tissue so only the transgenic somatic embryos survived and germinated during maturation without selection. After the three-week-long stringent selection, excision of the Wus2/moCRE/NPTII cassette was carried out by inducing moCRE expression upon heat treatment at 45 °C and 70% humidity for 3 h (Supplementary Fig. 2). Following heat treatment, embryos were transferred to the maturation medium without selection for four weeks before rooting (Supplementary Fig. 2). As shown in Table 1, similar results were obtained for both binary vectors, pPHP94632 and pPHP94292. The three-week-long selection pressure during somatic embryo formation was stringent; except for two non-transgenic escapes (the escape frequency was between 0 and 3.7%), all 106 events from both experiments were transgenic. The heat shock-induced moCRE/LoxP-mediated excision system was highly efficient as well, with more than 59% of the events showing complete Wus2/moCRE/NPTII cassette excision (Table 1). Although the transformation efficiency was lower (in the range of 29.2–31.9%) compared to the excision-induced, selection-activation system described above (>60%), the overall QE frequency (single copy, backbone-free, and Wus2/moCRE/NPTII cassette-free) was in the range of 31.4–35.2% (Table 1). The high QE frequency of this marker-free system most likely resulted from the combined impact of highly efficient excision and extremely low non-transgenic escape frequency.

WUS protein is non-cell-autonomous, migrating between cells via plasmodesmata^9,23. It has been demonstrated that the movement of WUS protein stimulates somatic embryo formation of neighboring cells^11,12. Taking advantage of this observation, a novel transformation system known as “altruistic” transformation, has been developed that utilizes the mixing of two Agrobacterium (Agro-1 and Agro-2), both being the same strain (LBA4404 Thy-) containing the accessory plasmid pPHP71539, but each Agrobacterium carrying one of two different plasmids for transformation¹². Agro-1 carries an “altruistic” T-DNA binary plasmid with a Wus2 expression cassette but lacking a selectable marker gene, while Agro-2 carries a conventional binary T-DNA plasmid with a selectable marker and a gene-of-interest. Transformation is performed by mixing Agro-2:Agro-1 at a 9:1 ratio. During the initiation of transformation, the “altruistic” binary plasmid stimulates somatic embryogenesis in a non-cell-autonomous manner, and a morphogenic gene-enabled transformation protocol (Supplementary Fig. 2) can be followed for rapid and highly efficient transformation through somatic embryo formation from neighboring non-Wus2-containing cells. Using this method, the preponderance of regenerated plants contains the gene-of-interest T-DNA, but not the altruistic T-DNA containing Wus2 (hereby referred to as non-integrated Wus2-enabled altruistic transformation system).

Targeted genome modification using the CRISPR/Cas system has proven to be a powerful tool for crop engineering and has been successfully applied to maize, rice, sorghum, and numerous other plant species to generate stable genome-edited events with targeted modifications (e.g., insertions, deletions, and replacements)^25–27. An efficient transformation system is a prerequisite for CRISPR/Cas-mediated genome editing in plants²⁸. Taking advantage of the highly efficient transformation mediated by morphogenic gene expression, we further conducted CRISPR/Cas-mediated genome editing and compared the targeted editing frequency to that of conventional transformation using the same guide-RNA(s) in sorghum. All the T0 plants were analyzed by deep sequencing to identify mutations at each target site and gene dropout between the two target sites (“Methods”). In the first experiment, sorghum ortholog of Mtl²⁹ (Matrilinial Sobic.001G348600) gene was mutated by targeting with a single guide RNA (sgRNA) (Supplementary Fig. 3). As shown in Supplementary Table 8, using the same sgRNA (Sb-Mtl-CR1), the mutation frequency in Tx430 increased 2.7-fold from 27.5% for conventional transformation using plasmid pPHP86801 (Supplementary Fig. 1j) compared to 74.2% when using the integrated morphogenic gene-enabled transformation plasmid pPHP87098 (Axig1_pro:Wus2/Pltp_pro:Bbm) (Supplementary Fig. 1k). In the second experiment, dropout a segment of the Kaf (γ-Kafirin, Sobic.002G211700) gene³⁰ was tested using two sgRNAs (Sb-Kaf-CR1 and -CR3) (Supplementary Fig. 4) in Tx430. We compared both mutation frequency for each sgRNA and the gene-dropout frequency of those events generated by conventional transformation using plasmid PHP87018↗ (Supplementary Fig. 1l) and the non-integrated Wus2-enabled altruistic transformation (90% pPHP87018 + 10% altruistic vector pPHP88158 (Supplementary Fig. 1f)). We found that the mutation frequencies for each sgRNA, Sb-Kaf-CR1 and -CR3, were increased from 39.3 and 48.5% for conventional transformation to 63.6% and 66.7% for non-integrated Wus2-enabled altruistic transformation, respectively (Supplementary Table 9). One gene-dropout event was identified using non-integrated Wus2-enabled altruistic transformation, but none were recovered from conventional transformation (Supplementary Table 9). The third experiment was designed to delete the Lgs1 (Low germination stimulant 1, Sobic.005G213600) gene^13,31 by gene dropout using two sgRNAs, Sb-Lgs1-CR1 and -CR3 (Supplementary Fig. 5), in the Macia background. Macia is barely transformable without morphogenic genes (0.7% transformation efficiency, as shown in Table 1). To generate enough edited events in Macia, we conducted integrated morphogenic gene-enabled transformation and non-integrated Wus2-enabled altruistic transformation and then compared mutation frequency for each sgRNA as well as the gene-dropout frequency. As shown in Supplementary Table 10, higher mutation frequencies of 83.3 and 92.9% at each target site and a higher gene-dropout frequency of 52.4% were observed when using integrated morphogenic gene-enabled transformation with pPHP87984 (Supplementary Fig. 1m) carrying Axig1_pro:Wus2/Pltp_pro:Bbm. This compared to mutation frequencies of 40.4 and 57.8% and a 29.2% gene dropout frequency when using the non-integrated Wus2-enabled altruistic transformation method (90% pPHP87980 (Supplementary Fig. 1i) + 10% altruistic vector pPHP88158 (Supplementary Fig. 1f)).

