Targeted A-to-G base editing in the organellar genomes of Arabidopsis with monomeric programmable deaminases

Abstract Plastids and mitochondria are 2 intracellular organelles containing DNA-encoding partial but essential components for their roles, photosynthesis, and respiration. Precise base editing in both plastid and mitochondrial genomes would benefit their gene functional analysis and crop breeding. Targeted base editing in organellar genomes relies on a protein-based genome-editing system that uses the TALE-DNA recognition motif with deaminases. This is because the efficient delivery of guide RNA for clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 systems into organelles is currently impossible. Since TALE-based base editors used in organellar genomes are usually dimeric types, in this study, we used targeted A-to-G base editing in Arabidopsis (Arabidopsis thaliana) plastid and mitochondrial genomes with monomeric TALE-based deaminase for easier assembling of vectors. As a result, inheritable targeted A-to-G base editing of adenosine triphosphatase subunit 6-2 (atp6-2) in plant mitochondrial genomes and of 16S ribosomal RNA (16S rRNA) in plastid genomes of Arabidopsis was successfully induced by monomeric TALE-based adenine deaminase (AD) without off-target mutations. The monomeric TALE-based adenine deaminases also demonstrated a preference for editing the 8th T on the same strand from the recognition end. Phenotypic analysis showed that A-to-G conversion at 1139A of plastid 16S rRNA conferred substantial spectinomycin resistance in Arabidopsis, but not the other 2 potential-resistant mutations at 1131T and 1137T, predicted from the previous bacterial data. Our study demonstrated the feasibility of monomeric TALE-based ADs in plant organelles and their potential contribution to the functional analyses of plant organelles with easier assembling.


