Optimized dCas9 programmable transcriptional activators for plants

Abstract

Understanding how gene expression impacts plant development and physiology is important for crop engineering. Programmable transcriptional activators (PTAs), including CRISPR-Cas activators, have relied on a limited number of transcriptional activation domains (ADs) (Casas-Mollano et al., 2020; Lowder et al., 2018; Pan et al., 2021; Papikian et al., 2019). Usually, the VP64 domain, derived from herpes simplex virus, is fused to a DNA-binding domain to activate target gene expression. In dCas9-based PTAs, binding to a target DNA sequence is afforded by the dCas9-sgRNA ribonucleoprotein. We reasoned there was considerable space for PTA improvement by replacing VP64 with a plant-derived AD. We compiled a list of ADs to test in dCas9-based PTAs in Arabidopsis and Seteria protoplasts as model dicot and monocot species (Figure 1a, Table S1) (Sychla et al., 2022). We started with a literature search of diverse plant-derived transcription factors with known DNA-binding domains. We computationally removed native DNA-binding domains, as these regions are highly conserved and are expected to produce off-target effects if retained in PTA fusions. If an AD was not empirically determined in the literature, we selected motifs enriched in acidic and/or aromatic residues. Acidic and aromatic residue patches are often associated with ADs due to their propensity to form phase separation condensates upon recruitment to a core promoter (Boija et al., 2018). We also added plant-evolved AD sequences from plant pathogen effector proteins such as TALE proteins from Xanthomonas, along with plant-derived sequences from transcription preinitiation complexes such as 14-3-3 scaffolding proteins. The ability of these ADs to function in the context of a dCas9-based PTA was tested in protoplasts from monocot and dicot species (Tables S2 and S3, Figures S1–S7). Specifically, we confirmed that in the Arabidopsis protoplast assay, the SunTag-PTA, in which ADs are recruited to dCas9 using scFv-epitope interactions, outperformed direct translational fusions of ADs to dCas9 (Figure S1). Splitting the genetic components of a dual-luciferase reporter assay across two plasmids decreased the correlated expression of each luciferase reporter that was observed in both Arabidopsis (Figure S2) and Setaria (Figures S3 and S4). AD performance was measured in both the two-plasmid reporter assay in Arabidopsis (Figure S5) and Setaria (Figure 1b) and the one-plasmid assay (Figure S6 for Setaria and Figure S7 for Arabidopsis). Nonspecific activation of the dual-luciferase reporter assay when ADs were expressed without the dCas9 was negligible compared with a positive control (Figure S8). While the specific rank order varied slightly between assays, the ADs DREB2, AvrXa10, DOF1, AtHSFA6b and DREB1 showed transactivation activity comparable or better than VP64 (Figure 1b). We next tested the ability of strong ADs to activate gene expression from endogenous promoters in the Arabidopsis genome (Figure 1c,d) using qRT-PCR. Testing multiple sgRNAs for the FT and CLV3 target genes yielded similar rank-order results, but different fold changes in transactivation (Figure S9). Single guide RNAs targeting the promoters of WUS, CLV3, FT and PAP1 were all capable of driving significant increases in expression. These target genes were selected for their potential use in engineering genetic incompatibility (EGI) in plants. EGI relies on PTA-driven over- or ectopic expression of developmental genes to cause hybrid lethality (Maselko et al., 2020). The largest fold changes in expression were seen for FT, which was overexpressed ten-thousand-fold. The smallest fold changes were seen for PAP1 (10-fold), but this is likely due to the high basal expression level of PAP1 in the absence of a PTA (no-AD control in Figure 1d). Transactivation of both FT and PAP1 yielded expression levels comparable to our housekeeping control, PP2A. Cumulative analysis of different ADs across all four endogenous gene targets shows that AvrXa10, DOF1, and DREB2 significantly outperform VP64 in protoplast experiments (Figure 1e). These ADs also work well in the context of different DNA-binding domains, for example TALEs (Figure 1f, Appendix S1). No activation of native targets of the ADs was seen (Figure S10). We next made transgenic Arabidopsis in which the developmental gene FT was targeted with MoonTag PTAs (Casas-Mollano et al., 2023) containing our newly characterized ADs. We switched to the MoonTag system for the transgenic plant studies because of our previous report that it is more stable than SunTag in transgenic plants (Casas-Mollano et al., 2023). MoonTag and SunTag PTAs are similar but differ in the molecular components used for the recruitment of ADs to the dCas9 (MoonTag uses nanobody:epitope interactions in place of the scFv:epitope interactions of SunTag systems). The