Genome editing in plants using the compact editor CasΦ

Significance Plant genome engineering with Clustered regularly interspaced short palindromic repeats and CRISPR-associated proteins (CRISPR-Cas) systems is frequently used in both research and agriculture. Here, we demonstrate that the hypercompact CasΦ-2 nuclease is able to generate heritable gene edits in Arabidopsis. Two CasΦ protein variants vCasΦ and nCasΦ increased the editing efficiency in plants and yielded more offspring plants with heritable edits. CasΦ also has a wide range of working temperatures, and the editing by CasΦ is highly specific. We also observed that editing by CasΦ is sensitive to chromatin environment. The hypercompact size, T-rich minimal protospacer adjacent motif (PAM) and wide range of working temperatures make CasΦ an excellent supplement to existing plant genome editing systems.

observed, suggesting that these variants can be utilized for highly specific genome editing in plants.

CasΦ-2 Is Capable of Generating Heritable Mutations
in Arabidopsis. Previously, it was shown that CasΦ-2 ribonucleoproteins (RNPs) were capable of editing the AtPDS3 gene in Arabidopsis mesophyll protoplasts (19). To investigate if CasΦ-2 is able to edit a target gene in transgenic plants, we used the Arabidopsis UBQ10 promoter to drive the expression of an Arabidopsis codon optimized CasΦ-2 and the U6 promoter to drive transcription of guide RNAs (Fig. 1A). The ubiquitous expression by the UBQ10 promoter allows for assessment of editing efficiency in somatic tissues of transgenic plants as well as using mesophyll protoplasts for further optimization. We utilized AtPDS3 gRNA10, which had the highest editing efficiency of all AtPDS3 gRNAs tested in protoplasts (19). Version 1 and version 2 constructs were made with nuclear localization signal peptides either flanking N and C termini or only at the C terminus of the CasΦ-2 protein, respectively (Fig. 1A). Seventy-eight T1 transgenic plants of the version 1 construct were screened by Sanger sequencing, and a plant heterozygous for mutation of the gRNA10 targeted region was identified (T1 #33) (Fig. 1B). Further amplicon sequencing of different parts of this plant indicated that it was mosaic for the heterozygous mutation, with leaf a and leaf b showing about 50% of reads carrying mutation, but other leaves showing little to no editing (SI Appendix, Table S1). The dominant form of mutation detected in this plant by amplicon sequencing was a 6 bp deletion in the AtPDS3 gR10 targeted region, although a small number of reads with other forms of deletion were also detected (SI Appendix, Table S1). The AtPDS3 gene encodes a phytoene desaturase enzyme that is essential for chloroplast development. Disruption of this gene function results in albino and dwarfed seedlings (23). Indeed, albino and dwarfed seedlings were observed from the offspring (T2) population of plant #33 (Fig. 1C). Twenty albino/dwarf seedlings were tested individually for the AtPDS3 gR10 target region, and all of the tested seedlings were homozygous for a 6 bp deletion at the gR10 target region (Fig. 1D). This 6 bp deletion leads to the loss of two amino acids which are highly conserved among orthologous proteins (SI Appendix, Fig. S1A). In addition, two of these 20 albino/dwarf T2 seedlings had segregated away the transgene and were thus CasΦ-2 transgene-free (Fig. 1E), confirming the germline transmission (heritability) of the CasΦ-2 generated mutation in the AtPDS3 gene from the T1 to the T2 plants.
For the version 2 construct, six T1 transgenic plants were screened, this time directly by amplicon sequencing so that we could more sensitively detect plants with low editing efficiency. T1 plant #6 showed the highest number of edits, with 13.5% of reads carrying mutations, while the other five plants showed much lower or no editing (SI Appendix, Fig. S1B). Ninety-six offspring T2 plants of T1 plant #6 were analyzed, and six of them were heterozygous for mutations at the AtPDS3 gR10 targeted region (SI Appendix, Fig. S1C). As expected, albino seedlings were identified in the T3 offspring populations of these heterozygous T2 plants (SI Appendix, Fig. S1D). DNA sequencing of T3 albino seedlings derived from different T2 lineages showed deletions of 3 bp, 5 bp, 9 bp (two plants), or a 42 bp deletion plus a 1 bp deletion that was 16 bp upstream of the larger deletion (two plants) (SI Appendix, Fig. S2A). CasΦ-2 transgene-free albino seedlings were also identified from these T3 populations, further supporting the heritability of the  CasΦ-2 generated mutation in the AtPDS3 gene (SI Appendix, Fig. S2B). These results suggest that the initial T1 plant #6 was chimeric for different deletions which were inherited into the different T2 and T3 lines.