Introduction
In addition to the cell nucleus, there is also a portion of essential genetic information located in the organelles of cytoplasm in plants, namely chloroplasts and mitochondria.The targeted editing of organellar DNA is important for uncovering genuine functions of organellar-encoded genes.The clustered regularly interspaced short palindromic repeats (CRISPR)/Cas system has been widely used by scientists for genome editing from basic science to applications such as biomedicine and agriculture since its introduction (Lander 2016).However, it has yet to have an established application in organellar genome editing since it is still difficult to deliver guide RNA into organelles (Gammage et al. 2018;Schmiderer et al. 2022).On the other hand, the protein-only genomeediting enzymes, transcription activator-like effector nucleases (TALENs), are successfully used for editing mammalian or plant mitochondrial genomes by attaching the organelle targeting signals to the N-terminus of the enzymes (Bacman et al. 2013;Kazama et al. 2019).The cytoplasmic male sterility (CMS)-associated genes (ORF 79 [orf79] and orf125) of CMS varieties of rice (Oryza sativa L.) and rapeseed (Brassica napus L.), respectively, were successfully knocked out by TALENs with mitochondria localization signals (mitoTALENs; Kazama et al. 2019).Such a nuclease-based knock-out strategy in rice and rapeseed led to large deletions (100bp to 5 kb) and ectopic recombination, which altered the genomic structure and then might cause other potential problems.Compared with knocking out by nucleases, targeted base editing, which mainly involves cytosine base editing (CBE; C:G pairs to T:A pairs, C-to-T) and adenine base editing (ABE; A:T pairs to G:C pairs, A-to-G), allows mild and precise genome editing without double-stranded DNA cleavage causing insertions, deletions, and recombinations.CRISPR/Cas9-based CBEs and ABEs were developed in 2016 (Komor et al. 2016;Nishida et al. 2016) and 2017 (Gaudelli et al. 2017).In 2020, CRISPR-free DdCBE, composed of TALE array proteins, a split DddA half, which is an interbacterial toxin that catalyzes the deamination of cytidines, and a uracil glycosylase inhibitor (UGI), was developed and applied to targeted C-to-T base editing in human mitochondrial DNA (Mok et al. 2020).Based on this TALE-split DddA half-UGI motif, CRISPR-free DdCBEs were then applied to the chloroplast (Kang et al. 2021;Li et al. 2021;Nakazato et al. 2021Nakazato et al. , 2023) ) and mitochondrial genomes (Kang et al. 2021;Nakazato et al. 2022) in plants, and an efficient catalysis of C-to-T substitutions in the organelles was reported.In 2022, A-to-G base editing was also reported in mitochondrial DNA from human cells (Cho et al. 2022) and in plant chloroplast DNA (Mok et al. 2022).Plant mitochondrial genomes are known to be AT-rich, and therefore, A-to-G base editing would be more useful, but to our knowledge, targeted A-to-G base editing has yet to be reported in plant mitochondrial genomes.Also, all the tools that successfully implement base editing in plant organelles are based on the dimeric TALE-linked deaminase architecture (Li et al. 2021;Nakazato et al. 2021Nakazato et al. , 2022Nakazato et al. , 2023;;Mok et al. 2022).Monomeric TALE-linked adenine deaminase (AD; it is originally called mTALED, but we call it mTALEAD in this study to distinguish it easily from TALECD, the cytidine deaminase that catalyzes C-to-T) has only been reported to achieve A-to-G base editing in human mitochondrial DNA (Cho et al. 2022).mTALEAD consists of TALE arrays, a deoxyadenosine deaminase variant TadA8e engineered from Escherichia coli TadA that catalyzes A-to-G, and a catalytically inactive cytidine deaminase DddA tox E1347A.The DddA tox E1347A was set to open the double-stranded DNA for easy access and for the activity of TadA8e, which is known to act on single-stranded DNA rather than doublestranded DNA.Monomeric base editors are worth developing in plants because they are half the length of sequences recognized by the dimeric editors but are accurate enough to recognize the individual sites in the small organellar genomes of a few hundred kilobytes.Monomeric-based editors also do not require the simultaneous expression of 2 molecules with multiple or single-complicated dual expression vectors.
To test the feasibility of monomeric TALE-based base editors in plant organelles and to induce A-to-G conversion in plant mitochondrial genomes, we attempted to induce A-to-G conversion in NADH dehydrogenase subunit 7 (nad7) and adenosine triphosphatase (ATPase) subunit 6-2 (atp6-2) in mitochondria and 16S ribosomal RNA (16S rRNA) in the chloroplast of Arabidopsis (Arabidopsis thaliana) using mTALEADs.

