Towards engineering hybrid incompatibility in plants

The potential for gene flow between genetically modified organisms (GMOs) and non-GMO relatives poses a significant challenge to the development and regulatory approval of GMO crops (Wedger et al., 2024), for example, the spread of herbicide resistance transgenes from crops such as rice or sorghum to cross-pollinating weedy species. Addressing this concern, we developed Engineered Genetic Incompatibility (EGI) (Maselko et al., 2017), a system that establishes species-like barriers to gene flow between otherwise sexually compatible populations. EGI employs Programmable Transcriptional Activators (PTAs) to drive lethal over- and/or ectopic expression of tightly regulated genes following undesired hybridization events (Figure 1a,b). A benign mutation of the target promoter in the EGI organism protects it from ill effects of the PTA, which acts as a sentinel for the wild-type (WT) promoter sequence. Given numerous potential PTA targets, multiple mutually incompatible subpopulations are feasible (Maselko et al., 2020). EGI has been demonstrated in yeast as a proof-of-concept (Maselko et al., 2017) and in insects as a strategy for genetic biocontrol of pests (Maselko et al., 2020; Upadhyay et al., 2022). EGI in plants would provide a strategy to halt gene flow between engineered crops and their domestic and wild relatives without altering normal cultivation or propagation practices. Here, we present promising results towards the demonstration of EGI in plants and highlight technical challenges that still need to be overcome. In the model plant, Arabidopsis thaliana, we targeted the WUSCHEL (AtWUS) gene for the development of EGI. AtWUS is crucial for early embryo formation and development. Ectopic, postembryonic AtWUS expression induces aberrant somatic embryogenesis, resulting in callus formation or post-germination growth arrest and death (Zuo et al., 2002). Prior work has shown that single guide RNAs (sgRNAs) targeting PTAs to regions upstream of the transcriptional start sites (TSS) trigger strong transcriptional activation (Casas-Mollano et al., 2023). Thus, we targeted four sites within 500 bp of the TSS in AtWUS (Figure 1c) for activation with the SunTag PTA (Papikian et al., 2019). SunTag utilizes activation domain (AD) scaffolding, where a VP64 AD is fused to a scFv antibody while a multimeric GCN4 epitope tail is fused to dCas9. Each scFv antibody recognizes a GCN4 epitope, recruiting multiple ADs to a single dCas9 molecule (Papikian et al., 2019). We crossed a PTA expressing line with a line expressing 4 sgRNAs targeting the AtWUS promoter. The expected ectopic embryo formation in the F1 progeny was observed (Figure S1), giving us confidence to pursue EGI using WUS overexpression. Individual sgRNAs were expressed in Arabidopsis protoplasts with sgRNA1 showing the strongest AtWUS activation (Figure 1c), guiding subsequent EGI construction. We then mutated the AtWUS promoter to block PTA binding in the EGI plant. A vector (Figure S2) expressing a catalytically active Cas9, sgRNA1 and the RFP-based FAST marker gene (Shimada et al., 2010) for Cas9 selection was used to transform Arabidopsis plants. Screening RFP-positive T1 plants confirmed multiple unique fixed biallelic AtWUS promoter indels across 30 T1 lines. After segregating out Cas9 in the T2 generation (RFP-negative plants), we advanced two lines for EGI prototyping that have homozygous AtWUS promoter indels (Figures 1d and S3) and no apparent phenotype. We transformed these lines with a T-DNA containing the PTA, sgRNA1 targeting the WT AtWUS promoter and FAST marker (Figure S4). Evidence of successful transformation included accumulation of RFP in seeds (Figure S5). This EGI system uses MoonTag PTAs, which we have found to outperform SunTag PTAs in transgenic plants (Casas-Mollano et al., 2023; Zinselmeier et al., 2024). After floral dip transformation, RFP+ T1 seeds were germinated in soil, showing no obvious phenotype, suggesting that the promoter mutations successfully prevent PTA binding and subsequent AtWUS over- or ectopic expression. We next wanted to test if the molecular components of EGI are functional in these T1 lines even though they do not contain a complete EGI genotype. They are homozygous for the protective promoter mutation, but presumably hemizygous for the PTA construct. Homozygosing the PTA construct would generate a complete EGI genotype. However, the T1 plants provide a convenient screening strain to test for hybrid lethality via outcrosses to WT plants. Such crosses yield approximately equal numbers of RFP+ seeds (that inherit the PTA construct) and RFP- seeds (that do not inherit the PTA construct) to serve as age- and treatment-matched negative controls. We expect poor fitness in the RFP+, but not in the RFP- seedlings. We crossed