RDR6 Mediated Transgene Silencing Attenuates CasΦ-2
Mediated Editing in Transgenic Plants. In Arabidopsis, RNAdependent RNA polymerase 6 (RDR6) plays an important role in the initiation of transgene silencing (24,25).  (19). Altogether, these data suggest that the CasΦ-2 protein has a wide functional temperature range, which should make it useful for editing in a wide variety of eukaryotic systems.

CasΦ-2 Mediated Editing Is Sensitive to DNA Methylation Status.
Previously, it was observed that DNA methylated chromatin showed a lower efficiency of editing by Cas9 (28,29). To examine this for CasΦ-2 editing, we took advantage of the FWA gene and the fwa-4 epiallele. In WT Arabidopsis plants, the FWA gene is silent in all adult plant tissues owing to DNA methylation in the promoter. FWA is normally only expressed by the maternal allele in the developing endosperm where it is imprinted and demethylated (30). In the epiallele fwa-4, the promoter is heritably unmethylated, and thus, the FWA gene is expressed ectopically (31). Ten guide RNAs were designed targeting the promoter region of the FWA gene, immediately upstream of the transcription start site ( Fig. 2A). In WT plants, this region contains DNA methylation and is targeted by the RNA directed DNA methylation machinery (32,33), as evidenced by Pol V occupancy ( Fig. 2A). Assay for Transposase-Accessible Chromatin using sequencing (ATAC-seq) signals are also significantly lower in the WT plant compared to the fwa-4 epi-mutant (34) ( Fig. 2A), suggesting that the accessibility of the FWA promoter region is lower in the WT plant. An in vitro cleavage assay with CasΦ-2 RNPs showed that all 10 FWA guide RNAs led to cleavage of the PCR amplified FWA gene fragment, with the majority of the substrate DNA cleaved by RNPs of gRNA1, gRNA4, gRNA5, gRNA6, and gRNA7 (SI Appendix, Fig. S4). When CasΦ-2 RNPs were transfected into fwa-4 epi-mutant protoplasts, gene editing events were detected with gRNA1, gRNA4, gRNA5, and gRNA6 (Fig. 2B). To compare the editing efficiency at FWA in different chromatin states, in an independent experiment, protoplasts of WT and fwa-4 epi-mutant plants were prepared and transfected with CasΦ-2 RNPs with FWA gRNA1, gRNA4, gRNA5, and gRNA6. Much higher editing efficiency was observed for each of the four gRNAs in the fwa-4 protoplasts compared to the WT protoplasts ( Fig. 2C), suggesting that the CasΦ-2 mediated editing is more efficient when the target DNA has a more open chromatin state compared to a repressive chromatin state. pnas.org

CasΦ-2 Guide RNAs Can Be Expressed with Pol-II Promoters and
Ribozyme Processing Machineries. As Pol-II promoters have been successfully used to drive the gRNA transcription for the Cas9 and Cas12 systems (14,35), we wanted to test if Pol-II promoters are able to drive CasΦ-2 guide RNA transcription. Three Pol-II promoter and terminator sets were tested, the CmYLCV promoter and 35S terminator, a 2×35S promoter and HSP18.2 terminator, and a UBQ10 promoter with a rbcS-E9 terminator (Fig. 3A). Since CasΦ-2 has intrinsic pre-crRNA processing activity (19), AtPDS3 gRNA10 without additional RNA processing machinery in three configurations was tested, a single CasΦ-2 repeat followed by AtPDS3 gRNA10 spacer, a CasΦ-2 repeat followed by AtPDS3 gRNA10 spacer followed by a second CasΦ-2 repeat, and a triple array of CasΦ-2 repeats/AtPDS3 gRNA10 spacers followed by a fourth CasΦ-2 repeat (Fig. 3A). Among the three combinations of Pol-II promoters and terminators, the CmYLCV promoter with the 35S terminator led to the highest editing efficiency, while the UBQ10 promoter with the rbCS-E9 terminator led to the lowest editing efficiency in protoplasts (Fig. 3B). Out of the three different gRNA configurations, the single CasΦ-2 repeat followed by the AtPDS3 gRNA10 exhibited the highest editing efficiency, while the CasΦ-2 repeat followed by the AtPDS3 gRNA10 with another CasΦ-2 repeat at the end exhibited the lowest editing efficiency (Fig. 3B). When combining the CmYLCV promoter/35S terminator with the single CasΦ-2 repeat followed by the AtPDS3 gRNA10, the target gene editing efficiency was much higher than that of the AtU6-26 AtPDS3 gRNA10 cassette in protoplasts (Fig. 3B). Consistent with the higher levels of editing observed, a higher level of AtPDS3 gRNA10 was detected in protoplasts transfected with plasmid carrying the cassette with the CmYLCV promoter AtPDS3 gRNA10 construct (SI Appendix, Fig. S5). This data suggests that boosting the levels of gRNAs using a strong promoter can increase the efficiency of gene editing by CasΦ-2.
