Production of knockout mice by DNA microinjection of various CRISPR/Cas9 vectors into freeze-thawed fertilized oocytes

Construction of an all-in-one vector system for FokI-dCas9-mediated genome editing

We previously established a system for creating all-in-one CRISPR/Cas9 vectors expressing
Cas9 nuclease or nickase and up to seven gRNAs 23], and have distributed these plasmids as a “Multiplex CRISPR/Cas9 Assembly System
Kit” via Addgene (Kit #1000000055). To expand the system for FokI-dCas9-mediated genome
editing, we first created pX330A_FokI-1x(2–7) vectors. Using these newly constructed
plasmids and pX330s-(2–7) vectors included in the Addgene Kit, we can create an all-in-one
vector for FokI-dCas9-mediated genome editing, not only for single but also for multiplex
gene editing (Figure 1). The new vectors will expectedly be distributed by Addgene as a supplemental package
of the current kit (Multiplex CRISPR dCas9/FokI-dCas9 Accessory Pack).

Figure 1. Schematic illustration of the all-in-one FokI-dCas9 vector construction system. Single
gRNA-expressing vectors are constructed by the insertion of annealed oligonucleotides.
Subsequently, the gRNA cassettes are assembled using Golden Gate cloning. Amp, ampicillin;
Spec, spectinomycin; U6, human U6 promoter; CBh, chicken beta-actin short promoter.

Design of CRISPR/Cas9 nuclease, Cas9 nickase, and FokI-dCas9 vectors to target the
interleukin-11 (IL11) gene

To compare the applicability of the three CRISPR/Cas9 systems with Cas9 nuclease,
Cas9 nickase, or FokI-dCas9 in mouse genome editing, we designed and constructed CRISPR/Cas9
vectors as shown in Figure 2. Although the three systems require different design parameters and it is difficult
to compare the genome editing efficacy with exactly the same gRNAs, we tried to target
almost the same genomic region, which is around exon 3 of the IL11 gene. For Cas9 nuclease, two gRNAs, gRNA_A and gRNA_B, were designed within the exon
and oligonucleotides for these gRNAs were cloned separately (Figure 2A; Nuclease_A and Nuclease_B vectors). For Cas9 nickase, gRNA_B and gRNA_C were designed
with a 7-bp offset (Figure 2B: Nickase_BC vector). When inducing double nick-mediated genome modification, the
optimal range of the offset length is around 0–10 bp 23], whereas FokI-dCas9 requires 13- to 18-bp offsets for highly-efficient targeted mutagenesis
14]. Thus, we designed another gRNA, gRNA_D, in combination with gRNA_B for FokI-dCas9
(Figure 2C; FokI-dCas9_BD vector).

Figure 2. The target sequence and the constructed vectors. The genomic region around exon 3
of the IL11 gene was targeted by Cas9 nuclease (A), Cas9 nickase (B), or FokI-dCas9 (C) vectors. The target sequence of each gRNA is indicated by colored bases. The PAM
sequence is underlined.

Activity validation of the all-in-one FokI-dCas9 vector

Since the genome editing potential of the all-in-one vector containing the FokI-dCas9
cassette was unknown, we evaluated the activity of the FokI-dCas9_BD vector in mouse
embryos. The FokI-dCas9_BD vector was microinjected into the pronucleus of mouse fertilized
oocytes. Twenty-one survival oocytes were cultured for 3.5 days. We observed seven
blastocysts, and six of these embryos developed into expanded blastocysts. Each blastocyst
was collected into a microtube, and then PCR was performed to amplify the targeted
locus. We verified that at least one mutation had been induced in three embryos by
direct sequencing (Figure 3A). In addition, bacterial cloning of the PCR products followed by DNA sequencing
revealed high mutation rates (?50%) (Figure 3B), confirming the powerful potential of the all-in-one FokI-dCas9 vector for GEM
generation.