Generation of A-to-G conversion in rrn 16S, nad7, and atp6-2
As targets for adenine base editing, we selected 3 sites (1131T, 1137T, and 1139A) in 16S rRNA in the plastid genome, which are predicted to confer spectinomycin resistance (spm r ) from the homologous sequences in the bacterial information (Miyazaki and Kitahara 2018; Supplementary Table S1).To simplify, both conversions of T-to-C and A-to-G are basically described as A-to-G in this study.In the mitochondrial genome, we selected genes nad7 and atp6-2.An isoform of the ATP synthase subunit 6 gene, atp6-2, was chosen as a nonlethal target, without any genetic burden, because of its redundancy of genes in Arabidopsis ecotype Columbia (Arimura et al. 2020).Two vectors were constructed on either side of the target region (nad7-1/2, atp6-2-1/2, and 16S rRNA-1/2) for each target gene (Supplementary Fig. S1).Ten T 1 progenies were initially genotyped for each construct, and among the 6 constructs, only 16S rRNA-1 and atp6-2-2 caused A-to-G conversion in the targeted region (Fig. 1A).For rrn 16S rRNA, no targeted A-to-G conversions were observed from leaves at 11 d after stratification (DAS), but the targeted base 1137T appeared to be homoplasmically (homo) substituted (10th T, which refers to the 10th T away from the TALE binding sequence) in a leaf sample of 1 plant with the vector 16S rRNA-1 at 24 DAS.At the same time, five 11 DAS plants had heteroplasmic and/ or chimerical (h/c; i.e. not homoplasmic) C-to-T conversion, but they were all not detected in any leaves at 24 DAS (Fig. 1,  A and B), suggesting that C-to-T induced by the catalytically inactive cytidine deaminase DddtoxA E1347A (Cho et al. 2022) is not stable.The 3rd sequencing (33 DAS) was performed to see the genotype of this sole homo T 1 mutant 16S rRNA-1-9, and it showed the wild-type base (Fig. 2A), suggesting that the homoplasmic mutation occurred at some parts as a chimera in a plantlet.Compared with 16S rRNA-1, 4 T 1 plants with minor heteroplasmic or chimerical (h/c) A-to-G conversion at 8th T were found in atp6-2-2 at 11 DAS, and 2 of them (atp6-2-2-7 and atp6-2-2-9) also showed the A-to-G conversion in the leaves sampled at 24 DAS (Fig. 1A).Encouraged by this result, more T 1 seeds were sowed and genotyped (Supplementary Fig. S2).This time A-to-G conversion was observed in 4 constructs: 16S rRNA-1, 16S rRNA-2, atp6-2-2, and nad7-1 (Supplementary Fig. S2A), and 13 of 46 T 1 plants showed h/c A-to-G conversion at the 8th T in atp6-2-2 (Supplementary Fig. S2B).Eight plants with nad7-1 showed targeted A-to-G base editing at the 12th T and 17th T, but 7 of them except for nad7-1-22 had the wild-type base at these positions in another leaf sampled at 24 DAS (Supplementary Fig. S3).Collectively, 4 of 6 mTALEAD constructs functioned in plastids and mitochondria, and atp6-2-2 showed stable and the highest editing efficiency at the 8th T of the targeted sequence.

Inheritance of mutations by seed progeny
To test whether the A-to-G mutations induced by mTALEADs can be inherited by seed progeny, we genotyped 62 T 2 plants from a T 1 line 16S rRNA-1-9 and 47 T 2 plants from a T 1 line atp6-2-2-7, with and without transfer DNA (T-DNA) insertion in their nuclei.All of the h/c A-to-G conversions in T 1 plants were segregated to the wild-type base or homo conversion in T 2 progeny for both genes (Fig. 2 10  24DAS   1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19  accounts for 47% for 16S rRNA (Fig. 2A) and 50% for atp6-2 (Fig. 2B).No h/c genotype was observed for both genes in the T 2 generation.These results suggested that the A-to-G conversion in the plastid or mitochondrial genomes in T 1 parents induced by mTALEADs can be inherited and fixed to homoplasmy, independently of the inheritance of the mTALEAD expression cassettes in the nuclei.

Off-target analysis
Off-target effects of mTALEADs on the plastid and mitochondrial genomes were checked by using total DNA from 6 homoplasmically base-edited T 2 null segregants (3 individual plants of each segregant for atp6-2-2-7 and 16S sRNA-1-9) and 3 Col-0 via Illumina next-generation sequencing (NGS; Fig. 3; Supplementary Table S2).No dominant off-target point mutations were detected in all the T 2 lines.mTALEAD targeting 16S rRNA-1 introduced the targeted mutation with 100% allele frequency (AF) in the chloroplast genome of A. thaliana, and similar results were obtained for mTALEAD targeting atp6-2 in the mitochondrial genome.In addition to the target sites, mTALEAD exhibited minor frequent (<10% of the reads) off-target A-to-G mutations/polymorphisms at 5 sites in the mitochondrial genomes.However, they were also detected in the wild-type plants with a similar frequency.Four of the 5 off-target mutations/polymorphisms were likely to be artifacts caused by the nuclear-encoded mitochondrial DNA sequences in chromosome 2 (Naish et al. 2021;Fields et al. 2022;Wang et al. 2022; Supplementary Table S2).All the results indicated that mTALEADs could specifically convert T:A to C:G in the target windows without substantial off-targets in organellar genomes.