Collectively, those results suggested a trend for improved gene editing frequency when utilizing Wus2 during transformation. Namely, the highest frequency of genome editing was observed when using an integrated morphogenic gene-enabled transformation compared to that of non-integrated Wus2-enabled altruistic transformation, while the conventional transformation without Wus2 had the lowest gene-editing frequency. Assuming that plant cells harboring an integrated morphogenic gene accumulate more WUS2 protein than neighboring cells receiving WUS2 protein via migration from a cell expressing Wus2-enabled altruistic T-DNA, we hypothesize that CRISPR/Cas-mediated genome editing frequency may correlate with the level of WUS2 protein accumulated in the transgenic cells and that WUS2 facilitates more efficient genome editing.

The premise that Cas-mediated genome editing was stimulated during Wus2-mediated somatic embryogenesis was strengthened by being based on the edited data for four different genomic loci using seven total sgRNAs across four different chromosomes in two sorghum varieties, Tx430 and Macia. Though Wus2 expression is sufficient to stimulate somatic embryo formation in tissue culture, the mechanism of how Wus2 enhances CRISPR/Cas-targeted genome editing is not fully understood. One possibility could be that Wus2-enabled transformation may increase the chance of multi-copy T-DNA integration, in turn leading to enhanced Cas9 protein accumulation in the cells and resultant higher editing frequency. We excluded this possibility based on two observations. We first compared the QE frequencies (single-copy for T-DNA) among those transformation systems. As shown in Table 1, similar QE frequencies 58.5, 52.7, and 56.3% were observed for all three transformation systems using pPHP86655, 90% pPHP86655 plus 10% altruistic vector pPHP87078 and pPHP96564 (Pltp:Wus2), respectively, indicating that despite enhancing transformation efficiency, Wus2-enabled transformation had no impact on integrated T-DNA copy number in sorghum. We then compared the targeted genome editing frequency among the single-copy QEs exclusively to eliminate any potential issues related to multi-copy events. Consistent with above observation, the integrated Wus2-enabled transformation treatment once again showed the highest mutation frequency at both target sites (89.9 ± 4.0%) and gene-dropout frequency (47.9 ± 18.4%) when calculated for single-copy QE events alone (Table 2). This is followed by the non-integrated Wus2-enabled altruistic transformation treatment with an mutation frequency of 62.4 ± 4.2 and 65.8 ± 1.3% for each target site and a gene-dropout frequency of 22.8 ± 6.5% when calculated for the single-copy QE events (Table 2). Finally, conventional transformation produced mutation frequencies of 70.8 ± 12.9% at both target sites and the lowest gene-dropout frequency of 3.1 ± 2.3% when calculated for the single-copy QE events (Table 2). The consistency of these multiple lines of evidence described above strongly supports that Wus2 expression enhances CRISPR/Cas-mediated genome editing in sorghum.