Figure 3. Sequence analysis of blastocysts injected with the FokI-dCas9_BD vector. The PCR products
were analyzed by direct sequencing (A), followed by sequencing of subcloned plasmids (B). The wild-type IL11 sequence is shown at the top (Wild) with the gRNA target sequences (underlined).
The PAM sequence is enclosed in red boxes. Deletions are indicated by dashes and substitutions
and insertions are enclosed in a black box. A blue box indicates the partial sequence
of exon 3. Only one or two types of mutations were detected in direct sequencing (A), whereas various mutation patterns were observed in subcloned sequencing (B). Numbers in red letters on the right side of each sequence indicate the frequencies
of the corresponding alleles (number of the allele/number of total clones).

GEM generation by microinjection of all-in-one CRISPR/Cas9 plasmids into freeze-thawed
fertilized oocytes

We next conducted microinjection of the three types of CRISPR/Cas9 vectors into freeze-thawed
fertilized oocytes for efficient and convenient genome engineering in mice. Using
reproductive engineering techniques, such as IVF and freeze-thawing, fertilized oocytes
could be obtained with high efficiency (Additional file 1). Each CRISPR/Cas9 vector was directly injected into the pronucleus of freeze-thawed
fertilized oocytes prepared by mating or IVF. The survival oocytes after microinjection
were transferred into the oviducts of pseudopregnant ICR female mice. All the surrogate
female mice gave birth to pups (Table 1), and all the pups excluding one pup died from cannibalism were genotyped by direct
sequencing analysis (Table 1 and Additional file 2). Although we have shown that the combinatorial screening strategy is required for
complete genotyping 28], direct sequencing is reportedly able to screen mutant mice with high mutation rates
7].

Table 1. Generation ofIL11mutants using freeze-thawed fertilized oocytes

Sequencing analysis revealed that all three types of CRISPR/Cas9 vectors could produce
mutant mice, but the birth rates and mutation rates of each method were not comparable.
In the Cas9 nuclease group, the birth rates were relatively low (7.4–12.0%), whereas
the mutation rates were high (80–100%). Conversely, more pups were born in the Cas9
nickase group (19.1% and 22.2%), while the number of mutants were fewer than in the
Cas9 nuclease group (5.9% and 0%). Interestingly, the FokI-dCas9 group exhibited moderate
birth rates (18.2% and 17.6%) and mutation rates (66.6% and 33.3%). To confirm the
superiority of FokI-dCas9-mediated method, we additionally injected the FokI-dCas9_BD
vector into freeze-thawed fertilized oocytes collected from mature female mice (8 weeks
of age) mated with male mice. Consistent with the results using the oocytes collected
from immature female mice (5 weeks of age), the birth rate and the mutation rate were
sufficiently high (38.2% and 84.6%, respectively). These results suggest not only
the robustness of FokI-dCas9-mediated mouse genome editing, but also the applicability
of oocytes from female mice at various weeks of age, as shown previously using TALENs
24]. Although these results do not necessarily reflect the general properties of the
three CRISPR/Cas9 systems, they represent an important practical example of CRISPR/Cas9-mediated
GEM generation. Importantly, the low birth rates of Cas9 nuclease plasmid-injected
mice are comparable with previous studies using inbred strains 29],30], although high birth rates can be achieved using B6D2F1 (BDF1) hybrid mice 7]. We thus concluded that FokI-dCas9 plasmid injection might be a powerful strategy
for generating knockout mice, especially for inbred strains.

Transgene analysis and off-target analysis

Since plasmid DNA was used for microinjection, we investigated whether DNA vectors
were integrated into the genome. Genomic PCR of the coding sequence of Cas9 variants
and FokI was performed to detect genomic integrants. In the two founders generated
by the injection of Nuclease_B vector, the Cas9 gene fragment was amplified (Additional
file 3). Altogether, 10% (2/20) of the mutant mice carried the transgene. Although the observed
transgenicity is slightly higher than in Mashiko’s report (4.3%) 7], this may not be a major issue because the transgene can be removed by mating before
expansion of the mutant strain.

