Programmable genome editing tools, which include zinc-finger nucleases (ZFNs)1, transcription activator-like effector nucleases (TALENs)2, clustered regularly interspaced short palindromic repeat (CRISPR) systems3–6, and base editors7–9 composed of the catalytically-deficient CRISPR-associated protein 9 (Cas9) variant and a nucleobase deaminase protein, have been developed for plant genetic studies and crop improvements through manipulation of genomic DNA sequences. However, these tools come short of editing DNA sequences in plant organelles, including mitochondria and chloroplasts, possibly because there are no efficient DNA double-strand break (DSB) repair systems in organelles and also because it is difficult to deliver both guide RNA and the Cas9 protein to organelles or to express the two components in organelles simultaneously. Plant organelle genomes encode hundreds of genes essential for photosynthesis and respiration. Methods or tools for editing these genes in organelles are highly desired for studying the functions of these genes and improving crop productivity. Recently, Mok et al.10 demonstrated that CRISPR-free DddA-derived cytosine base editors (DdCBEs) enable targeted C∙G-to-T∙A base substitutions in mitochondrial DNA in mammalian cells. DdCBEs composed of non-toxic split domains of the bacterial cytidine deaminase toxin, DddAtox, a custom-designed transcription activator-like effector (TALE) array, and a uracil glycosylase inhibitor (UGI), function as heterodimers to catalyze cytosine deamination, inducing C-to-T conversions, within a spacer region between the two TALE protein binding sites in target DNA. In this study, we present a rapid and convenient system to assemble DdCBE plasmids for expression in mitochondria and chloroplasts and use the resulting DdCBEs to edit mitochondrial DNA and chloroplast DNA, respectively, demonstrating highly efficient organelle base editing in plants for the first time.
To this end, we first developed a Golden Gate assembly system to construct chloroplast-targeting DdCBE (cp-DdCBE) plasmids or mitochondrial-targeting DdCBE (mt-DdCBE) plasmids (Fig. 1). Our expression plasmids encode fusion proteins composed of a chloroplast transit peptide (CTP) or a mitochondrial targeting sequence (MTS), the TAL effector N- or C-terminal domains, split-DddAtox halves (G1333N, G1333C, G1397N and G1397C) and UGI, which are codon-optimized for expression in dicot plants, under the control of the parsley ubiquitin (PcUbi) promoter and pea3A terminator. DdCBE plasmids with custom-designed TALE DNA-binding arrays can be constructed in a single sub-cloning step by mixing an expression vector and six TALE sub-array plasmids in an Eppendorf tube. A total of 424 (= 6 x 64 tripartite + 2 x 16 bipartite + 2 x 4 monopartite) modular TALE sub-array plasmids11 are available for making cp-DdCBEs or mt-DdCBEs that recognize DNA sequences of 16–20 base pairs (bps) in length, including a conserved T at the 5’ terminus. As a result, a functional DdCBE heterodimer recognizes 32- to 40-bp DNA sequences.
To assess whether our DdCBEs can promote base editing in chloroplasts, we constructed four pairs of cp-DdCBE plasmids specific to the chloroplast 16S ribosomal RNA (rRNA) gene encoding the RNA component of the 30S ribosomal subunit, co-transfected each pair into lettuce and rapeseed protoplasts, and measured base editing efficiencies using targeted deep sequencing at day 7 post-transfection (Fig. 2a, 2b). The best-performing cp-DdCBE pair (Left-G1397-N + Right-G1397-C) induced C∙G-to-T∙A conversions in the 15-bp spacer region between the two TALE array-binding sites at frequencies of 30.2% in lettuce protoplasts and 8.6% in rapeseed protoplasts (Fig. 2b). In line with the previous results in mammalian cells, cytosines (C9 and C13) in a 5’-TC motif were converted to thymine preferentially by this cp-DdCBE. Interestingly, a cytosine (C7) in a 5’-AC context was changed to thymine at a frequency of 4.0% in lettuce protoplasts by another cp-DdCBE (Left-G1333-N + Right-G1333-C).
We also tested base editing in two additional chloroplast genes, psbA and psbB, which encode the photosynthetic proteins, D1 and CP-47, respectively, of Photosystem II (PSII) (Fig. 2c, d, Supplementary Fig. 2). Among four cp-DdCBEs targeted to the psbA gene, the most active one (Left-G1397-C + Right-G1397-N) was able to induce C∙G-to-T∙A conversions in lettuce protoplasts with frequencies of up to 24.2 % (Fig. 2d). Only the two cytosines (C11 and C12) in a 5’-TCC context were efficiently converted to thymines by this base editor. It is possible that 5’-TCC was first converted to 5’-TTC and then to 5’-TTT. In rapeseed protoplasts, the other split pair (Left-G1333-N + Right-G1333-C) was most active at four cytosine positions (C3, C4, C11, and C12) with editing efficiencies of up to 3.5% (C3). Note that C3 and C4 are in a 5’-TCC context in the rapeseed gene, whereas they are in a 5’-ACC context in the lettuce counterpart, owing to a single nucleotide polymorphism, which is responsible for efficient editing of the two cytosines (C3 and C4) in the rapeseed gene but not in the lettuce gene by this DdCBE. Likewise, the cp-DdCBE pair targeted to the psbB gene catalyzed conversion of two cytosines in a TCC context at editing frequencies of 1.2–4.1% in rapeseed protoplasts (Supplementary Fig. 2). Taken together, these results suggest that editing efficiencies are dependent on cytosine positions and contexts within a spacer region as well as DddAtox split positions (G1333 vs. G1397) and orientations (Left-G1333-N vs. Left-G1333-C) and demonstrate that our cp-DdCBEs enable efficient base editing in the chloroplast genome in plants.