Cell division and dedifferentiation are two important processes for de novo shoot regeneration^8,34–36. The activation of the homeodomain transcription factor WUS has been demonstrated to facilitate cell division and dedifferentiation to establish shoot stem cell niches during de novo shoot regeneration^8,37. Compared to non-dividing cells that typically have more condensed DNA and stable nucleosomes, a constantly dividing cell rapidly replicates its DNA and undergoes constant chromatin reconfiguration during each cell cycle³⁸. It is also known that dedifferentiation is regulated by various epigenetic mechanisms, such as histone acetylation which increases the accessibility of transcriptional regulators to target DNA and contributes to the regulation of gene expression³⁹. Therefore, dedifferentiating plant cells possess more open, decondensed chromatin architecture^37,39,40. Recently it is reported that WUS can also act via regulation of histone acetylation at its target loci in Arabidopsis⁴¹, indicating that WUS activation may alter chromatin accessibility. Since CRISPR/Cas9-mediated editing is more efficient in open chromatin regions than in closed chromatin regions^42–44, it is reasonable to speculate that WUS may, either directly or indirectly, increase chromatin accessibility thereby facilitating double-strand break (DSB) targeted by CRISPR/Cas among those cells stimulated to become somatic embryos during tissue culture. Higher DSB frequency leads to higher mutation efficiency at each target site and therefore, results in higher gene-drop out efficiency through the NHEJ chromosomal repair pathway. We did notice that although similar mutation efficiency was observed at each target site for both non-integrated Wus2-enabled altruistic transformation and conventional transformation, the gene-drop efficiency was significantly higher for non-integrated Wus2-enabled altruistic transformation as shown in Table 2. One explanation is that the gene-drop out process is more complex and requires coordinated and synchronized DSB at both target sites. The open chromatin in that region could provide a better environment for the gene fragment to be released once DSB on both targeted sites happens simultaneously. Otherwise, NHEJ repair may occur quickly and result in indel mutation at each target site preventing the gene fragment from being released. A carefully designed ATAC-seq assay may shed light on delineating the mechanism of how Wus2 enhances CRISPR/Cas-mediated genome editing in the future.

Agrobacterium auxotrophic strain LBA4404 Thy- carrying a ternary vector transformation system was used to generate transgenic sorghum plants and for CRISPR/Cas-mediated genome modifications. The ternary vector system contains the T-DNA binary vector and pVIR accessory plasmid (pPHP71539) as previously described by Che et al.¹⁴ and Anand et al.¹⁵ The T-DNA binary plasmid pPHP79066 (Supplementary Fig. 1a) carrying both Wus2 and Bbm (Axig1_pro:Wus2/Pltp_pro:Bbm) as morphogenic genes, Zs-YELLOW1 (Ltp2_pro:Zs- YELLOW1) as the visual marker and Hra (Sb-Als_pro:Hra) as the selectable marker was reported previously by Lowe et al.³ The T-DNA binary plasmid pPHP81814 (Supplementary Fig. 1b) reported by Chu et al.¹⁸ was a multi-gene cassette containing the Wus2 (Axig1_pro:Wus2), Bbm (Pltp_pro:Bbm), moCRE (Glb1_pro:moCRE) and Zs-GREEN1 (Sb-Ubi_pro:Zs-GREEN1) inserted in the first intron of Hra gene. The Wus2/Bbm/moCRE/GREEN1 cassette was flanked by two directly repeated LoxP sites. Binary vectors pPHP94632 and pPHP94292 were used for marker-free transformation in which the excision cassette contains Wus2 (Pltp_pro:Wus2), a heat shock inducible moCRE gene (Hsp17.7_pro:moCRE) and a selectable marker (Zm-Ubi_pro:NPTII) flanked by the directly repeated loxP sites (Supplementary Fig. 1c, d). Both marker-free T-DNA binary plasmids pPHP94632 and pPHP94292 were identical plasmids, except the same proprietary trait gene (Trait) was driven by different maize promoters (Supplementary Fig. 1c, d). Altruistic binary vectors pPHP87078 and pPHP88158 (Supplementary Fig. 1e, f) were reported previously by Hoerster et al.¹².

CRISPR/Cas gene modification was also achieved using the LBA4404 Thy- Agrobacterium strain carrying ternary vector transformation system described above. The SpCas9 and sgRNA gene modification machinery were expressed on a TDNA expressing binary vector, such as pPHP96564, pPHP86655, pPHP87980, pPHP86801, pPHP87098, pPHP87018, and pPHP87984 genome modification binary vectors described in this study (Supplementary Fig. 1g–m). The sgRNA target sites, sequences, and corresponding PAM for Sb-Mtl, Sb-Kaf, Sb-Lgs1, and Sb-Bmr6 genes were illustrated in Supplementary Figs. 3–6.