Finally, potential off-target sites for each gRNA target site were analyzed by direct
sequencing. The top three candidates for each site were selected using the CRISPR
design tool (http://crispr.mit.edu/) (Additional file 4). We then sequenced all of the 20 founders (seven founders for Nuclease_A, six founders
for Nuclease_B, one founder for Nickase_BC, and six founders for FokI-dCas9_BD), but
no off-target mutations were detected. However, our analysis and all of the previously
reported off-target analyses in mice provide limited genomic context, and the potential
risk of off-target mutagenesis in disparate parts of the genome is undeniable. In
fact, Tsai and colleagues recently revealed that CRISPR/Cas9 can induce various genomic
rearrangements including megabase-scale large deletions and chromosomal translocations
in cultured cells 31]. They also showed that most of these experimentally demonstrated off-target sites
were not identified as off-target candidates using any computational prediction tool
including the CRISPR design tool. Also in mice, these concealed off-target effects
can exist, but they are difficult to detect because they cannot be predicted by in silico analysis. We thus believe that a highly specific gene targeting strategy, such as
FokI-dCas9, is important not only for cultured cell applications but also for the
creation of GEM.

Multiplex genome editing in mouse embryos using a single all-in-one FokI-dCas9 vector

We finally examined whether multiple gene targeting could be applied using microinjection
of a single all-in-one FokI-dCas9 vector simultaneously expressing four gRNAs and
FokI-dCas9. We designed gRNAs to target exon 3 of regenerating islet-derived 3 beta
(Reg3b) and regenerating islet-derived 3 gamma (Reg3g) genes, which are located in the chromosome 6 with 96.1-kb distance (Figure 4A and B). The all-in-one FokI-dCas9 vector was microinjected into the pronucleus of
fertilized oocytes. Twenty-eight survival oocytes were cultured for 3.5 days, and
nine embryos developed into expanded blastocysts. Each blastocyst was collected into
a microtube, and then PCR was performed to amplify around the both target loci. Direct
sequencing analyses revealed that more than 50% of the embryos possessed mutated alleles
in the both target loci (8/9 in Reg3b and 5/9 in Reg3g) (Figure 5A and B). More importantly, five embryos (#1, 3, 6, 7, and 8) possessed mutations
in the both genes, showing high potential for creating double knockout mice mediated
by microinjection of all-in-one FokI-dCas9 vector. We further performed out-out PCR
to investigate whether chromosomally deleted alleles exist, but no intended amplicons
were observed using blastocyst PCR genotyping. Further examinations are needed to
clarify the possible existence of chromosomally deleted alleles in pups.

Figure 4. Schematic design of multiple gene targeting using all-in-one FokI-dCas9 vector. (A) The genomic context around the targeted loci. The target sites are located in exon
3 of Reg3b and Reg3g genes. (B) The target sequence and the constructed vectors. The target sequence of each gRNA
is indicated by colored bases. The PAM sequence is underlined.

Figure 5. Sequence analysis of blastocysts injected with all-in-one FokI-dCas9 vector simultaneously
targeting Reg3b and Reg3g genes. The PCR products were analyzed by direct sequencing to identify mutations
in Reg3b(A) and Reg3g(B) loci. The wild-type sequence of Reg3b and Reg3g are shown at the top (Wild) with the gRNA target sequences (underlined). The PAM
sequence is enclosed in red boxes. Deletions are indicated by dashes. Blastocyst numbers
on the left side of each sequence in (A) and (B) are identical. Most blastocysts had multiple types of mutations at the Reg3b locus. The wild-type Reg3b sequence was not detected in blastocyst #4 and #6. In the other blastocysts, the
wild-type Reg3b sequence was detected along with mutant sequence. Regarding Reg3g locus, the wild-type sequence was detected in all blastocysts with mutant sequence.