Comparison of efficiency and specificity of CRISPR-associated (Cas) nucleases in plants: An expanded toolkit for precision genome engineering
Comparison of efficiency and specificity of CRISPR-associated (Cas) nucleases in plants: An expanded toolkit for precision genome engineering
Oleg Raitskin 0 1 2
Christian SchudomaID 0 1 2
Anthony West? 0 1 2
Nicola J. PatronID 0 1 2
0 Editor: Muthamilarasan Mehanathan, ICAR- National Research Centre on Plant Biotechnology , INDIA
1 Current address: Isogenica Ltd. The Mansion , Chesterford Research Park, Little Chesterford, Saffron Walden , United Kingdom
2 Earlham Institute , Norwich Research Park, Norwich, Norfolk , United Kingdom
Molecular tools adapted from bacterial CRISPR (Clustered Regulatory Interspaced Short Palindromic Repeats) systems for adaptive immunity have become widely used for plant genome engineering, both to investigate gene functions and to engineer desirable traits. A number of different Cas (CRISPR-associated) nucleases are now used but, as most studies performed to date have engineered different targets using a variety of plant species and molecular tools, it has been difficult to draw conclusions about the comparative performance of different nucleases. Due to the time and effort required to regenerate engineered plants, efficiency is critical. In addition, there have been several reports of mutations at sequences with less than perfect identity to the target. While in some plant species it is possible to remove these so-called 'off-targets' by backcrossing to a parental line, the specificity of genome engineering tools is important when targeting specific members of closely-related gene families, especially when recent paralogues are co-located in the genome and unlikely to segregate. Specificity is also important for species that take years to reach sexual maturity or that are clonally propagated. Here, we directly compare the efficiency and specificity of Cas nucleases from different bacterial species together with engineered variants of Cas9. We find that the nucleotide content of the target correlates with efficiency and that Cas9 from Staphylococcus aureus (SaCas9) is comparatively most efficient at inducing mutations. We also demonstrate that 'high-fidelity' variants of Cas9 can reduce off-target mutations in plants. We present these molecular tools as standardised DNA parts to facilitate their re-use.
Data Availability Statement: All relevant data are
within the manuscript, the Supporting Information
files or have otherwise been deposited to online
repositories. Plasmid sequences and materials are
available from Addgene plasmid repository https://
www.addgene.org/Nicola_Patron/. Scripts are
https://github.com/EICoreBioinformatics/CRISPRanto. Sequences have
been deposited in the EMBL Nucleotide Sequence
Database (ENA) accession number PRJEB3044.
Components of bacterial CRISPR (Clustered Regulatory Interspaced Short Palindromic
Repeats) systems for adaptive immunity have been repurposed for engineering the genomes of
Funding: This work was funded by the UK
Biotechnological and Biological Sciences Research
Council (BBSRC) and Engineering and Physics
Research Council (EPSRC) Synthetic Biology for
Growth program [OpenPlant Synthetic Biology
Research Centre; BB/L014130/1]. Next-generation
sequencing and library construction and
automated DNA assembly was delivered via the
BBSRC National Capability in Genomics at the
Earlham Institute [BB/CCG1720/1]. CS was
supported by BBSRC strategic funding, Core
Capability Grant [BB/CCG1720/1, BBS/E/T/
000PR9816]. The funders had no role in study
design, data collection and analysis, decision to
publish, or preparation of the manuscript.
eukaryotic organisms [
]. These molecular tools have been rapidly and successfully applied
in many organisms, including plants (for recent reviews see [
], primarily due to the ease at
which they can be programmed to recognise new genomic targets. The majority of plant
genome engineering studies have utilised Cas9 (CRISPR Associated Protein 9) from
Streptococcus pyogenes (SpCas9), a monomeric nuclease found in the Type II CRISPR system of that
species. The Cas9 protein can be directed to selected genomic targets by an engineered RNA
moiety known as the single guide RNA (sgRNA) [
]. One or more sgRNAs can be
programmed to recognise new genetic targets by recoding the ~19 base pairs at the 5? end of the
RNA molecule, known as the spacer . The sgRNA forms a ribonuclease complex with Cas9
enabling it to scan DNA, pausing when it encounters a cognate sequence known as the
protospacer adjacent motif (PAM) [
]. On recognition of a cognate PAM, the ribonuclease complex
probes the adjacent sequence to determine identity to the spacer region. If complementary, the
spacer region of the sgRNA forms a Watson-Crick base-pair with the target DNA forcing the
Cas9 protein to undergo a conformational change that enables the nuclease domains to cleave
each of the DNA strands [
]. The most common application in plants has been targeted
mutagenesis achieved following transgenic expression of sgRNA-guided Cas9 to introduce double
strands breaks (DSBs) at selected genomic loci. These induced breaks are predominantly
repaired by the endogenous mechanism of non-homologous end-joining (NHEJ), which
sometimes introduces errors [
]. Since a perfect repair will continue to be recognised by the
Cas9/sgRNA complex, and therefore cut again, constitutive expression eventually results in
mutations, typically small insertions or deletions at the target. In many transgenic events, these
mutations occur sufficiently early in the development and regeneration of the transgenic plant
that all cells of the plant contain the same mutant genotype. A smaller number of studies have
successfully leveraged the induction of targeted DSBs to increase the efficiency of targeted
integration via the mechanism of homology directed repair (HDR), enabling transgenes to be
inserted at a precise locus or for genomic DNA sequences to be specifically recoded [
Additional Cas proteins from other bacterial species have been adapted for genome
engineering in eukaryotes and have also been applied to plants. These include Cas9 from the
Staphylococcus aureus Type II CRISPR system (SaCas9) [
] and Cas12a (previously Cpf1) from
the Type V CRISPR systems found in Francisella novicida, Acidaminococcus sp. and
Lachnospiraceae bacterium [
] (FnCas12a and LbCas12a). Cas proteins from different species
generally show preferences for different PAMs and, therefore, these additional Cas proteins
have increased the number of genomic sites that can be targeted for engineering. While
SpCas9 most efficiently cleaves DNA adjacent to NGG PAMs, SaCas9 is reported to show
preference for NNGRRT [
]. Cas12a-RNA complexes efficiently cleave target DNA preceded
by a short T-rich PAM, with data suggesting a preference for TTTV [
]. In addition,
engineered versions of Cas9 and Cas12a with mutations in their PAM recognition domains have
further expanded the repertoire of target sites to include sequences adjacent to NGAG and
NGCG PAMS (with variants of SpCas [
] and NNNRRT PAMs (with variants of SaCas9
Cas-mediated genome engineering in plants has often utilised established transformation
methods (e.g. Agrobacterium-mediated) to deliver DNA molecules encoding expression
cassettes for the Cas9 protein, one or more guide RNAs and, typically, a plant selectable marker
cassette, to the plant cell aiming for integration of all components at a single genetic locus. To
achieve this, several systems have been developed for the assembly of multigene constructs,
many of which employ Type IIS restriction cloning, also known as Golden Gate assembly
]. Following the recovery of transgenic plants, most studies report the ?efficiency? of
targeted mutagenesis as the percentage of transgenic plants in which mutations are found at
the intended target. Most published studies have utilised different genetic targets, construct
2 / 17
designs and regulatory elements making it difficult to draw conclusions about the comparative
efficiencies and specificities of each nuclease. A few recent studies have attempted to compare
Cas9 and Cas12a nucleases in similar experimental conditions (e.g.. However, they have
necessarily used different targets with each type of nuclease, which are known to exert
significant influence on efficiency [
Another area of interest has been the specificity of Cas nucleases. With careful
bioinformatic analysis it is often possible to identify target sites in the genomic region of interest that
are unique within the genome, thereby reducing the likelihood of mutations being induced at
alternative loci. However, a number of different studies, including several in plants, have
reported so-called "off-target" mutations at sites in the genome with less than perfect identity
to the spacer [
]. In some cases, small numbers of off-target mutations are of no concern,
especially as mutations are known to occur naturally during the process of tissue culture and
]. Further, many species can be back-crossed to the parent to remove
unwanted mutations. However, when the aim is to induce mutations in specific members of a
gene family, specificity is desirable, especially for recent paralogues that may be in close
proximity and therefore impossible to segregate. In addition, the introduction of additional
unwanted mutations in lineages that are not typically sexually propagated, such as cultivars of
Solanum tuberosum (potato), is particularly undesirable.