spm r test of T2 progenies of 16S rRNA-1
Verification of spm resistance in T 2 progenies of 16S rRNA-1 with a homoplasmic substitution of T1137C was performed.T 2 seeds of 16S rRNA-1-9 were sowed on plates containing 50 mg L −1 of spm.The same number of Col-0 seeds and T 3 seeds of 16S rRNA C1015T spm r mutants (Nakazato et al. 2021) were sowed.Consequently, all the T 2 progenies of 16S rRNA-1-9 showed a spm-sensitive (spm s )-like phenotype like wild-type plants (Supplementary Fig. S4A), even in the lower concentration of spm, 5 and 0.5 mg L −1 (Supplementary Fig. S4, B and C).T1137C in 16S rRNA may confer no spm r or very weak spm r in A. thaliana, which is also consistent with the previous study that reported that the corresponding 1137T site in E. coli 16S rRNA U1189 conferred a modest spm r to E. coli (Miyazaki and Kitahara 2018).

Shifting the TALE arrays to obtain A-to-G conversions at 1131T and 1139A in 16S rRNA
Since the substitution of 16S rRNA_1137T conferred very weak spm r in A. thaliana, we sought to induce A-to-G substitution into 2 other sites in rrn 16S rRNA that are expected to confer stronger spm resistance in A. thaliana: 1131T and 1139A (O'Neill et al. 1993Criswell et al. 2006;Filipenko et al. 2011;Miyazaki and Kitahara 2018).Although we succeeded in mutating the 10th T in the 1st trial in rrn 16S rRNA, it is suggested from the previous data (Fig. 1B; Supplementary Fig. S2B) that the 8th T is the most efficient site for the introduction of A-to-G conversion by mTALEADs.Therefore, to induce A-to-G conversion in 1131T and 1139A efficiently, 4 new mTALEADs were designed by placing 1131T and 1139A at the 8th or 10th positions of T. They are as follows: (i) rrn 16S rRNA_1131T8C, (ii) rrn 16S rRNA_1131T10C, (iii) rrn 16S rRNA_1139T8C, and (iv) rrn 16S rRNA_1139T10C (Fig. 4A).T 1 plants with each of the new constructs were sowed, and their genotype was checked twice (11 and 28 DAS).As shown in Fig. 4B, plenty of A-to-G conversions were found at the 8th T in 16S rRNA_1131T8C.In 16S rRNA_1139T8C, 5 h/c A-to-G conversions at the 8th T were detected out of 53 T 1 plants.We also observed 3 h/c A-to-G conversions at the 10th T at 28 DAS in 16S rRNA_1131T10C.There were no A-to-G conversions detected in 16S rRNA_1139T10C in all the tested 54 T 1 plants (Fig. 4B) In a word, mTALEADs exhibited higher editing efficiency at the 8th T than at the 10th T, and we successfully obtained mutants at 1131T and 1139A of 16S rRNA via the strategy of shifting TALE arrays to place 1131T and 1139A at the 8th T.

spm r test of 16S rRNA_T1131C and 16S rRNA_A1139G
To check whether spm r was conferred by substitution of 1131T and 1139A of 16S rRNA, 20 seeds were sowed for each 16S rRNA mutant (16S rRNA_T1131A, 16S rRNA_A1139G, and 16S rRNA_T1137A) on 1/2 MS medium with or without 10 mg L −1 spm (Fig. 5, A and B).Consequently, all the T 2 seeds of 16S rRNA_1131T8C-22 (Fig. 5, Line 2) showed an obvious spm-sensitive phenotype like 16S rRNA_1137T8C and the wild-type Col-0 plants.On the other hand, all of the tested 20 plants with 16S rRNA_A1139G (Fig. 5, Line 1) exhibited spm r , which is consistent with the previous study that induced A-to-G conversion at 1139A by dimeric TALEADs (Mok et al. 2022).The spm r test indicated that the substitution at 1139A in the 16S rRNA was capable of conferring spm r in A. thaliana, while the substitutions at 1131T and 1137T, which provided 8-fold and 32-fold resistance to E. coli, respectively, conferred little resistance to Arabidopsis (Supplementary Table S1).None of these plants contained mTALEAD genes.The x-and y-axes show the positions and frequencies of the mutated SNPs that differed by at least 5% from the reference genomes BK010421.1 (mitochondria) and AP000423.1 (chloroplast).AF was calculated as AF mu −AF WT , where AF mu is an AF of a SNP of each mutant and AF WT is the average of the AFs of the same SNP of 3 WT plants.