the T1 lines with WT plants to generate hybrid lines, named T1-Hyb (Figures 1e and S5). RFP+ T1-Hyb seedlings were smaller than parental EGI lines, exhibiting cotyledon browning or ectopic embryos from roots, whereas RFP- seedlings appeared normal. Seedlings with cotyledon browning died upon longer incubation. Upon transfer of non-necrotic seedlings to soil, some died, while survivors were smaller, with abnormal flowers and siliques (Figure 1f). Because we tested independent T1 EGI lines, the number of seedlings with phenotypes varied for each cross (Table S1). Many RFP+ seedlings that appeared normal after germination showed delayed or abnormal development when transferred to soil. Phenotypic observations suggest AtWUS over- or ectopic expression in T1-Hyb seedlings. Endogenous AtWUS expression is restricted to meristematic regions (Mayer et al., 1998) and is absent from most differentiated tissues. AtWUS showed minimal expression in WT, EGI and RFP- T1-Hyb seedlings. RFP+ T1-Hyb exhibited AtWUS overexpression, albeit to differing degrees (Figure S6a). The variation is partially explained by variations in expression of the MoonTag components (Figure S6b). AtWUS expression in T1-Hyb lines was used to select which T1 lines to homozygose at the PTA allele to make complete EGI genotypes. Several seedlings from crosses between T2 homozygous EGI lines and WT plants (named T2-hyb) displayed a yellowing phenotype, delayed germination and slower growth compared to WT plants (Figure 1g–i). Once again, the yellowing seedlings died on the plates before they could be transferred to soil. However, no ectopic embryogenesis was observed, unlike in T1-Hyb. T2-Hyb plants showed ~100-500-fold AtWUS activation compared to Col-0 and EGI lines, which correlates with the yellowing phenotype (Figure 1j). T2-hyb seeds were planted directly to soil, and they were severely delayed in germination and growth (Figure S7). EGI T2 lines also exhibited delayed germination, albeit to a lesser extent (Figure S7). EGI has been demonstrated to provide complete hybrid lethality in model insects. Substantial differences in the developmental biology of plants and insects may require modifications to the system before gene flow can be controlled via this approach in plants. We were able to drive up to 500-fold overexpression of our target gene, AtWUS, only in hybrid plants (Figure 1j). Unfortunately, this was not sufficient for complete hybrid lethality. New target genes whose over- or ectopic expression is more potently lethal may need to be identified. Alternatively, driving higher expression of AtWUS by targeting the promoter with multiple sgRNAs, or driving MoonTag expression using more effective promoters, is expected to increase hybrid lethality. Interestingly, there is a striking growth phenotype in the T2 EGI lines compared to WT that is only apparent when seeds are germinated directly in soil (e.g. Figure 1i vs. Figure S7). Similar fitness defects were not observed in other systems (Maselko et al., 2017, 2020). This cannot be explained by activation of the target gene, as AtWUS is expressed at WT levels in the EGI lines. Potentially, off-target effects of expressing the MoonTag PTA are the cause. Controlling gene flow in plants is a challenging but impactful goal that will have myriad applications in agriculture and biomanufacturing. These promising results not only confirm that the underlying components needed to engineer a genetic incompatibility function as required in plants, but also show that the system needs to be optimized prior to use for biocontainment. MHZ and DFV are supported by the Department of Energy Office of Science, Office of Biological and Environmental Research, Grant No. DE-SC0023160. MHZ, JAC-M, JC, SSF and MJS are supported by the Department of Energy Office of Science, Office of Biological and Environmental Research, Grant No. DE-SC0023142, USDA grant 2018-33522-28747 and by the Advanced Plant Technologies program, DARPA Award HR001118C0146. MHZ is supported by an NIH NIGMS Biotechnology Training grant NIHT32GM008347. M.H.Z., D.F.V. and M.J.S. conceived this study. M.H.Z., J.A.C.-M. and J.C. designed and conducted experiments. M.H.Z., J.A.C.-M., S.S.F., D.F.V. and M.J.S. analysed data. S.S.F. and M.J.S. wrote the manuscript and all authors provided revisions. The editors have granted permission for both M.H.Z. and J.A.C.-M. to list their names in the first position when reporting this manuscript on their CVs. All authors read and approved the final manuscript. M.H.Z., J.A.C.-M. and M.J.S. are co-inventors of a provisional patent, No. 63/693796, regarding the work in this publication. The data that supports the findings of this study are available in the supplementary material of this article. Tables S1–S4. Figures S1–S7. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.