To increase the specificity of Cas9 proteins for their target, researchers have engineered the
nuclease and PAM-recognition domains [
]. The resulting proteins were initially tested
in mammalian cell cultures and reported to maintain similar levels of efficiency to the
wildtype protein at sequences with an exact match to the spacer, but limited activity at sequences
with less than perfect identity to the spacer. Here we report the outcomes of experiments that
compare the efficiency and specificity of multiple wild-type (SpCas9, SaCas9, FnCas12a and
LbCas12a) and engineered variants of Cas nucleases (eCas9 1.0, eCas9 1.1, eSaCas9 [
xCas9 3.7 ). All constructs used exactly the same regulatory elements and are assembled
into identical vectors for delivery. Importantly, the efficiency and specificity of each nuclease
are initially compared at the same target and, by quantifying the frequency of mutagenesis
induced during transient expression, we avoid the influence of transgene insertion location on
transgene expression. We also present an analysis of all potential targets as well as their
off-targets in the coding exons of Arabidopsis and test our tools at a larger number of targets to
identify factors that influence efficiency.
Analysis of targets in Arabidopsis coding sequences
Coding sequences were extracted from the Arabidopsis thaliana TAIR10 annotated whole
chromosome datasets (ftp://ftp.arabidopsis.org/home/tair/Sequences/whole_chromosomes/
gff3/TAIR10_GFF3_genes.gff). Jellyfish (v.2.2.6)  was combined with a script (available at
https://github.com/EI-CoreBioinformatics/CRISPRanto) to extract all possible legitimate
candidate target sequences for each Cas nuclease [N20NGG (SpCas9), N21-NNGGGT (SaCas9),
N20-NGAG (SpCas9-VQR), N20-NGCG (SpCas9-VRER), TTTV-N23 (Lb/AsCas12a)].
Potential off-targets (target sequences that are either identical to another target or differ by, at most,
one base pair in the spacer region) were detected by mapping all identified targets against the
TAIR10 genomic sequences using bbmap (v.38.06) with the following parameters (k = 8,
minid = 0.75, ambig = all, mappedonly = t, secondary = t, sssr = 0.75, ssao = t, saa = f, mdtag =
t, nhtag = t, xmtag = t, amtag = t, nmtag = t, xstag = t, indelfilter = 0, subfilter = 4, editfilter = 4).
Alignments were filtered by Hamming distance allowing at most one mismatch in the spacer
3 / 17
(plus mismatches at the ambiguous positions of the respective PAM sequences). Any potential
off-targets located in coding sequences were identified with Bedtools intersect (v. 2.26.0; 
requiring an overlap equal to the length of the target including the PAM. All identified targets
are provided S1 Table. All scripts and a snakemake pipeline  containing the whole
workflow are available at https://github.com/EI-CoreBioinformatics/CRISPRanto.
Selection of targets and off-targets for assessment of targeted mutagenesis
To remove any potential variability that might be associated with gene expression, we selected
targets/off-target pairs located in genes expressed in leaves. To do this, we compared our
candidate target/off-target pairs to gene expression data from the Expression Atlas (https://www.
ebi.ac.uk/gxa/home, experiment E-GEOD-38612), retaining those targets within genes
expressed in leaves. We then selected candidate target/off-target pairs that differed within the
first 3 bp distal to the PAM (S2 Table). Finally, we either selected target/off-target pairs in
which the expected cleavage site overlapped with the recognition site for a Type II restriction
endonuclease, or searched 1000 bp of sequence each side of the target/off-target pairs to
identify the presence of an additional target, identical in both sequences, that would enable the
creation of a deletion when a second sgRNA recognising this target was delivered.
Assembly of constructs
Constructs were assembled using the Plant Modular Cloning (MoClo) plasmid toolkit , a
gift from Sylvestre Marrillonet (Addgene Kit # 1000000044). New Level 0 parts were made
according to the standards described in . All Level 0 parts encoding Cas9 proteins
contained a nuclear localisation (NLS) KKRKVKKRKVKKRV signal and were assembled together
with a C-terminal yellow fluorescent protein (YFP). Cas12a proteins were fused to the same
Cterminal NLS tag, followed by a 3xHA tag (a YFP tag was not used as it has not been
determined if this reduces nuclease activity). Level 0 parts were assembled into transcriptional units
in Level 1 acceptor plasmids in a one-step digestion-ligation reaction, except for sgRNAs, in
which a U6 promoter L0 part was assembled into Level 1 acceptor plasmid together with a
PCR amplicon of the complete sgRNA (the spacer is fused to the RNA scaffold as a 5? extension
of the forward primer), as described in [
]. Subsequently, Level 1 transcriptional units were
assembled into the Level 2 acceptor plasmid pICSL4723 (Addgene #86172). The
digestion-ligation reactions were set up either manually or at nanoscale using laboratory automation. For
manual assembly, 15 fmol of each DNA part was combined with 7.5 fmol of the acceptor
plasmid in a final volume of 5 ?L dH20. This was combined with 5 ?L of reaction mix (3 ?L of
dH20, 1 ?L of T4 DNA ligase buffer 10x (NEB, Ipswich, MA, USA), 0.5 ?L of 1 mg/mL purified
bovine serum albumin (1:20 dilution in dH20 of BSA, Molecular Biology Grade 20 mg/mL,
NEB), 0.25 ?L of T4 DNA ligase at 400 U/?L (NEB) and 0.25 ?L of BsaI or BpiI restriction
enzyme at 10 U/?L (ThermoFisher, Waltham, MA, USA)) and incubated in a thermocycler for
26 cycles of 37?C for three minutes followed by 16?C for four minutes and a final incubation at
37?C for 5 minutes followed by 80?C for five minutes. A 2 ?L aliquot of each reaction was
transformed into 20 ?L electrocompetent JM109 or DH5alpha Escherichia coli cells and plated
on selective LB-agar plates. Automated reactions were scaled down to a final reaction volume
of 1 ?L using the Echo 550 liquid handler (Labcyte Ltd. San Jose, SA, USA), transformed into
2 ?L XL10-Gold Ultracompetent E. coli (Agilent Technologies, Santa Clara, CA, USA) and
plated onto eight-well selective LB-agar plates on a Hamilton STARplus platform. The
sequences of assembled plasmids were verified by complete sequencing using 150 base pair
paired-end reads on an Illumina MiSeq platform. Libraries were prepared using the Nextera
XT DNA Library Prep Kit (Illumina, San Diego, CA, USA) with a modified 2 ?L total volume
4 / 17
protocol using a one in 25 dilution of components. A complete list of all 132 plasmids,
comprising Level 0 DNA parts, Level 1 transcriptional units and Level 2 constructs, used in this
study is given in S2 Table. Samples, together with complete annotated sequence files for the
128 new plasmids generated for this study have been deposited at the AddGene plasmid
Transient expression in protoplasts
For each experiment, a sufficient number of protoplasts were prepared to enable the delivery
of four replicates of all constructs to be compared. Protoplasts were prepared from the leaf
tissues of Nicotiana benthamiana or A. thaliana as previously described . Protoplasts were
quantified and divided into aliquots of 200 ?L in transfection buffer (0.4M mannitol, 15mM
MgCl2, 4 mM MES, pH 5.6), each containing approximately 1 x 104/mL intact protoplasts,
such that four separate aliquots of protoplasts from the same preparation were available for
each of the plasmids to be compared. Plasmid DNA for delivery to protoplasts was prepared
using the Plasmid Plus Midi kit (Qiagen, Hilden, Germany) with a modified protocol
incorporating three additional wash steps prior to elution from the column. Freshly made PEG (2 g of
PEG (Mn 4000 (Sigma, 81240)) in 2 mL of 500 mM mannitol and 0.5 mL of 1M CaCl2) was
mixed with10 ?g of purified DNA and added to each aliquot of protoplasts. Subsequently,
protoplasts were washed and resuspended in 300 ?L of washing buffer (154 mM NaCl, 125 mM,
CaCl2, 5 mM KCl, 2 mM MES; pH5.6) and incubated for 24 hours at 24?C in an illuminated
incubator with light intensity of approximately 70 ?mol/m2/s. Transformation efficiency was
estimated by quantification of protoplasts in which YFP fluorescence was visible in the nuclei
using an inverted fluorescence microscope (Zeiss Axio Observer Z1 or ThermoFisher Evos).
