Genome Editing in Mice Using CRISPR/Cas9 Technology

Abstract

CRISPR-Cas9 technology has revolutionized genome editing in mice, allowing for simple and rapid development of knockouts and knockins. CRISPR relies on small guide RNAs that direct the RNA guided nuclease Cas9 to a designated genomic site using ~20bp of corresponding sequence. Cas9 then creates a double strand break in the targeted loci that is either patched in an error-prone fashion to produce a frame-shift mutation, a knockout, or is repaired by recombination with donor DNA containing homology arms, a knockin. This protocol covers the techniques needed to rapidly generate knockout and knockin mice with CRISPR via microinjection of Cas9, the guide RNA, and possible donor DNA into the mouse zygote.

INTRODUCTION

CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats) is a RNA guided means of genome editing that, in conjunction with pronuclear microinjection, has provided an efficient and simple means to create genetically engineered mice. CRISPR technology allows for the targeted introduction of a double strand break (DSB) within a designated location in the mouse genome. Repair of the DSB in the mouse zygote either results in a knockout mouse through introduction of an indel mutation (small insertion or deletion) by Non-Homologous End Joining (NHEJ), or, if a donor repair DNA template is present, the production of a knockin mouse through Homology Directed Repairs (HDR), where point mutations, loxP sites, protein tags, etc. can be inserted into the genome. Genome editing to make a knockout mouse using CRISPR just requires two components to be introduced into a donor mouse zygote: the Cas9 (CRISPR-associated) DNA endonuclease and a single guide RNA (sgRNA) to direct the introduction of a DSB at a selected site. Alternatively, creating a knockin mouse requires the addition of a donor DNA to serve as a template to repair the DSB and introduce a designated mutation.

Genetically engineered mice are a valuable research tool for understanding mammalian biology. Mice breed rapidly, are easy to house, have short lifespans, and share about 95% of genes with humans, all of which make the laboratory mouse an ideal animal model to study human diseases. Genome editing in mice originally relied on the random integration of a targeting vector in embryonic stem (ES) cells and subsequent injection of targeted clones into donor mouse blastocysts (Hall et al., 2009; Limaye et al., 2009; Longenecker and Kulkarni, 2009). Later, zinc finger technology (ZFNs) and Transcription activator-like effector nucleases (TALENs) were developed that allowed for genome editing by using DNA binding proteins that are fused to the Fok I nuclease. With ZFNs and TALENS, modules of DNA binding protein domains are essentially pieced together to target a specified region within the mouse genome. A pair of these proteins are needed to create a DSB, which would either result in a gene knockout from NHEJ or, when donor DNA is present, lead to HDR and result in a knockin mouse. ZFNs and TALENs accelerated genome engineering when compared to the inefficient rate of random homologous recombination in ES cells. ZFNs and TALENs provided one-step genome editing requiring only microinjection of their respective mRNAs into a donor zygote, analogous to making a transgenic mouse. Genome editing at the zygote stage, thereby, allowed for rapid generation of genetically engineered mice, while also avoiding a reliance on the germline competency of ES cells (Singh et al., 2015). These gene targeting techniques, along with the eventual development of CRISPR-Cas9 technology, reduced the time to develop genetically engineered mice from 8–10 months to about 3 months. CRISPR-Cas9 technology, however, has essentially supplanted ZFNs and TALENs in terms of ease of design and construction (Fig. 1). ZFNs and TALENs require the assembly of multimeric DNA binding domains for each new genomic target. CRISPR, in contrast, is an RNA-guided means of genome editing. CRISPR relies on the formation of a ribonucleoprotein complex between the DNA endonuclease Cas9 and a “guide” RNA that directs where a genomic DSB occurs using about 20bp of matching complementary sequence. A single Cas9 nuclease can then target multiple genetic loci by simply switching this guide RNA sequence to designate a different site. The components needed for CRISPR genome editing are inexpensive, easy to generate, and are designed around simple base-pair binding (Sander and Joung, 2014).

Figure 1. Gene Editing Strategy Using CRISPR/Cas9.