The fact that the single AtPDS3 gRNA10 without another CasΦ-2 repeat at the end exhibited the highest editing efficiency among the three gRNA configurations suggests that the CasΦ-2 processing of the transcript was not able to provide enough mature gRNAs. The 20 bp spacer could be too short for CasΦ-2 to bind simultaneously to both of the repeats flanking the spacer for pre-cr-RNA processing. To examine this further, AtPDS3 gRNA10 with a 30 bp spacer was used to test if a longer spacer might assist the self-processing of pre-crRNA by CasΦ-2. However, we observed no difference between the editing efficiencies of the 30 bp and the 20 bp AtPDS3 gRNA10 spacers when constructed in the A form (Fig. 3C). We also observe a slight decrease of target gene editing when the 30 bp spacer was constructed in the C form compared to the 20 bp spacer (Fig. 3C). These results suggest that a longer 30 bp spacer did not promote more efficient processing of pre-crRNA by CasΦ-2.
We tested whether adding a secondary gRNA processing mechanism might facilitate the release of mature gRNA. A ribozyme gRNA processing system, previously proven to be able to facilitate Cas9 and Cas12a guide RNA processing (14,36), was cloned to flank the CasΦ-2 repeat and the AtPDS3 gRNA10 spacer sequence. The hammerhead (HH) type ribozyme was added to the 5′ end of the CasΦ-2 AtPDS3 gRNA10, and a hepatitis delta virus (HDV) ribozyme was added to the 3′ end (Fig. 3D). Constructs with ribozymes led to significantly higher editing efficiency, with all three promoter-terminator combinations tested (Fig. 3D). These results suggest that ribozymes were able to promote the processing of gRNA from the Pol-II transcripts, leading to a higher editing efficiency.

The Engineered CasΦ-2 Variants vCasΦ and nCasΦ Have Higher
Target Gene Editing Efficiency in Protoplasts. It has been previously reported that the engineered CasΦ-2 variants vCasΦ and nCasΦ cleaved substrate DNA faster than the WTCasΦ in vitro (22). To test if these engineered CasΦ-2 variants are able to edit a target gene with higher efficiency in Arabidopsis, mesophyll protoplasts were transfected with plasmids expressing WTCasΦ, vCasΦ, or nCasΦ, as well as the desired guide RNAs. Version 2 constructs with U6 promoter-driven AtPDS3 gRNA8 (19) and gRNA10 were transfected into WT protoplasts, while version 2 constructs with U6 promoter-driven FWA gRNA1, gRNA4, gRNA5, and gRNA6 were transfected into fwa-4 protoplasts.
Higher editing efficiencies were observed with vCasΦ and nCasΦ variants for all guide RNAs tested (Fig. 4A). To statistically evaluate the differences between the editing efficiencies, for the six gRNAs tested, normalized editing efficiencies were calculated (ratio over WTCasΦ efficiency) and pooled for statistical tests. Both the vCasΦ and nCasΦ variants yielded significantly higher editing efficiencies than the WTCasΦ, with ~17-fold of increase in editing efficiency for vCasΦ and ~ten-fold of increase in editing efficiency for nCasΦ (Fig. 4B). Similar tests were performed with RNPs reconstituted with WTCasΦ, vCasΦ, or nCasΦ and the same guide RNAs (SI Appendix, Fig. S6A). A significant increase in the editing efficiency was detected with both the vCasΦ and nCasΦ variants when pooled analysis was performed (SI Appendix, Fig.  S6B), although the overall fold increase in editing efficiency was lower compared to the plasmid transfection experiments ( Fig. 4B and SI Appendix, Fig. S6B).