Detection and quantitation of targeted mutagenesis
DNA was extracted from the protoplasts using a cetyltrimethylammonium bromide (CTAB)
extraction protocol. Pellets of protoplasts were resuspended in 100 ?L of extraction buffer
(0.2M Tris-HCl, pH7.5; 0.05M EDTA; 2M NaCl; 2% CTAB, pH7.4) and incubated at 65?C for 1
hour prior to addition of 45 ?L chloroform. Following centrifugation, the upper aqueous phase
was precipitated with an equal volume of isopropanol. DNA pellets were washed with 70% w/v
ethanol, dried and resuspended in sterile distilled water with 5 ?g/?L RNAse A (Thermo
Fisher). Each target was amplified using a pair of primers, specific to the locus of interest (S3
Table). PCR reactions were performed using 70 ng DNA and Q5 High-Fidelity DNA
Polymerase (NEB) according to the instructions provided by the manufacturer. Mutations at the targets
were identified by either Illumina or Sanger sequencing. For preparation for Illumina
sequencing, amplicons were purified using Agencourt AMPure XP (Beckman Coulter) and indexed
using the Nextera XT Library Preparation kit (Illumina) according to the manufacturer?s
instructions. Reactions were analysed by microfluidic gel fractionation (LabChip GXII, Perkin
Elmer) and pooled. Primers were removed by fractionation (BluePippin, SAGE Science). The
concentration and quality of DNA was analysed by QUBIT (ThermoFisher) and qPCR (Kapa
Library Quantification Kit, Illumina). phiX Sequencing Control V3 (Illumina) was added to
final concentration of 1.75 pM. Sequencing was performed on an Illumina MiSeq using the
MiSeq Reagent Kit v2 Micro. Adapter sequences were removed and quality trimmed using
bbduk (v.37.24) (https://jgi.doe.gov/data-and-tools/bbtools/bb-tools-user-guide/bbduk-guide/)
(parameters ktrim = r k = 21 mink = 11 hdist = 2 qtrim = lr trimq = 3 maq = 10 ftr = 250). All
sequences have been deposited in the EMBL Nucleotide Sequence Database (ENA) accession
number = PRJEB3044. Quantification of mutations at the target was performed using
CRISPRESSO (http://crispresso.rocks/)  and automated with a custom script (available at https://
5 / 17
github.com/EI-CoreBioinformatics/CRISPRanto). The quantity of mutations in each sample
was then normalised to the quantified transfection efficiency.
For analysis by Sanger sequencing, amplicons were purified (Qiaquick PCR purification,
Qiagen) and incubated with a restriction enzyme for which the recognition sequence
overlapped the expected cleavage site prior to reamplification. This reduced the amount of
wildtype sequence in the sample, enabling detection of low-abundance amplicons. These
amplicons were sequenced directly (Eurofins) and evidence of mutagenesis was conferred by the
presence of multiple peaks, representing the different DSB-repair events across the population
of cells, visible after the expected cut site (S1 Fig). These chromatogram signals were analysed
using the ICE software that determines rates of CRISPR-Cas9 editing at a specific, sgRNA
directed genomic location within a cell population (Synthego, https://ice.synthego.com/).
Codon optimisation and sgRNA structure have minimal effects on the
efficiency of targeted mutagenesis in plants
Prior to comparison with other Cas nucleases, we first assessed several variables for
RNAguided Cas9. We assessed the effect of codon-optimisation, comparing human (SpCas-h)
versus plant (SpCas9-p), as well as variations in the single guide RNA sequence, comparing two
sgRNAs: sgRNA and a sgRNA with extended step-loops (sgRNA-ES; . We also compared
the use of a previously extended endogenous terminator for sgRNA expression cassettes
[50,51]. In these and subsequent experiments, all constructs were assembled similarly using
identical regulatory sequences (Fig 1A). All sgRNAs contained a spacer to direct Cas9 to a
target in the phytoene desaturate gene of Nicotiana benthamiana (NbPDS1) and constructs were
delivered to protoplasts isolated from leaves of N. benthamiana. Following a quantitative
assessment of the frequency of mutagenesis by Illumina sequencing, we found no significant
differences in the number of sequencing reads with mutations at the target (Fig 1B), indicating
that neither human codon optimisation, the shorter stems found in the original sgRNA, or the
minimal terminator significantly impaired the efficiency of mutagenesis in plants. In all
subsequent experiments, we used SpCas9-h together with its original sgRNA as we have had
previous success with these sequence in other species [
Cas proteins from different bacterial species show varied efficiencies of
targeted mutagenesis at the same target in identical experimental
To enable a direct comparison of the ability of four Cas proteins, SaCas9, SpCas9, AsCas12
and LbCas12a to induce targeted mutations, without the confounding influence of varied
efficiency across targets, we designed three variants of a synthetic target such that the same
recognition sequence was adjacent to the preferred PAM for each protein (Fig 1C). We then
codelivered the plasmid containing the target with the cognate PAM to protoplasts together with
plasmid DNA encoding each of the four nucleases and an appropriate sgRNA (Cas9) or
crRNA (Cas12a) with a spacer to the synthetic target. We observed significantly more
mutations at the NbPDS1 target with SaCas9 than with SpCas9-h, AsCas12 or LbCas12a (Fig 1D).