The most efficient base site for inducing A-to-G conversion via mTALEADs
Last but not least, we roughly reassessed editing efficiency at different T (nth T, where n refers to the number of bases in the targeting window away from TALE arrays) and A (nth A) positions in the sense strand, based on the results of base editing in T 1 plants from the constructs of mTALEADs (Supplementary Table S3; Fig. 6).Constructs that did not bring any mutations (even C-to-T) were excluded from the counting.It is notable that in 215 T 1 plants that had T at the 8th base, 58 had homo or h/c A-to-G conversion in the opposite strand, which is the highest ratio (58/215, 27%) among all the T/A sites (Fig. 6, A and B).Fifteen percent of the T 1 plants that had T at the 17th base had A-to-G conversion at the 17th T. It is also unexpected that mTALEADs displayed very little editing effects for As in the sense strand.In 75 and 78 1 plants that had As at the 21st and 23rd bases, respectively, only 3 plants had A-to-G mutation at these As (Fig. 6C).This result suggests that placing the targeted base at the 8th T from the TALE arrays would maximize the editing efficiency of the base in plant organellar DNA.

Discussion
In this study, we succeeded in targeted A-to-G editing in Arabidopsis in both mitochondria and chloroplasts by monomeric TALEADs under the control of the RPS5A promoter via Agrobacterium-mediated transformation.Analyses of T 2 progenies of 16S rRNA and atp6-2 indicated that the induced A-to-G conversion could be stably inherited independently of the inheritance of the mTALEAD expression vector in nuclei.For sequences without T at the 8th base, mTALEADs showed little editing efficiency (Fig. 1B; Supplementary Fig. S2), whereas for sequences with T at the 8th base, 27% of the tested T 1 plants had A-to-G conversions at the 8th T (Fig. 6).However, there is also the case in which mTALEAD showed low editing frequency at the 8th T (e.g.16S rRNA-1139T8C, Fig. 4), which might be caused by the differences of surrounding sequences and/or TALE repeats/protein structure of the enzymes.Such differences should be characterized more in the future, which will contribute to developing more accurate and useful A-to-G base editors.
Concurrent C-to-T and A-to-G edits were observed in 16S rRNA and atp6-2 in this study (Supplementary Fig. S5).In addition, while UGI-free dimeric TALEDs induced A-to-G edits without C-to-T edits in human mitochondria (Cho et al. 2022), they induced C-to-T edits in addition to A-to-G edits in protoplasts and all examined plants (Mok et al. 2022).It is speculated that uracil bases (the intermediates during C-to-T conversion) produced by UGI-free TALEDs in plant organelles are less efficiently removed by endogenous uracil DNA glycosylase than in human mitochondria, as proposed by a previous study (Kim and Chen 2023).The C-to-T edits observed in this study were likely to be caused by the full-length catalytically inactive cytidine deaminase DddA tox E1347A, which was engineered to act and open double-stranded DNA, making it more accessible to the AD TadA8e that operates on single-stranded DNA.Using DddA tox with multiple point mutations at active sites may help suppress C-to-T editing.It is also a future task to develop ABEs that induce only A-to-G in plant organelles.
Many T 2 plants inherited homoplasmic A-to-G mutations from their heteroplasmically or chimerically edited parental T 1 mutants (Fig. 2B).The obtained null segregant T 2 plants will be good materials to reveal the effect of targeted A-to-G base editing on phenotypes.In addition, these null segregant T 2 plants are suitable for practical use because they are exempt from Genetically Modified Organism regulations in some countries.The rapid segregation of homoplasmic base-edited T 2 plants in this study would compensate for, in some cases, the low efficiency of A-to-G observed in the T 1 generation.Such rapid segregation of homoplasmy in T 2 plants is supported by the study showing that heteroplasmic sorting out occurs relatively rapidly, even in 1 to 2 generations for both mitochondrial and plastid genomes (Broz et al. 2022).
Taken together, our study demonstrated the feasibility of mTALEADs in plant organelles.The inclined high efficiency of base editing at the 8th T (Fig. 6) contributes to the design of targeted editing enzymes with high accuracy and suppression of unintentional off-target mutations.This work contributes to the future development and application of monomeric organellar genome editing tools in plant organelles.