We also tested if the vCasΦ and nCasΦ variants were able to enhance the editing efficiency at more compact chromatin utilizing the FWA promoter region as the target. To test this, protoplasts from WT plants were used for plasmid transfections with FWA gRNA1, gRNA4, gRNA5, and gRNA6. Although zero to very low editing was observed for WTCasΦ, both vCasΦ and nCasΦ variants yielded readily detectable and higher editing frequencies for all four FWA guide RNAs tested (Fig. 4C). Thus, in compact chromatin environments, the vCasΦ and nCasΦ variants appear to dramatically enhance the editing efficiency compared to WTCasΦ.
To confirm that the vCasΦ and nCasΦ variants, like WTCasΦ, are compatible with guide RNAs driven by Pol-II promoters, the CmYLCV promoter and the UBQ10 promoter were used to drive transcription of AtPDS3 gRNA10 flanked by ribozymes. For both the CmYLCV promoter and the UBQ10 promoter, the editing efficiencies of the vCasΦ and nCasΦ variants were higher than that of the WTCasΦ (Fig. 4D).
Despite the differences in editing efficiency, the editing profiles of WTCasΦ, vCasΦ, and nCasΦ were similar. Consistent with previous data for WTCasΦ (19), the vCasΦ and nCasΦ variants also generate deletions primarily around 8 to 10 bp in size (SI Appendix, Fig. S6C).

vCasΦ and nCasΦ Show Higher Editing Efficiency in Transgenic
Plants. Like in protoplasts, we observed significantly higher editing efficiencies in T1 transgenic plants expressing the vCasΦ and nCasΦ variants compared to WTCasΦ for AtPDS3 gRNA10 driven by the U6 promoter (Fig. 5A)  with white sectors were observed from the vCasΦ and nCasΦ expressing T1 populations (Fig. 5B), indicating strong editing activity in somatic cells. These white sectors were not observed in transgenic T1 plants expressing WT CasΦ (Fig. 5B). High target gene editing efficiencies were also observed in the T1 plants with vCasΦ and nCasΦ combined with the CmYLCV promoter or UBQ10 promoter-driven AtPDS3 gRNA10 flanked by ribozymes (Fig. 5C). Consistent with the higher editing efficiency in the T1 generation, vCasΦ and nCasΦ with AtPDS3 gRNA10 also generated more albino seedlings in T2 populations compared to the WTCasΦ (Fig. 5D and SI Appendix, Fig. S7A and Table S2). Specifically, only one out of 31 T2 populations of WTCasΦ had albino seedlings at a percentage of 0.32% of total seedlings, while six out of 30 vCasΦ T2 populations and 14 out of 29 nCasΦ T2 populations showed albino seedlings, with highest percentage of albino seedlings at 6.12% and 6.07%, respectively (SI Appendix,   , Fig. S7B).

Off-Target Editing by the Engineered CasΦ-2 Variants vCasΦ and
nCasΦ Is Rare. To evaluate in vivo off-target editing frequencies by CasΦ-2 in Arabidopsis, we performed whole genome sequencing on transgene-free T2 albino seedlings from T1 transgenic plants expressing vCasΦ or nCasΦ U6::AtPDS3 gRNA10 in the rdr6-15 mutant background. In T1 transgenic plants with ubiquitous CasΦ expression, unless off-target editing happens at a significant level in sequence contexts similar to the on-target loci, it is difficult to distinguish off-target editing effects from sequence variants generated from spontaneous mutation, PCR amplification errors, or sequencing errors. In these transgene-free T2 albino seedlings, inherited offtarget edits will exist as heterozygous or homozygous mutations, thus making their detection robust by deep sequencing analysis at the whole genome level. Genomic DNA of albino transgene-free T2 seedlings from three independent T2 populations of vCasΦ and nCasΦ U6::AtPDS3 gRNA10 in rdr6-15, together with four rdr6-15 control seedlings, was sequenced to between 116-fold and 613fold genome coverage, with >99% of genome covered by mapped reads (Fig. 6A and SI Appendix, Table S3). Variations relative to the reference Col-0 TAIR10 genome sequence were detected by The Genome Analysis Toolkit (GATK) and Strelka2 (37,38), and similar to previous observations (39)(40)(41)(42)(43)(44), a large number of single nucleotide polymorphisms (SNPs) and additions or deletions of nucleotides (indels) were detected (SI Appendix, Table S3). However, the number of variants detected in the rdr6-15 background plants relative to the Col reference genome was similar to that in the CasΦ-2 edited plants, suggesting that most of the variants detected are variations which already existed in the rdr6-15 mutant background. We therefore filtered out all variants in the CasΦ-2 edited plants that were also present in the rdr6-15 background. In addition, to select for heterozygous or homozygous mutations, variants with ratios of reference allele reads over variant allele reads larger than 3 (Ref/Alt>3) were discarded (Fig. 6A). Between 66 and 203 variants remained within the six albino seedlings tested after these filtering steps (SI Appendix, Table S3). On-target site mutations (AtPDS3 gRNA10 target region) were reliably detected with this pipeline for all six albino seedlings, with two representative plants shown in Fig. 6   (SI Appendix, Table S3). An example of detailed reads at a predicted off-target site is shown in Fig. 6 B, Right panel and SI Appendix, Fig.  S8 A, Right panel.