Efficiency of SpCas9-induced mutagenesis correlates to GC content of the
Analysis of the Arabidopsis genome for potential target sequences identified 3,853,090
potential targets for SpCas9, SaCas9 or Lb/AsCas12a in coding exons (Table 1). Of the 2,695,798
6 / 17
Fig 1. Comparison of Cas9 and Cas12 nucleases at a single target. (A) All constructs were assembled in the same backbone, using the same regulatory elements.
Asterisks ( ) represent elements compared in this study. The ?dummy? consists of 15 random nucleotides and enables the Cas9 and sgRNA expression cassettes to be
reassembled with, e.g. a selectable marker cassette, in Position 1. (B) Assessment of codon-optimisation and sgRNA scaffold on the efficiency of Cas9-mediated targeted
mutagenesis at PDS1 in Nicotiana benthamiana protoplasts (h = human; p = plant; ES = extended stem; T = terminator). (C) An identical target sequence was used to
test Cas proteins from different bacteria. The target was flanked by the preferred cognate protospacer adjacent motif (PAM) for SpCas9 (red box), SaCas9 (blue box) or
Cas12a (black box). The spacer sequence used in each sgRNA (Cas9) or crRNA (Cas12a) are indicated by dotted lines. (D) Under identical experimental conditions,
SaCas9 induced more mutations at the target than SpCas9, AsCas12a or LbCas12a. Error bars = standard error of the mean; n = 4. Post hoc comparisons using Tukey?s
Honest Significant Difference test indicted significant differences: = p<0.05, = p<0.01.
targets recognised by SpCas9, 61,739 (2.29%) have at least one potential off-target (identical or
differing at only a single base) also in a coding exon, 7,201 of which differed by a single base in
the first three positions. The total number of targets is expectedly lower for SaCas9 and Lb/
AsCas12a, which recognise longer PAMs, (Table 1). We filtered these targets for those in genes
previously shown to be expressed in leaves (see methods) and selected sequences for functional
analysis in Arabidopsis leaf-derived protoplasts.
We were able to detect evidence of induced mutations at 14 out of 33 candidate targets with
prefect identity to the spacer for SpCas9 (the sequences of all targets tested are provided in S1
Table). We were able to detect evidence of induced mutations at seven out of nine candidate
of targets in coding exons
Identical to target
Number and similarity of potential
off-targets in coding exons
One SNP in positions
1, 2 or 3
(distal to PAM)
One SNP at
any other position
7 / 17
Fig 2. Nucleotide content analysis of 33 spacer sequences. All spacers have cognate targets in the coding sequences of
leaf-expressed genes tested in sgRNAs with SpCas9-h. Each spacer was incorporated into a sgRNA and tested in
Arabidopsis protoplasts. Blue bars indicate mutations were detected at target; red bars indicate no mutations detected
at target. Seed = six base pairs adjacent to PAM. = p-value 0.044591. The result is significant at p <0.05.
targets with perfect identity to the spacer for SaCas9. The sample set for SaCas9 is too small for
meaningful analysis, however, analysis of nucleotide composition of the 33 SpCas9 targets
found that the %GC content of spacers used where mutagenesis was detected was significantly
greater than those for which no activity was detected (single-tailed T-test, P<0.05) (Fig 2). The
data indicates that GC content of spacers should, for maximal efficiency, be greater than 40%.
In contrast, we found no differences in the GC content within the seed region.
?High-fidelity? variants of Cas9 show improved specificity at some targets
To directly compare the efficiency and specificity, in planta, of wild-type SpCas9 and the
engineered variants, SpCas9 DE [
], SpCas9 KA, eCas9 1.0 and eCas9 1.1 [
], we used a similar
experimental process as reported by [
]. We designed a set of five sgRNAs each with a
mutation in a different base of a spacer designed to target the NbPDS gene (Fig 3A). A sgRNA with
an exactly matching spacer, or one of the five variants was delivered to N. benthamiana
protoplasts in combination with each of the five variants of SpCas9 and the number of targeted
mutations was quantified using Illumina sequencing. While the number of mutations induced
by wild-type Cas9 was significantly reduced when the spacer contained a mutation in the
region close to the PAM, the presence of a mismatch between the spacer and target in the distal
region had minimal effects (Fig 3B). In contrast, the frequency of mutations induced by the
8 / 17
Fig 3. Assessment of Cas9 specificity at an identical target. (A) Schematic showing transversions in four regions of the spacer sequences of sgRNAs targeting NbPDS1
(B) The efficiency of targeted mutagenesis is impacted by transversions in the spacer. Shaded bars represent a perfect match of the spacer to the target, black bars show
sgRNAs with spacers with transversions relative to the target. Error bars = standard error of the mean; n = 4. Post hoc comparisons using Tukey?s Honest Significant
Difference test indicted significant differences: = p<0.05, = p<0.01.
variants eCas9 1.0 and 1.1 was significantly reduced by a mismatch in any region of the sgRNA
(Fig 3B). This experiment, although quantitative, was difficult to scale across a larger number
of targets. To compare the efficiency of mutagenesis across a larger number of targets and to
observe the performance of spacers at non-identical endogenous targets, we conducted an
analysis of the Arabidopsis genome to identify pairs of targets that differed by a single base
pair. For each target at which mutagenesis was detected using SpCas9-h or SaCas9, the same
sgRNAs were also tested with a number of variants of SpCas9 (eCas9 1.0, eCas9 1.1, eSaCas9
] and the recently reported xCas9 3.7 ) all described to have reduced activity at targets
9 / 17
Differences between target (bold text) and off-target pairs are shown in red text. The Protospacer Adjacent Motif (PAM) is shown in blue text. Numbers in parentheses
are scores from Inference of CRISPR Editing (ICE) software that quantifies the efficiency of induced mutations at the target from the sequence chromatogram. Y/N
indicates that mutations at the target were/were not detected
indicates assessment made by detection of deletion induced with a second sgRNA.
| indicates expected point of cleavage.
with a less than perfect match to the spacer. In addition to analysis at the targets, we also
analysed the identified off-target locus for evidence of mutagenesis (Table 2). Of eight targets at
which SpCas9-h was able to induce mutations, the SpCas9 variants were only able to induce
mutations at three (xCas9 3.7 and eCas9 1.0) or four (eCas91.1) targets (Table 2). SpCas9-h
also induced mutations at all off-targets that differed from the spacer by one nucleotide in the
first three positions. Analysis of the sequence chromatogram using the Inference of CRISPR
Editing (ICE) software indicated that, in most cases, SpCas9-h induced mutations at a similar
efficiency at the off-target as at the target (Table 2 and S1 Fig). In some cases, however, the
efficiency of mutagenesis was reduced at off-target loci when the sgRNA was delivered with a
variant of Cas9 (Table 2 and S1 Fig). In agreement with the data obtained for NbPDS (Fig 3), the
efficiency of mutagenesis at some targets was similar for SpCas9-h and eCas9 1.0, while
efficiency at the off-target was reduced with eCas9 1.0 (Table 2). However, at several other targets,
the SpCas9 variants were unable to induce mutations. In addition, the use of variants did not
always reduce mutagenesis at the off-target locus (Table 2).