Next, we sought to achieve base editing in plant mitochondrial DNA using our custom-designed mt-DdCBEs. To this end, we constructed mt-DdCBE-encoding plasmids, using our Golden Gate cloning system, targeted to the ATP6 and RPS14 genes in the rapeseed mitochondrial genome, transfected the resulting plasmids into rapeseed protoplasts, and measured base editing frequencies using targeted deep sequencing at day 7 post-transfection (Fig. 2e, f, Supplementary Fig. 3). Our mt-DdCBEs were able to catalyze C∙G-to-T∙A conversions at a frequency of 13.2% in ATP6 (Fig. 2f) or 9.5% in RPS14 (Supplementary Fig. 3). These results show that mitochondrial DNA in plants is amenable to base editing with mt-DdCBEs.
To investigate whether DdCBE-mediated edits in cpDNA and mtDNA were maintained during regeneration, we collected lettuce and rapeseed calli regenerated from DdCBE-treated protoplasts, four weeks after transfection (Fig. 3a), and measured base editing efficiencies in each callus using targeted deep sequencing and Sanger sequencing (Fig. 3b and Supplementary Fig. 4, 5). Base edits induced by the DdCBE specific to the chloroplast or mitochondrial genes were detected in 22 out of 26 lettuce calli or 7 out of 14 rapeseed calli with frequencies of up to 38.4% or 24.9%, respectively (Fig. 3c). Also, base edits in the chloroplast psbA gene were observed with frequencies of up to 3.9% in lettuce calli (Supplementary Fig. 4). Likewise, mitochondrial base edits were detected in rapeseed calli with frequencies of up to 24.9% and 1.9% in the ATP6 and RPS14 target sites, respectively (Supplementary Fig. 5). These results show that DdCBE expression in plant protoplasts is not cytotoxic and that organelle base edits induced by DdCBEs in protoplasts remain intact during regeneration.
Encouraged by the stable maintenance of organelle edits in calli regenerated from protoplasts, we investigated whether the chloroplast DNA edits in the 16S rRNA gene could confer resistance to streptomycin, an antibiotic that binds to 16S rRNA irreversibly, leading to inhibition of protein synthesis. Several SNPs in the 16 s rRNA gene are commonly observed in streptomycin-resistant prokaryotes and eukaryotes; in particular, the 16S rRNA C860T (Escherichia coli coordinate C912) mutation endows Nicotiana tabacum (tobacco) with resistance to streptomycin12. The nucleotide affected by the C860T point mutation in tobacco and the equivalent nucleotide in lettuce correspond to the C9 position in Fig. 2b and 3b. We transferred lettuce calli regenerated from DdCBE-treated protoplasts to medium containing streptomycin (Fig. 3d). Mock-treated calli turned white, indicative of protoplast dysfunction, upon exposure to streptomycin. In contrast, DdCBE-treated calli remained greenish, showing resistance to streptomycin. Apparently, the heteroplasmic state harboring the C860T mutation was enough to confer resistance to the lettuce calli. Taken together, these results show that plant organelle heteroplasmy generated by DdCBEs in protoplasts can be maintained after cell division and callus development.
Last but not least, we sought to demonstrate DNA-free base editing in organelles using in vitro transcribed cp-DdCBE mRNA rather than expression plasmids. We transfected in vitro transcripts encoding the cp-DdCBE targeted to the 16S rRNA gene into lettuce protoplasts and analyzed base editing frequencies at the target site (Fig. 3a). C-to-T mutations were detected in protoplasts with frequencies of up to 19.0% (Fig. 3e and Supplementary Fig. 6). This method can avoid potential integration of plasmid DNA fragments in the host genome.
In conclusion, we have developed a Golden Gate cloning system, which employs a total of 424 TALE sub-array plasmids and 16 expression plasmids, to assemble DdCBE-encoding plasmids for organelle base editing in plants. Our DdCBEs custom-designed to target three genes in chloroplast DNA and two genes in mitochondrial DNA achieved C-to-T conversions at high frequencies in lettuce and rapeseed protoplasts. Importantly, the edits in plant organelles were maintained during cell division and callus development. Furthermore, we were able to obtain streptomycin-resistant lettuce calli by inducing a C860T mutation in the chloroplast 16S rRNA gene. We expect that our Golden Gate cloning system will be a valuable resource for organelle DNA editing in plants.