The ternary design was assembled by first mobilizing the accessory plasmid pPHP71539 in the Agrobacterium auxotrophic strain LBA4404 Thy- and selecting on media supplemented with gentamycin (25 mg l⁻¹). Subsequently, the binary constructs were electroporated into Agrobacterium strain LBA4404 Thy- containing the accessory plasmid and recombinant colonies were selected on media supplemented with both gentamycin and spectinomycin. All constructs were then subjected to next-generation sequencing and sequence confirmation before conducting transformation experiments.

Sorghum genotypes Tx430, Tx623, and Tx2752 and African genotypes (Macia, Malisor 84-7, and Tegemeo) grown in a greenhouse were used in this study. Tx430, Tx623 and Tx2752, Malisor 84-7 (PI 656048), and Macia (PI 565121) were requested from USDA GRIN. Tegemeo is a public line obtained from Mycogen Seeds. Immature embryo explants isolated from those sorghum plants were transformed with Agrobacterium auxotrophic strain LBA4404 Thy- a carrying a ternary vector transformation system to generate transgenic sorghum plants^14,15. The conventional transformation method was performed as previously described by Wu et al.¹⁶ and Che et al.¹⁴ Morphogenic gene-mediated transformation was conducted by modifying the conventional sorghum transformation protocol, eliminating nine-weeks from the protracted selective callus proliferation phase as illustrated in Supplementary Fig. 2. The Wus2/CRE-mediated marker-free transformation tissue culture procedure is illustrated in Supplementary Fig. 2. As illustrated Supplementary Fig. 2, following the Agrobacterium infection and co-cultivation steps, the immature embryos were subcultured on the multi-purpose medium with selection for three weeks to induce somatic embryo formation. After three weeks of stringent selection, Wus2/moCRE/NPTII cassette excision was carried out by heat-shock treatment at 45 °C for 3 h (Supplementary Fig. 2). Following heat-shock treatment, embryos were transferred to the maturation medium without selection for four weeks before rooting (Supplementary Fig. 2). All the media described in this study for sorghum transformation were indicated in Supplementary Fig. 2 and attached in the supplementary tables (Supplementary Tables 1–6).

The integrated copy number of the T-DNA and the vector backbone in these transgenic plants were determined by a series of qPCR analyses^16,45. For conventional and morphogenic gene-mediated transformations, the transgenic plant carrying a single copy of the intact T-DNA integrations without vector backbone was defined as ‘quality event (QE)’. For morphogenic gene-mediated excision-induced selection-activation transformation, the transgenic plant carrying a single copy of the intact T-DNA integrations without vector backbone and with no Wus2/Bbm/moCRE/GREEN1 sequences was defined as ‘quality event (QE)’. For Wus/CRE-mediated marker-free transformation, the transgenic plant carrying a single copy of the intact T-DNA integrations without vector backbone and Wus2/moCRE/NPTII cassette was defined as ‘quality event (QE)’. For altruistic transformation, the transgenic plant carrying a single copy of the intact T-DNA integrations without vector backbone and altruistic T-DNA co-integration was defined as ‘quality event (QE)’.

Images were taken using a dissecting Leica M165 FC stereo-epifluorescence microscope, with YFP or GFP filters for detection of fluorescence, using the PLANAPO 1.0× objective, 0.63× zoom, and Leica Application Suite V4.7 acquisition software.

To detect mutations and gene-dropouts induced by the CRISPR/Cas9, T0 events generated through transformation were characterized by amplicon sequencing from DNA extracted from a single fresh leaf punch from each plant as per manufacturer recommendations via the sbeadexTM tissue extraction system (LGC Limited, UK). DNA was normalized to 10 ng/µl and twenty cycle target region PCR was performed on 50 ng of genomic DNA with PhusionTM Flash 2x master mix (Thermo Scientific, Waltham, MA) as per the manufacture’s recommendations. Five microliters of primary PCR product were transferred to twenty cycles secondary amplification containing primers to attach individual sample indexes and sequencing components, again with PhusionTM Flash 2x master mix. Sequencing was performed on Illumina MiSeq^®, paired-end 150 cycles per read according to Illumina standard operating procedure. Sequence reads were aligned to the wildtype reference sequence via Bowtie2. The allele return threshold was set to 5%. Mutation is defined as mutagenesis at target site, including both in-frame and frameshift mutations. Mutation frequency was calculated as the number of mutated events divided by the total number of events generated in a transformation experiment. Gene-dropout is defined as gene deletion between the two target sites. Gene-dropout frequency was calculated as the number of gene-drop events divided by the total number of events generated in a transformation experiment.

Further information on research design is available in thelinked to this article. Nature Research Reporting Summary

Peer Review File SUPPLEMENTAL MATERIAL Supplementary Data 1 Reporting Summary