10 / 17
We also compared the efficiency and specificity of SaCas9 with an engineered variant
]. Of four targets at which SaCas9 was able to induce mutations, eSaCas9 variants
were only able to induce mutations at three. In similarity with SpCas9-h, SaCas9 also induced
mutations at all off-targets that differed from the spacer by one nucleotide in the first three
positions. In contrast, eSaCas9 only induced mutations at one target without perfect identity
to the spacer.
Components of bacterial CRISPR/Cas systems have been applied to a wide variety of model
and economically-important plant species including dicotyledonous fruit crops such as
strawberries , tomatoes  and oranges  as well as monocotyledonous grain crops such as
], wheat  and maize . In most cases, engineered plants are regenerated via
somatic embryogenesis in tissue culture; a process that is both laborious and time-consuming.
Consequently, researchers are interested in the efficiency of the molecular tools they deliver.
Since the first reports of Cas9 from S. pyogenes being applied as tool for genome engineering,
there has been much interest in novel properties associated with Cas nucleases from addition
CRISPR systems [
]. These include comparative efficiency and the potential for engineering
at an increased number of genomic targets due to the recognition of additional PAMs. Because
of this, Cas nucleases have often, necessarily, used different targets. Together with other
differences introduced by different construct designs and experimental conditions, it is difficult to
draw conclusions about relative efficiencies. Our experimental strategies allowed us to
compare the efficiency of different components for genome engineering on the same target and
indicate that SaCas9 may have the highest efficiency of targeted mutagenesis in plants (Fig 1).
The drawback of our experimental design, however, is that that the same experimental
conditions are unlikely to be optimal for each Cas nuclease. For example, recent work has shown
that temperature can increase the activity of Cas12a [
]. It is also important to note that our
experimental system, necessarily, used the same regulatory elements in order to compare
different nucleases. Since our experiments were performed in leaf-derived protoplasts of N.
benthamiana and A. thaliana, we utilised the frequently-used 35s promoter from Cauliflower
Mosaic Virus (CaMV-35s), which expresses strongly in this cell type, to drive expression of all
nucleases. This promoter will not be optimal for all species and the choice of promoters is
likely to contribute to efficiency in different species. Even in Arabidopsis, where stable
transgenesis utilises a floral-dip method, it is preferential to use a promoter that expressed strongly
in developing embryos to minimise the recovery of plants that are genetically-chimeric due to
expression of the Cas-nuclease in somatic cells during development [50,57,58].
Much of the work on sequence determinants of spacers performed to date has been
performed in mammalian cell cultures [
]. These systems provide the advantage of being
able to delivery libraries of sgRNAs targeting genes involved in a particular process followed
by the selection of cells with the expected phenotype correlating to disruption of those genes to
be selected (e.g. by staining and FACS). In an analysis of a number of datasets, including one
of 4,000 sgRNA targeting 17 genes,  concluded that the patterns of activity are complex
and are not likely to be apparent by examining smaller numbers of sgRNA:DNA interactions.
Current methods of DNA delivery to plant cells include the preparation of separate strains of
Agrobacterium tumefaciens for delivery to callus or tissues, or the direct delivery of plasmids
either to protoplasts or to tissues (using biolistic delivery). Delivery of a large-scale combined
library of constructs using these methods would result in populations of cells with multiple
induced mutations from which it would not be possible to identify or separate cells with
individual genotypes. In our study, we delivered individual constructs to aliquots of protoplasts
11 / 17
prepared from the same batch of leaf tissue. This allowed us to work with larger number of
constructs, however, since cultures of protoplasts cannot be perpetuated (they reform cell
walls and form masses of cells if allowed or induced to divide), it is not possible to select
populations of cells beyond the first 24?48 hours. The advantages of this system, however, allowed
us to compare the impact of a number of different components by allowing similar levels of
expression across different experiments that are difficult to achieve with integrated transgenes,
the expression of which is subject to copy number and the location of integration [
this method, we attempted to induce mutations at over 45 targets in the Arabidopsis genome,
using sgRNAs with spacers with both perfect and imperfect identity. This enabled us to
compare the features of functional spacers (those that were able to induce targeted mutations) with
non-functional spacer noting a correlation with GC content (Fig 2), broadly in agreement with
that reported for mammals . We did not, however, note any determinants for sequence
composition in the seed region or at specific bases, likely because of the comparatively small
size of our dataset. The correlation of efficiency with GC content may be related to nucleosome
occupancy, previously reported to influence the ability of Cas9 to recognise its targets [
nucleosome-depleted regions in Arabidopsis tend to have a higher GC content [
]. As this
negative correlation between nucleosome occupancy and GC content is not known to be
common across all organisms, further studies in a wider range of plant species will need to be
performed to determine if efficiency can be predicted by GC content alone.
Analysis of Arabidopsis coding sequences revealed that many targets for Cas-mediated
genome engineering have one or more potential off-targets in other coding sequences that differ
by just a single base (Tables 1 and S1). We provide experimental evidence that SpCas9 and SaCas9
are readily able to induce mutations at targets with less than perfect identity to the spacer (Table 2
and Fig 3). Different laboratories have engineered variants of SpCas9 and SaCas9, reportedly
reducing their activity at targets to which the spacer does not have 100% identity [
compare these proteins and determine their function in plants, we measured their efficiency and
specificity across a number of targets (Table 2 and Fig 3). Previous reports have found that the
engineered variants showed increased specificity and retained a similar efficiency as the wild-type
protein when the spacer was identical to the target . We found this to be true at some targets
but at others the engineered variants were often either less efficient or non-functional.
Additionally, we found that use of the variants did not always increase specificity and that the efficiency of
mutagenesis at some off-targets was similar to wild-type Cas9.
To perform these experiments, we utilised standardised modular cloning techniques to
create a suite of comparable constructs to enable direct comparison of multiple different tools for
genome engineering. Type IIS mediated assembly methods have been widely utilised to
facilitate the construction of the complex plasmids required for multiplexed genome editing
]. In addition to standardising our experimental process, we are able to provide an
expanded toolkit of modular, reusable parts for plant genome engineering that will facilitate
their application in new studies (Fig 4). The basic (Level 0) parts are flanked with inverted BsaI
sites that will release parts with overhangs in the common genetic syntax for plants ,
making them amenable for reuse with a number of different assembly toolkits for plants.
This study has provided data that will be useful to plant scientists when selecting targets
and molecular components for Cas-mediated targeted mutagenesis in plants. We found that
SaCas9 is likely to be the most efficient nuclease. When selecting targets for mutagenesis with
SpCas9, spacers with a GC content above 45% may be preferential. Regarding specificity, both
SaCas9 and SpCas9 are highly likely to induce mutations at any off-target loci that differ from
the target by a single nucleotide close to the 5? end. Specificity may be improved by the use of
eCas9 1.0, eCas9 1.1 or xCas9 3.7, however, a greater number of targets should be tested as
efficiency is likely to be reduced or eliminated at some targets.
12 / 17
Fig 4. An expanded toolkit for Cas-mediated genome engineering in plants. Regulatory elements and coding sequences are cloned as Level 0 parts, enabling one-step
assembly into transcriptional units mediated by BsaI (e.g. using the Plant MoClo Toolkit, ) and subsequently into multigene constructs. Numbers in parentheses
represent catalogue number at Addgene. For simplicity, all cassettes are shown assembled on the forward strand. However, the orientation of any cassette in the final
construct can be altered by use of a reverse Level 1 acceptor. ?+S? = with stop codon ?-S? = no stop codon.