Plant materials
Arabidopsis (A. thaliana) Columbia-0 (Col-0) and transgenic plants were grown at 22 °C and under long-day conditions (16 h light, 8 h dark).Col-0 seeds were sown on 1/2 MS medium (pH 5.7) containing 2.3 g L −1 of MS Plant Salt Mixture (Wako), 500 mg L −1 of MES, 10 g L −1 of sucrose, 1 mL L −1 of Plant Preservative Mixture (Plant Cell Technology), 1 mL L −1 of Gamborg's Vitamin Solution (Sigma-Aldrich) and 5 g L −1 of agar.Seedlings at 2 to 3 wk old were transferred to Jiffy-7 (Jiffy Products International) in the greenhouse at 22 °C under long-day conditions and thereafter subjected to Agrobacterium transfection.

Designing the TALE binding sequence and plasmid construction
The TALE targeting sequence was designed to be on one side of the AD targeting window, and we prepared 2 TALEAD vectors for each target gene (Supplementary Fig. S1).The length of TALE-binding sequences was designed between 15 and 20 bp (Supplementary Table S3), so that TALE repeats would specifically recognize and bind to the targeting sequence.ptpTALEADs or mtTALEADs in Ti-plasmids for each target was constructed by using a Platinum Gate assembling kit (Sakuma and Yamamoto 2016) and multisite Gateway (Thermo Fisher) as described in our previous study of mitochondria-targeted TALEN (Kazama et al. 2019).The DNA-binding domains of ptpTALEAD or mtTALEAD were assembled with the Platinum Gate TALEN system based on our previous study (Nakazato et al. 2021).The construct was composed of DNA-recognizing sequence TALE arrays, a deoxyadenosine deaminase variant for catalyzing A-to-G conversion, termed TadA8e, engineered from the E. coli TadA (Richter et al. 2020; shown as AD in the figures) and a catalytically inactive, nontoxic, full-length cytidine deaminase DddA tox variant with an active-site E1347A mutation (Fig. 1A).The N terminus of the mTALEADs was linked to a chloroplast targeting signal peptide of the A. thaliana RecA1 protein (51 aa; Cerutti et al. 1992;Cao et al. 1997) or the mitochondrial targeting peptide of the Arabidopsis ATPase delta prime subunit (Arimura et al. 2020) to introduce the expressed mTALEAD proteins into chloroplast or mitochondria.Each CD half and UGI coding sequence in the previous vectors of ptpTALECDs used for assembly-step2 was replaced in advance by TadA8e and DddA tox E1078K coding sequences.TadA8e and DddA tox E1078K coding sequences were designed to encode same amino acid as Cho's experiment (Cho et al. 2022) and artificially synthesized by Twist Bioscience, with restriction enzyme cutting sites bglII and pstI at the beginning and end of the sequence.Both TadA8e coding sequence and E1 vector pENTR_L1-L4_HD_G1397-DddtoxA-N were subjected to double enzyme digestion with bglII and pstI and ligated by Quick Ligase Enzyme.The reading frame in the assembled entry vectors including the TALE sequence and TadA8e and DddA tox E1078K coding sequences were transferred into the Ti