We also manually inspected each of the sequence variants detected genome wide, which indicated that most of them are located within long stretches of repeated single or di-nucleotides, and reads with similar variants were readily detected in the rdr6-15 plants, suggesting that most of the detected variants are likely due to imprecise PCR amplification, sequencing reaction or mapping outcomes. After removing these sequence variants at simple repeats, between 6 and 18 high-confidence variants remained in the six albino plants, with more SNP than indels (SI Appendix, Fig. S8B). However, there was no overlap of the variant loci except for a single SNP between two albino seedlings (SI Appendix, Table  S4). These data suggest that the high-confidence variants we detected are rare and likely due to spontaneous mutation events, meaning that CasΦ-2 shows little or no off-target activity.

CasΦ-2 Is Able to Perform Gene Editing in Maize Protoplasts.
To test gene editing by CasΦ-2 in maize, an agronomically important monocot species with a more complex genome than Arabidopsis, 11 guide RNAs were designed targeting the maize Lc gene (SI Appendix, Table S5). RNPs of WTCasΦ, vCasΦ, and nCasΦ with Lc gRNAs were transfected into maize protoplasts. By amplicon sequencing, editing of target loci by Lc gRNA3 to gRNA7 was detected (SI Appendix, Fig. S9). Consistent with results in Arabidopsis, vCasΦ and nCasΦ yielded much higher target editing efficiency compared to the WTCasΦ, with up to about 7% editing for the best guide RNA. These data suggest that CasΦ systems might be useful in many different plant species.

Discussion
In this study, we explored the use of CasΦ-2 for plant genome engineering. We showed that heritable gene edits could be generated in Arabidopsis using WTCasΦ at a low frequency. We also showed that the vCasΦ and nCasΦ protein variants exhibited much higher editing efficiency than WTCasΦ in Arabidopsis and maize protoplasts and also a higher frequency of heritable edits in transgenic Arabidopsis plants. Given that CasΦ-2 was able to edit the dicot plant Arabidopsis, the monocot plant maize, as well as mammalian cells (19), it is likely that CasΦ-2 could be utilized in most eukaryotic systems. The wide temperature range of enzymatic activity of CasΦ-2 may also make it useful for plants and other organisms that grow at cooler temperatures. Since Pol-II promoters can be used for CasΦ-2 guide RNA transcription, the vast array of Pol-II promoters could be useful for expression of gRNAs in specific conditions, such as in different tissues, at higher levels with strong promoters, or in an inducible fashion. CasΦ-2-based editing can also be applied in both RNP and plasmid expression formats. We also found that CasΦ-2 editing was highly specific, with no convincing off-target editing detected by genome wide analysis.
We observed that transgene silencing limited the editing efficiency of CasΦ-2, because editing frequencies were higher when utilizing the rdr6 mutant. This suggests that the level of expression of either the CasΦ-2, its gRNA precursors, or both, is limiting for high efficiency editing. This also suggests that tuning down the transgene silencing machinery and improvements in transgene designs are likely to lead to more efficient editing. By comparing editing efficiencies at the FWA locus in protoplasts derived from the WT methylated strain or the fwa unmethylated epiallele strain, we also found that methylated compact chromatin negatively affected editing by CasΦ-2, presumably because the more highly compact chromatin limits access of CasΦ-2 to the DNA. Fusion of CasΦ-2 protein to chromatin remodelers, as previously reported for the Cas9 protein (46), or other methods, could facilitate successfully editing at difficult target sites, or might increase editing efficiency at all sites.