S1 Table. Genomic targets for Cas nucleases in Arabidopsis thaliana coding exons.
S2 Table. Details of plasmid constructs used in this study.
S3 Table. Oligonucleotide primers used to amplify genomic targets.
S1 Fig. Chromatograms and gel images of Cas-induced targeted mutations and deletions.
Conceptualization: Oleg Raitskin, Nicola J. Patron.
Data curation: Nicola J. Patron.
Formal analysis: Oleg Raitskin, Christian Schudoma, Nicola J. Patron.
Funding acquisition: Nicola J. Patron.
Investigation: Oleg Raitskin, Christian Schudoma, Anthony West, Nicola J. Patron.
Methodology: Oleg Raitskin, Christian Schudoma, Anthony West, Nicola J. Patron.
Project administration: Nicola J. Patron.
13 / 17
Resources: Oleg Raitskin, Christian Schudoma, Nicola J. Patron.
Software: Christian Schudoma.
Supervision: Nicola J. Patron.
Validation: Oleg Raitskin.
Writing ? original draft: Nicola J. Patron.
Writing ? review & editing: Oleg Raitskin, Christian Schudoma, Anthony West, Nicola J.
14 / 17
15 / 17
Hu JH, Miller SM, Geurts MH, Tang W, Chen L, Sun N, et al. Evolved Cas9 variants with broad PAM
compatibility and high DNA specificity. Nature. 2018; 556(7699):57?63. https://doi.org/10.1038/
nature26155 PMID: 29512652
Kleinstiver BP, Pattanayak V, Prew MS, Tsai SQ, Nguyen NT, Zheng Z, et al. High-fidelity CRISPR?
Cas9 nucleases with no detectable genome-wide off-target effects. Nature. 2016; 529(7587):490?5.
https://doi.org/10.1038/nature16526 PMID: 26735016
Marc?ais G, Kingsford C. A fast, lock-free approach for efficient parallel counting of occurrences of
kmers. Bioinformatics. 2011; 27(6):764?70. https://doi.org/10.1093/bioinformatics/btr011 PMID:
Quinlan AR, Hall IM. BEDTools: A flexible suite of utilities for comparing genomic features.
Bioinformatics. 2010; 26(6):841?2. https://doi.org/10.1093/bioinformatics/btq033 PMID: 20110278
K o?ster J, Rahmann S. Snakemake-a scalable bioinformatics workflow engine. Bioinformatics. 2012; 28
(19):2520?2522. https://doi.org/10.1093/bioinformatics/bts480 PMID: 22908215
Engler C, Youles M, Gr u?etzner R. A Golden Gate modular cloning toolbox for plants. ACS Synth Biol.
2014; 3(11):839?43. https://doi.org/10.1021/sb4001504 PMID: 24933124
16 / 17
1. Mali P , Yang L , Esvelt KM , Aach J , Guell M , DiCarlo JE , et al. RNA-Guided Human Genome Engineering via Cas9 Prashant . Science. 2013 ; 339 ( 6121 ): 823 - 6 . https://doi.org/10.1126/science.1232033 PMID: 23287722
2. Cong L , Ran FA , Cox D , Lin S , Barretto R , Habib N , et al. Multiplex genome engineering using CRISPR/ Cas systems . Science 2013 ; 339 ( 6121 ): 819 - 23 . https://doi.org/10.1126/science.1231143 PMID: 23287718
3. Zetsche B , Gootenberg JS , Abudayyeh OO , Regev A , Koonin E V. , Zhang F , et al. Cpf1 Is a Single RNA-Guided Endonuclease of a Class 2 CRISPR-Cas System . Cell 2015 ; 163 ( 3 ): 759 - 71 . https://doi. org/10.1016/j.cell. 2015 . 09 .038 PMID: 26422227
4. Yin K , Gao C , Qiu J-L. Progress and prospects in plant genome editing . Nat Plants . 2017 ; 3 : 17107 . https://doi.org/10.1038/nplants. 2017 .107 PMID: 28758991
5. Gao C. The future of CRISPR technologies in agriculture . Nature Reviews Molecular Cell Biology . 2018 ; 19 : 275 - 276 https://doi.org/10.1038/nrm. 2018 .2 PMID: 29382940
6. Ricroch A , Clairand P , Harwood W. Use of CRISPR systems in plant genome editing: toward new opportunities in agriculture . Emerg Top Life Sci . 2017 ; 1 ( 2 ): 168 - 92 .
7. Barrangou R , Fremaux C , Deveau H , Richards M , Boyaval P , Moineau S , et al. CRISPR provides acquired resistance against viruses in prokaryotes . Science 2007 ; 315 ( 5819 ): 1709 - 12 . https://doi.org/ 10.1126/science.1138140 PMID: 17379808
8. Sternberg SH , Redding S , Jinek M , Greene EC , Doudna JA . DNA interrogation by the CRISPR RNAguided endonuclease Cas9 . Nature . Nature Publishing Group; 2014 ; 507 ( 7490 ): 62 - 7 . https://doi.org/ 10.1038/nature13011 PMID: 24476820
9. Schiml S , Puchta H . Revolutionizing plant biology: multiple ways of genome engineering by CRISPR/ Cas . Plant Methods. BioMed Central; 2016 ; 12 ( 1 ): 8 .
10. Hu Z , Cools T , De Veylder L. Mechanisms Used by Plants to Cope with DNA Damage . Annu Rev Plant Biol . 2016 ; 67 ( 1 ): 439 - 62 .
11. Cermak T , Curtin S , Gil-Humanes J , Cegan R , Starker CG , Kono TJY , et al. A multi-purpose toolkit to enable advanced genome engineering in plants . Plant Cell . 2017 ;
12. Gil-Humanes J , Wang Y , Liang Z , Shan Q , Ozuna C V. , Sanchez-Leon S , et al. High-efficiency gene targeting in hexaploid wheat using DNA replicons and CRISPR/Cas9 . Plant J. 2017 ; 89 ( 6 ): 1251 - 62 . https://doi.org/10.1111/tpj.13446 PMID: 27943461
13. Baltes NJ , Gil-humanes J , Cermak T , Atkins PA , Voytas DF . DNA Replicons for Plant Genome Engineering . Plant Cell . 2014 ; 26 (January): 151 - 63 . https://doi.org/10.1105/tpc.113.119792 PMID: 24443519
14. Begemann MB , Gray BN , January E , Gordon GC , He Y , Liu H , et al. Precise insertion and guided editing of higher plant genomes using Cpf1 CRISPR nucleases . Sci Rep . Springer US; 2017 ; 7 ( 1 ): 1 - 6 .
15. Li J-F , Norville JE , Aach J , McCormack M , Zhang D , Bush J , et al. Multiplex and homologous recombination-mediated genome editing in Arabidopsis and Nicotiana benthamiana using guide RNA and Cas9 . Nat Biotechnol . 2013 , Aug 8 ; 31 ( 8 ): 688 - 91 . https://doi.org/10.1038/nbt.2654 PMID: 23929339
16. ?erma?k T , Baltes NJ , ?egan R , Zhang Y , Voytas DF . High-frequency, precise modification of the tomato genome . Genome Biol . 2015 ; 16 ( 1 ): 232 .
17. Kaya H , Mikami M , Endo A , Endo M , Toki S . Highly specific targeted mutagenesis in plants using Staphylococcus aureus Cas9 . Sci Rep . 2016 ; 6 ( 1 ): 26871 .