23 24 A A A A A A A A A A A A A A A A A A A A A
), and in the case of T-DNA (i.e.no T-DNA), the homo conversion 5'……ACCGCCGGTGATAAGCCCCAGGAAGGTGAGGATGACGTCAAGTCATCATGCCCCTTA……3'

Figure 1 .
Figure 1.Genotyping of T 1 plants.A) Numbers of T 1 plants with A-to-G or C-to-T mutation for 6 constructs.The numbers in parentheses represent the number of T 1 plants with homoplasmic substitution.B) Schematic of the mTALEAD construct of 16S rRNA-1 and the number of T 1 plants with bases edited and their positions for the construct 16S rRNA-1 at 11 and 24 DAS.The triangles indicate the special base-editing targets predicted to confer spm r .CTP, chloroplast targeting peptide; TALE arrays: DNA-binding sequence; AD, adenine deaminase TadA8e; DddA tox E1347A, a catalytically inactive cytidine deaminase; homo, homoplasmic substitution.The number of T 1 plants with A-to-G conversion is marked in shadow, and the nonshadow number indicates C-to-T conversion.C) Base editing results in the target of 11 and 24 DAS T 1 plants of the construct atp6-2-2.

Figure 2 .
Figure 2. Genotyping of T 1 s and their progeny T 2 plants for 16S rRNA-1-9 and atp6-2-2-7.A) Number of T 2 plants with wild-type (WT) or homoplasmic (homo) A-to-G conversion at the target 1137T, the 10th T of 16S rRNA-1 with or without T-DNA.B) Number of T 2 plants with WT or homoplasmic A-to-G conversion at the target, the 8th T of atp6-2-2-7 with or without T-DNA.Target sites are bordered.

Figure
FigureOn-target and off-target SNPs in mitochondrial genomes and chloroplast genomes in 6 representative T 2 plants (3 progenies of 2 T 1 lines of atp6-2-2 A) and 16S rRNA-1 B).None of these plants contained mTALEAD genes.The x-and y-axes show the positions and frequencies of the mutated SNPs that differed by at least 5% from the reference genomes BK010421.1 (mitochondria) and AP000423.1 (chloroplast).AF was calculated as AF mu −AF WT , where AF mu is an AF of a SNP of each mutant and AF WT is the average of the AFs of the same SNP of 3 WT plants.
A A G G T G A G G A T G A C G T C A A G T C A T C A T G8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 C T T G A C G T C A T C C T C A C C T T C C T C C G G C T T A T

Figure 4 .
Figure 4. of T 1 plants of 16S rRNA_1131T and 16S rRNA_1139A.A) Shifting of TALE arrays to make the new target bases (1131T and 1139A) at the 8th or 10th T. TALE arrays of 4 new constructs are showed and the 1st tested constructs (16S rRNA-1 and 16S rRNA-2) are marked within the sequence.B) Base editing results in the target windows of 16S rRNA of 11 or 28 DAS T 1 plants for 3 new constructs.The number of T 1 plants with A-to-G conversion is marked in shadow, and the nonshadow number indicates C-to-T conversion.Vacant means 0. homo, homoplasmic substitution.

Figure 6 .
Figure 6.The percentage of T 1 plants with C or G substitution in the different positions of T or A induced by mTALEADs in this study.A) The percentage of T 1 plants with substitution in the nth T, summarized from B). B) The number of total detected T 1 plants and T 1 plants with C substitution in different T positions, accumulated from all the results from all constructs examined.C) The number of total detected T 1 plants and T 1 plants with G substitution in different A positions.