We observed that more than half of the guide RNAs we designed yielded no detectable editing events by amplicon sequencing, and only a few of the guide RNAs lead to efficient editing. Similarly, strong differences in editing efficiencies between different guide RNAs were observed when CasΦ-2 was used to target an enhanced green fluorescent protein (EGFP) gene in HEK293 cells (19). In addition, variable activities were observed with in vitro substrate DNA digestion assays with CasΦ-2 RNPs reconstituted with different guide RNAs (SI Appendix, Fig. S4). However, with the best performing guide RNAs, reasonable editing efficiencies by CasΦ-2 and its variants were observed. In mammalian cells, more than 30% of cells lost EGFP signal due to the editing of the WTCasΦ-2 with gRNA8 (19). In Arabidopsis, with the CasΦ-2 protein variants and AtPDS3 gRNA10, 10% to 30% editing efficiency was observed in protoplasts (Fig. 4D) and 40 to 80% of somatic editing efficiency was observed in multiple T1 transgenic plants (as indicated by the upper quartile in Fig. 5C). However, in the future it will be important to systematically define gRNA design principles to increase these frequencies.
Amplicon sequencing and in vivo editing results indicated that CasΦ-2 edits consist of small deletions of roughly nine base pairs, consistent with biochemical data showing that CasΦ-2 produces an 8 to 12 nucleotide staggered cut (19). CasΦ-2 could therefore be useful for making in-frame deletions to create partial loss of function alleles, or to remove small sections of specific domains of proteins. Indeed, one of the heritable edits found in this study was a two amino acid deletion resulting in an in-frame deletion of the coding region of the AtPDS3 gene. The ability to make small deletions may also be useful for promoter analysis by removing individual transcription factor binding sites, or by promoter deletion scanning by tiling gRNAs across a promoter.
In summary, this study demonstrates that the compact CasΦ editing system can be used for gene editing in plants and can generate heritable mutations with high specificity. Compared to the Cas9 and Cas12a systems, which has been optimized with community efforts for years, CasΦ-2 is still less efficient in overall editing efficiency. However, with well-performing guide RNAs, a reasonable level of editing efficiency can be obtained with the CasΦ system. In addition, the vCasΦ and nCasΦ variants demonstrated the effectiveness of protein engineering to potentially improve the editing efficiency further in the future. With its hypercompact protein size, wide working temperature range, and T-rich minimal PAM, CasΦ should be a useful supplement to the plant gene editing toolbox and should be useful in both basic research and agricultural biotechnology.

Plant Materials and Growth Condition.
To grow Arabidopsis plants for protoplast preparation, Col-0 and fwa-4 epi-mutant plants were grown under a 12-h light/12-h dark photoperiod and with low light condition for 3 to 4 wk. To grow maize plants for protoplast isolation, B73 seeds were soaked in water overnight and planted in half peat moss and half vermiculite. After 3 d of light incubation, emerged seedlings were transferred to dark until the second leaf was 10 to 15 cm long.
Agrobacterium mediated Arabidopsis plant transformation was performed as previously described (47), and transgenic T1 plants were screened with half MS plates with 40 μg/mL hygromycin B. For 28 °C treatment of T1 plants, including the T1 plants in SI Appendix, Fig. S1B (version2 construct AtPDS3 gRNA10), stratified seeds on half MS plates with hygromycin B were incubated at 28 °C and resistant T1 plants were transferred to soil after about a week and put back to 28 °C for a total of 2 wk incubation at 28 °C. T1 plants were then moved to a greenhouse (23 °C) for the rest of the life cycle. To support the growth of albino seedlings in T2 generation, 3% sucrose was supplemented to half MS plates when needed.
Protein Purification. WTCasΦ, vCasΦ, and nCasΦ proteins with nuclear localization signal (NLS) were purified as previously described (19,22). RNP Reconstitution. Guide RNAs were synthesized as 25 nucleotide repeat + 20 nucleotide spacer as shown in SI Appendix, Table S5. Lyophilized RNA was dissolved by adding DEPC-treated H 2 O to a concentration of 0.5 mM. The dissolved RNA was incubated at 65 °C for 3 min, then cooled down to room temperature. For RNP reconstitution, heated and cooled RNA was added to 2×Cleavage Buffer (2×CB buffer, 20 mM Hepes-Na, 300 mM KCl, 10 mM MgCl 2 , 20% glycerol, 1 mM TCEP, pH 7.5) to a final concentration of 5 µM and vortexed to mix. Then, WTCasΦ, vCasΦ, or nCasΦ proteins were added to a final concentration of 4 µM and mixed by pipetting. This solution was then incubated at room temperature for 30 min. The resulting solution contains 4 µM of RNP in 2×CB buffer.