18. Steinert J , Schiml S , Fauser F , Puchta H . Highly efficient heritable plant genome engineering using Cas9 orthologues from Streptococcus thermophilus and Staphylococcus aureus . Plant J . 2015 ; 84 ( 6 ): 1295 - 305 . https://doi.org/10.1111/tpj.13078 PMID: 26576927
19. Ran FA , Cong L , Yan WX , Scott DA , Gootenberg JS , Kriz AJ , et al. In vivo genome editing using Staphylococcus aureus Cas9 . Nature . 2015 ; 520 ( 7546 ): 186 - 91 . https://doi.org/10.1038/nature14299 PMID: 25830891
20. Xu R , Qin R , Li H , Li D , Li L , Wei P , et al. Generation of targeted mutant rice using a CRISPR-Cpf1 system . Plant Biotechnol J . 2017 ; 15 ( 6 ): 713 - 7 . https://doi.org/10.1111/pbi.12669 PMID: 27875019
21. Tang X , Lowder LG , Zhang T , Malzahn AA , Zheng X , Voytas DF , et al. A CRISPR-Cpf1 system for efficient genome editing and transcriptional repression in plants . Nat Plants . 2017 ; 3 : 17018 . https://doi.org/ 10.1038/nplants. 2017 .18 PMID: 28211909
22. Kleinstiver BP , Prew MS , Tsai SQ , Nguyen NT , Topkar V V. , Zheng Z , et al. Broadening the targeting range of Staphylococcus aureus CRISPR-Cas9 by modifying PAM recognition . Nat Biotechnol . 2015 ; 33 ( 12 ): 1293 - 8 . https://doi.org/10.1038/nbt.3404 PMID: 26524662
23. Xie H , Tang L , He X , Liu X , Zhou C , Liu J , et al. SaCas9 Requires 50 -NNGRRT-30 PAM for Sufficient Cleavage and Possesses Higher Cleavage Activity than SpCas9 or FnCpf1 in Human Cells . Biotechnol J . 2018 ; 14 ( 4 ): e1700561 .
24. Friedland AE , Baral R , Singhal P , Loveluck K , Shen S , Sanchez M , et al. Characterization of Staphylococcus aureus Cas9: A smaller Cas9 for all-in-one adeno-associated virus delivery and paired nickase applications . Genome Biol . 2015 ; 24 ( 16 ): 257 .
25. Yamano T , Zetsche B , Ishitani R , Zhang F , Nishimasu H , Nureki O. Structural Basis for the Canonical and Non-canonical PAM Recognition by CRISPR-Cpf1 . Mol Cell . 2017 ; 67 ( 4 ): 633 - 45 . https://doi.org/ 10.1016/j.molcel. 2017 . 06 .035 PMID: 28781234
26. Kleinstiver BP , Prew MS , Tsai SQ , Topkar V V. , Nguyen NT , Zheng Z , et al. Engineered CRISPR-Cas9 nucleases with altered PAM specificities . Nature . 2015 ; 523 ( 7561 ): 481 - 5 . https://doi.org/10.1038/ nature14592 PMID: 26098369
27. Werner S , Engler C , Weber E , Gruetzner R , Marillonnet S. Fast track assembly of multigene constructs using Golden Gate cloning and the MoClo system . Bioeng Bugs . 2012 ; 3 ( 1 ): 38 - 43 . https://doi.org/10. 1371/journal.pone. 0016765 PMID: 22126803
28. Engler C , Kandzia R , Marillonnet S. A one pot, one step, precision cloning method with high throughput capability . PLoS One . 2008 ; 3 ( 11 ):e3647. https://doi.org/10.1371/journal.pone. 0003647 PMID: 18985154
29. Xing H-L , Dong L , Wang Z-P , Zhang H-Y, Han C-Y , Liu B , et al. A CRISPR/Cas9 toolkit for multiplex genome editing in plants . BMC Plant Bio . 2014 ; 14 ( 1 ): 327 .
30. Ma X , Zhang Q , Zhu Q , Liu W , Chen Y , Qiu R , et al. A Robust CRISPR/Cas9 System for Convenient, High-Efficiency Multiplex Genome Editing in Monocot and Dicot Plants . Mol Plant . 2015 ; 8 ( 8 ): 1274 - 84 . https://doi.org/10.1016/j.molp. 2015 . 04 .007 PMID: 25917172
31. Vazquez-Vilar M , Bernabe?-Orts JM , Fernandez-del-Carmen A , Ziarsolo P , Blanca J , Granell A , et al. A modular toolbox for gRNA-Cas9 genome engineering in plants based on the GoldenBraid standard . Plant Methods . 2016 ; 12 ( 1 ): 10 .
32. Lee K , Zhang Y , Kleinstiver BP , Guo JA , Aryee MJ , Miller J , et al. Activities and specificities of CRISPR/ Cas9 and Cas12a nucleases for targeted mutagenesis in maize . Plant Biotechnol J . 2018 ; 1 - 11 .
33. Moreno-Mateos MA , Fernandez JP , Rouet R , Vejnar CE , Lane MA , Mis E , et al. CRISPR-Cpf1 mediates efficient homology-directed repair and temperature-controlled genome editing . Nat Commun . 2017 ; 8 ( 8 ): 2024 .
34. Li Z , Liu Z , Xing A , Moon BP , Koellhoffer JP , Huang L , et al. Cas9 -Guide RNA Directed Genome Editing in Soybean . Plant Physiol . 2015 ; 169 ( 2 ): 960 - 70 . https://doi.org/10.1104/pp. 15 .00783 PMID: 26294043
35. Lawrenson T , Shorinola O , Stacey N , Li C , ?stergaard L , Patron NJ , et al. Induction of targeted, heritable mutations in barley and Brassica oleracea using RNA-guided Cas9 nuclease . Genome Biology . 2015 ; 30 ( 16 ): 258 .