behaviour of ADs might be altered in this new context. Both the dCas9::epitope-tail and the nanobody::AD components of the MoonTag PTA were expressed from a strong Ubi10 promoter, and two sgRNAs targeting the FT promoter were expressed from the AtU6-26 promoter. Following agrobacterium floral dip, T1 RFP-positive transgenic seeds were germinated on 1/2 MS agar plates and one to five seedlings from AvrXa10, DOF1, DREB2, VP64 and Col-0 genotypes were transplanted to soil. Four of five AvrXa10 plants displayed an early-flowering phenotype, compared with a single early-flowering plant of each DOF1 and DREB2 genotypes and none from the Col-0 control. We analysed RFP-positive transgenic plants (without determining hetero- or homozygosity) from the T2 generation to measure the flowering phenotype (Figure 1g, Figure S11). The four early-flowering AvrXa10 AD lines displayed varied degrees of FT overexpression via qRT-PCR in T2 plants, but all were statistically significant compared with wild type (Figure 1h). Similarly, all four lines showed a significant reduction in the days required to bolt (Figure 1i) and in the number of rosette leaves present at the time of bolting (Figure S12). This transgenic plant work is intended to demonstrate that overexpression of FT by the new AvrXa10 AD can produce the expected phenotype. A rigorous comparison of activation strength afforded by the plant-evolved ADs versus VP64 was not performed. Lastly, we explored the synergism between targeting sgRNAs at both a core promoter and enhancer region, using the AtFT gene as a model (Figure 1j–m). In combination with a core-promoter sgRNA, we tested second sgRNAs that targeted the Block C, B or E enhancer regions. For each enhancer, we also tested potential synergism from a second sgRNA that bound at an intermediate site outside of the enhancer (Figure 1j). For the Block B/E enhancers, there was a substantial synergism only when the sgRNA targeting the enhancer was co-expressed with a sgRNA targeting the core promoter. Block C showed a high amount of variance between experimental replicates, but the greatest expression of FT was similar to the co-expression of sgRNAs targeting the core promoter and enhancer. No activation of the neighbouring gene FAS1 was observed (Figure S13). Cumulatively, this work makes several key contributions to the field. We present multiple ADs that function well in the context of dCas9-based or TALE-based PTAs. While they provide stronger activation than the commonly used VP64 domain when tested in protoplasts, this quantitative difference is not the only impactful aspect. Sequences derived from endogenous plant genes typically face qualitatively different levels of regulatory scrutiny compared with sequences derived from human pathogens. Even among the new ADs we tested in PTAs, there exist potential qualitative differences in regulatory scrutiny. AvrXa10 is derived from Xanthomonas oryzae, the causal agent of rice bacterial blight. However, DOF1 and DREB2 are endogenous domains from maize and Arabidopsis, respectively. The pathway for transgenic PTA regulatory approval using endogenous plant domains like these would likely look different compared with domains from known pathogens. This work focused on only the AD components of PTAs, but future work that characterizes alternative DNA-binding domains from non-pathogenic sources would similarly benefit regulatory approval of PTAs. A second intriguing contribution of this work is the potential use of PTAs to map enhancer regions in plant genomes (Ricci et al., 2019). Discovering and mapping enhancer sequences can be challenging. Enhancer-capture reporter assays have traditionally been used but these change the genetic context of the regulatory elements. Conversely, PTA-mediated enhancer activation retains the original genomic architecture and chromatin state of the enhancer-promoter pair. As enhancers function through a DNA looping mechanism, retaining this genetic architecture is important. Our approach to enhancer mapping would simultaneously confirm the promoter in an enhancer-promoter pair, which can be difficult to determine with reporter assays. The true impact of PTAs in determining enhancer boundaries will require more extensive experiments at multiple loci; however, our preliminary data demonstrate a quick and effective method for activating known regulatory elements in protoplasts. MHZ and DFV are supported by U.S. Department of Energy grants DE-SC0018277 and DE-SC0023160. MHZ, JAC-M, AS, and MJS are supported by the USDA grant 2018-33522-28747 and are supported by the Advanced Plant Technologies program, DARPA Award HR001118C0146. MHZ and AS are supported by an NIH NIGMS Biotechnology Training grant NIHT32GM008347. MHZ, DFV, and MJS are co-inventors of a patent, WO2023215259A1, 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–S5 Figures S1–S13 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.

Topics & Concepts