Plasmids Used in this Study. Plasmids generated in this study are listed in SI Appendix, Table S6. The CASΦ-2×SV40NLS-2×FLAG coding sequence (without IV2 intron) was Arabidopsis codon optimized and synthesized by IDT (Integrated DNA Technologies). The HBT-pcoCAS9 vector (addgene52254) backbone with the FLAG and SV40 NLS (for version1) or without the FLAG and SV40 NLS (for version 2) was amplified and assembled by TAKARA in-fusion HD cloning kit (cat639650) with the synthesized CasΦ sequence amplified as two fragments as well as the IV2 intron sequence amplified from the HBT-pcoCAS9 vector. pCAMBIA1300-pYAO-cas9-MCS vector (48) was digested by KpnI and EcoRI to remove the pYAO-cas9 cassette. Then, pcoCasΦ with IV2 intron, NLS, and FLAG tag coding sequence was amplified from HBT_pcoCASphi_version 1 and version 2 plasmids and cloned into the digested vector together with PCR amplified pUB10 and rbcs-E9 terminator by TAKARA in-fusion HD cloning kit (cat639650). Subsequently, various guide RNA cassettes were cloned into the SpeI site of the obtained vectors by TAKARA in-fusion reaction or by restriction digestion and ligation of PCR amplified or IDT synthesized DNA fragments. Construction of the nCasΦ and vCasΦ binary vectors was performed by first digesting the pC1300_pUB10_pcoCASphi_E9t_MCS_version2 vector with KpnI to remove pUB10 promoter and the first part of CasΦ coding sequence. Then, pUB10 promoter and CasΦ sequences containing the desired mutations were PCR amplified as two fragments with desired mutations added by overlapping primers and cloned back into the digested vector by an in-fusion reaction. Primers used for cloning and synthesized DNA fragments used for cloning are listed in SI Appendix ,  Table S7. The control plasmid HBT-sGFP for protoplast transfection was previously described (49) and obtained from ABRC (stock CD3-911).
Protoplast Isolation and Transfection. Arabidopsis mesophyll protoplast isolation was performed as previously described (49). For RNP transfection into Arabidopsis protoplasts, 26 µL of 4 µM RNP was first added to a round bottom 2 mL tube, followed by 200 µL of protoplasts (2 × 10 5 cells/mL). Then, 2 µL of 5 µg/µL salmon sperm DNA was added and mixed gently by tapping the tube 3 to 4 times. Finally, 228 µL of fresh, sterile, and RNase free PEG-CaCl 2 solution (49) was added to the protoplast-plasmid mixture and mixed well by gently tapping the tube. The protoplasts with PEG solution were incubated at room temperature for 10 min, then 880 µL of W5 solution (49) was added and mixed with the protoplasts by inverting the tube two to three times to stop the transfection. Protoplasts were harvested by centrifuging the tubes at 100 rcf for 2 min and resuspended in 1 mL of WI solution. They were then plated in 6-well plates pre-coated with 5% calf serum. These 6-well plates were then incubated at room temperature for 48 h.
For plasmid transfections into Arabidopsis protoplasts, the concentrations of plasmids were determined by nanodrop. Then 20 to 50 µg of plasmids (plasmid amounts were the same within each experiment) were added to the bottom of each transfection tube, and the volume of the plasmids was supplemented with H 2 O to reach 20 µL. 200 µL of protoplasts were added followed by 220 µL of fresh and sterile polyethylene glycol (PEG)-CaCl 2 solution. The mixture was mixed well by gently tapping tubes and incubated at room temperature for 10 min. 880 µL of W5 solution was added and mixed with the protoplasts by inverting the tube two to three times to stop the transfection. Protoplasts were harvested by centrifuging the tubes at 100 rcf for 2 min and resuspended in 1 mL of WI solution. They were then plated in 6-well plates pre-coated with 5% calf serum. These 6-well plates were then incubated at room temperature or at 28 °C for 48 h.