36. Zhang Q , Xing HL , Wang ZP , Zhang HY , Yang F , Wang XC , et al. Potential high-frequency off-target mutagenesis induced by CRISPR/Cas9 in Arabidopsis and its prevention . Plant Mol Biol . 2018 ; 96 ( 4- 5 ): 445 - 56 . https://doi.org/10.1007/s11103-018-0709 -x PMID : 29476306
37. Zhang D , Wang Z , Wang N , Gao Y , Liu Y , Wu Y , et al. Tissue culture-induced heritable genomic variation in rice, and their phenotypic implications . PLoS One . 2014 ; 9 ( 5 ):e96879. https://doi.org/10.1371/ journal.pone. 0096879 PMID: 24804838
38. Phillips RL , Kaeppler SM , Olhoft P . Genetic instability of plant tissue cultures: breakdown of normal controls . Proc Natl Acad Sci U S A . 1994 ; 91 ( 12 ): 5222 - 6 . PMID: 8202472
39. Slaymaker IM , Gao L , Zetsche B , Scott DA , Yan WX , Zhang F. Rationally engineered Cas9 nucleases with improved specificity . Science . 2016 ; 351 ( 6268 ): 84 - 8 . https://doi.org/10.1126/science.aad5227 PMID: 26628643
Patron NNJ , Orzaez D , Marillonnet S , Warzecha W , Matthewman C , Youles M , et al. Standards for Plant Synthetic Biology: A Common Syntax for Exchange of DNA Parts . New Phytol. 2015 ; 208 ( 1 ): 13 - 9 . https://doi.org/10.1111/nph.13532 PMID: 26171760
Yoo SD , Cho YH , Sheen J . Arabidopsis mesophyll protoplasts: A versatile cell system for transient gene expression analysis . Nat Protoc . 2007 ; 2 ( 7 ): 1565 - 72 . https://doi.org/10.1038/nprot. 2007 .199 PMID: 17585298
Pinello L , Canver MC , Hoban MD , Orkin SH , Kohn DB , Bauer DE , et al. Analyzing CRISPR genomeediting experiments with CRISPResso . Nat Biotechnol . 2016 ; 34 ( 7 ): 695 - 7 . https://doi.org/10.1038/nbt. 3583 PMID: 27404874
Chen B , Gilbert L a , Cimini B a , Schnitzbauer J , Zhang W , Li G-W , et al. Dynamic imaging of genomic loci in living human cells by an optimized CRISPR/Cas system . Cell . 2013 ; 155 ( 7 ): 1479 - 91 . https://doi. org/10.1016/j.cell. 2013 . 12 .001 PMID: 24360272
Wang Z-P , Xing H , Dong L , Zhang H , Han C , Wang X , et al. Egg cell-specific promoter-controlled CRISPR/Cas9 efficiently generates homozygous mutants for multiple target genes in Arabidopsis in a single generation . Genome Biology ; 2015 ; 16 ( 1 ): 144 .
Xu R , Li H , Qin R , Wang L , Li L , Wei P , et al. Gene targeting using the Agrobacterium tumefaciensmediated CRISPR-Cas system in rice . Rice . 2014 ; 7(1):5 . https://doi.org/10.1186/s12284-014 -0005-6 PMID: 24920971
Zhou J , Wang G , Liu Z. Efficient genome editing of wild strawberry genes, vector development and validation . Plant Biotechnology Journal . 2018 ; 16 ( 11 ) 1868 - 1877 . https://doi.org/10.1111/pbi.12922 PMID: 29577545
Brooks C , Nekrasov V , Lippman ZB , Van Eck J , Eck J Van. Efficient Gene Editing in Tomato in the First Generation Using the Clustered Regularly Interspaced Short Palindromic Repeats/CRISPR-Associated9 System . Plant Physiol . 2014 ; 166 ( 3 ): 1292 - 7 . https://doi.org/10.1104/pp. 114 .247577 PMID: 25225186
Jia H , Wang N. Targeted genome editing of sweet orange using Cas9/sgRNA . PLoS One . 2014 ; 9 ( 4 ): e93806. https://doi.org/10.1371/journal.pone. 0093806 PMID: 24710347
Liang Z , Chen K , Zhang Y , Liu J , Yin K , Qiu JL , et al. Genome editing of bread wheat using biolistic delivery of CRISPR/Cas9 in vitro transcripts or ribonucleoproteins . Nat Protoc . 2018 ; 13 ( 3 ): 413 - 30 . https://doi.org/10.1038/nprot. 2017 .145 PMID: 29388938
Char SN , Neelakandan AK , Nahampun H , Frame B , Main M , Spalding MH , et al. An Agrobacteriumdelivered CRISPR/Cas9 system for high-frequency targeted mutagenesis in maize . Plant Biotechnol J . 2017 ; 15 ( 2 ): 257 - 68 . https://doi.org/10.1111/pbi.12611 PMID: 27510362
Yan L , Wei S , Wu Y , Hu R , Li H , Yang W , et al. High-Efficiency Genome Editing in Arabidopsis Using YAO Promoter-Driven CRISPR/Cas9 System . Mol Plant . 2015 ; 8 ( 12 ): 1820 - 3 . https://doi.org/10.1016/j. molp. 2015 . 10 .004 PMID: 26524930
Xu P , Su H , Chen W , Lu P. The Application of a Meiocyte-Specific CRISPR/Cas9 (MSC) System and a Suicide-MSC System in Generating Inheritable and Stable Mutations in Arabidopsis . Front Plant Sci . 2018 ; 9 (July): 1 - 12 . https://doi.org/10.3389/fpls. 2018 .00001
Doench JG , Hartenian E , Graham DB , Tothova Z , Hegde M , Smith I , et al. Rational design of highly active sgRNAs for CRISPR-Cas9-mediated gene inactivation . Nat Biotechnol . 2014 Sep 3 ; 32 : 1262 - 7 . https://doi.org/10.1038/nbt.3026 PMID: 25184501
60. Xu H , Xiao T , Chen CH , Li W , Meyer CA , Wu Q , et al. Sequence determinants of improved CRISPR sgRNA design . Genome Res . 2015 ; 25 ( 8 ): 1147 - 57 . https://doi.org/10.1101/gr.191452.115 PMID: 26063738
61. Doench JG , Fusi N , Sullender M , Hegde M , Vaimberg EW , Donovan KF , et al. Optimized sgRNA design to maximize activity and minimize off-target effects of CRISPR-Cas9 . Nat Biotechnol . 2016 ; 34 ( 2 ): 184 - 91 . https://doi.org/10.1038/nbt.3437 PMID: 26780180
62. Gagnon JA , Valen E , Thyme SB , Huang P , Ahkmetova L , Pauli A , et al. Efficient mutagenesis by Cas9 protein-mediated oligonucleotide insertion and large-scale assessment of single-guide RNAs . PLoS One . 2014 ; 9 ( 8 ): e106396 .
63. Wang Y , Liu KI , Sutrisnoh NAB , Srinivasan H , Zhang J , Li J , et al. Systematic evaluation of CRISPRCas systems reveals design principles for genome editing in human cells . Genome Biol . 2018 ; 19 ( 1 ): 62 . https://doi.org/10.1186/s13059-018-1445 -x PMID : 29843790
64. Kohli A , Gonza? lez -Melendi P , Abranches R , Capell T , Stoger E , Christou P. The quest to understand the basis and mechanisms that control expression of introduced transgenes in crop plants . Plant Signaling and Behavior . 2006 ; 1 ( 4 ): 185 - 95 . PMID: 19521484
65. Horlbeck MA , Witkowsky LB , Guglielmi B , Replogle JM , Gilbert LA , Villalta JE , et al. Nucleosomes impede cas9 access to DNA in vivo and in vitro. Elife . 2016 ; 5:e12677 . https://doi.org/10.7554/eLife. 12677 PMID: 26987018
66. Liu M-J , Seddon AE , Tsai ZT -Y, Major IT , Floer M , Howe GA , et al. Determinants of nucleosome positioning and their influence on plant gene expression . Genome Res . 2015 ; 25 : 1182 - 95 . https://doi.org/ 10.1101/gr.188680.114 PMID: 26063739
67. Slaymaker IM , Gao L , Zetsche B , Scott DA , Yan WX , Zhang F. Rationally engineered Cas9 nucleases with improved specificity . Science . 2015 ; 10 ( 6268 ): 1126 .
68. Zhang D , Zhang H , Li T , Chen K , Qiu JL , Gao C . Perfectly matched 20-nucleotide guide RNA sequences enable robust genome editing using high-fidelity SpCas9 nucleases . Genome Biol . 2017 ; 18 ( 1 ): 191 . https://doi.org/10.1186/s13059-017 -1325-9 PMID: 29020979