For maize protoplast isolation, middle 8 to 10 cm of the second leaf from etiolated maize seedlings were sliced into thin strips (about 0.5 mm wide) and immersed in enzyme solution (0.6 M mannitol, 10 mM MES pH 5.7, 1.5% cellulase R10, 0.3% macerozyme R10, 1 mM CaCl 2 , 5 mM BME, 0.1% BSA). Then the enzyme solution with leaf stripes was put under vacuum for half an hour and then placed in dark for 2.5 h. The solution was then gently swirled to release the protoplasts and filtered through a 40 µm cell strainer. The cells were collected by centrifugation at 150 g for 5 min and resuspended in 10 mL washing solution (0.6 M mannitol, 4 mM MES pH 5.7, 20 mM KCl). Then the cells were rested on ice for at least half an hour. After resting, cell pellets were resuspended with washing solution to 2 × 10 5 cells/mL in cold washing buffer. For RNP transfection into maize protoplasts, 13 µL 4 µM RNP and 1 µL 5 µg/µL salmon sperm DNA were added to 100 µL protoplasts. Then 114 µL PEG solution (40% PEG 4000, 0.2 M mannitol, 0.1 M CaCl 2 ) was added and mixed by tapping. After 10 min of incubation, transfections were stopped by adding 880 µL washing solution and gently inverting several times. Transfected protoplasts were collected by 3 min of centrifugation at 150 g and resuspended in 1 mL incubation buffer (50). Then, the protoplasts were plated in 6-well plates pre-coated with 5% calf serum. These 6-well plates were then incubated at room temperature for 48 h. Amplicon Sequencing. DNA was extracted from protoplast samples and leaves of transgenic plants with Qiagen DNeasy plant mini kit (Qiagen 69106). The amplicon was obtained using two rounds of PCR. Amplification primers for the first round of PCR were designed to have the 3′ sequence of the primers flanking a 200 to 300 bp fragment of the genomic area targeted by the guide RNA of interest. Primers for the first round of amplification are listed in SI Appendix , Table S7. The 5′ part of the primer contained a sequence which will be bound by common sequencing primers. After 25 cycles of the first round of PCR amplification, the reaction was purified using 1× Ampure XP beads (Beckman Coulter A63881). The eluate was used as template for the second round of PCR amplification of 12 cycles. The second round of PCR was designed so that indexes were added to each sample. The samples were then purified using 0.8× Ampure XP beads. Part of the purified libraries were run on a 2% agarose gel to check for size and absence of primer dimer (fragments below 200 bp considered as primer dimer). Then amplicons were sent for next generation sequencing.
Amplicon Sequencing Result Analysis. Reads were first quality and adaptor trimmed using Trim Galore and then mapped to the target genomic region by the BWA aligner (v0.7.17, BWA-MEM algorithm). Sorted and indexed bam files were used as input files for further analysis by the CrispRvariants R package (v1.14.0). Each mutation pattern with corresponding read counts was exported by the CrispRvariants R package. After assessing all control samples, a criterion to classify reads as edited reads was established: Only reads with a >= 3 bp deletion or insertion (indel, mainly as deletions) of the same pattern (indels of same size starting at the same location) with >=100 read counts from a sample were counted as edited reads. This criterion is established due to the observation of 1 bp indels and occasionally 2 bp indels with read numbers >100 in control samples. Also, larger indels that occur at very low frequencies (much lower than 100 reads) were observed in control samples. These observations indicate that occasional PCR inaccuracy and low-quality sequencing in a small fraction of reads can result in the indel patterns with corresponding read number ranges as stated above in control samples with the typical sequencing depth in our experiments (1 to 5 million reads/sample). By employing such stringent criteria, it is believed that the editing signals counted are true signals indicating editing events. Additionally, for FWA gRNA5 and gRNA6 targeted regions, there are long stretches of adenines near these target regions. Due to the high error rate of polymerases amplifying long stretches of adenines, reads with indels only within these stretches of adenines were not counted as true deletions. For amplicon sequencing result analysis of maize protoplast, the criterion to classify reads as edited reads was established: Only reads with a >= 3bp deletion or insertion (indel, mainly as deletions) of the same pattern (indels of same size starting at the same location) with >=10 read counts from a sample were counted as edited reads. The adjustment of the criterion was based on the fact that this amplicon sequencing was of less depth (performed by the iSeq 100) compared to the amplicon libraries of Arabidopsis protoplasts.
Off-Target Analysis. DNA from single Arabidopsis seedlings was extracted with the Qiagen DNeasy plant mini kit and sheared to 300 bp size with a Covaris. Library