A) To conduct CRISPR mediated genome editing in a mouse, the Cas9 DNA endonuclease and an accompanying sgRNA (single guide RNA) designed with a specified targeting sequence are coinjected into the mouse zygote pronucleus. B) Cas9 and the sgRNA will subsequently combine to form a ribonucleotide particle. Upon recognition of the corresponding matching sequence within the mouse genome, Cas9 will then create a double stranded break (DSB). C) The resulting DSB is typically repaired through Non-Homologous End Joining (NHEJ). NHEJ often results in an indel (insertion/deletion) mutation that will cause a frameshift. Alternatively, the DSB can be fixed using Homology Directed Repair (HDR) if donor DNA is present, although at a lower frequency than NHEJ. The donor DNA is designed with enough homology (flanking homology arms) to trick the normal cellular repair machinery to use it as a template to fix the DSB. Large insertions or deletions, however, may require two sgRNAs to completely span the length of genomic DNA to be either replaced or removed.

CRISPR was first identified as a bacterial defense system against the intrusion of foreign DNA. CRISPR arrays in bacteria essentially consist of a series of palindromic repeats separated by unique spacers (Lander, 2016; Jinek et al., 2012; Gasiunas et al., 2012). The spacers were later identified to contain DNA corresponding to bacteriophages and plasmid genomes. The purpose of CRIPSR was then determined to work as an adaptive immune response in bacteria and archaea to protect against future bacteriophage infection (Barrangou et al. 2007). Essentially, CRISPR is a heritable record of past bacteriophage encounters that were adapted to provide a RNA guided means of re-identifying and silencing subsequent viral infections. The transcribed RNA from the CRISPR array then works in tandem with a Cas DNA endonuclease, also encoded on the CRISPR locus, to target and cleave foreign DNA. Two classes of CRISPR-Cas systems have been identified in bacteria and archea, with further divisions into multiple types and subtypes (Shmakov et al., 2017). Most known CRISPR-Cas loci are Class 1, which require a multi-Cas protein complex for site specific DNA silencing (Shmakov et al., 2017). Class 2 CRISPR-Cas systems, however, utilize only a single multidomain protein. Class 2 CRISPR-Cas systems include Cas9, the most commonly used DNA endonuclease for genome editing, as well as Cpf1, another class 2 CRISPR effector protein that produces a staggered cut rather than a blunt-ended DSB with Cas9 (Zetsche et al., 2015). This protocol will primarily focus on Cas9, which is derived from the class 2 CRISPR-Cas system of Streptococcus pyogenes.

Because of its potential as a programmable RNA guided endonuclease, Cas9 was later adapted for genetic engineering in mammalian cells (Cong et al., 2013; Mali et al., 2013). In bacteria, the CRISPR array of palindrome repeats and spacers are transcribed to form pre-CRISPR RNA (crRNA). The pre-crRNA binds to trans-activating crRNA (tracrRNA) to be subsequently processed by an RNase III. The resulting mature crRNA/tracrRNA hybrid then complexes with Cas9 for targeted gene silencing (Jinek et al., 2012). When applied for genome editing in eukaryotic cells, Cas9 was modified to contain a nuclear localization signal (NLS) and a synthetic single chimeric crRNA-tracrRNA, termed single guide RNA (sgRNA), was used to direct genome targeting. To design sgRNA, the recognition site selected for targeted DNA cleavage should consist of a unique genomic sequence of about 20 base pairs in length followed immediately by the Protospacer Adjacent Motif (PAM) that corresponds to the designated CRISPR-Cas system. With Cas9, the PAM is NGG, with a recognition sequence basically of 5’-N₂₀NGG-3’ (Sander and Joung, 2014). Upon target recognition, Cas9 has two nuclease domains, RuvC and HNH (histidine-asparagine-histidine endonuclease domain), to then generate a blunt-ended DSB in the DNA about 3 bp upstream of the PAM site (Yu et al., 2015). After generation of a DSB, the DNA is typically repaired through the error-prone process of NHEJ. Even if NHEJ does not initially produce DNA damage, the DSB repair process will continue until small insertions or deletions (indels) prevent further target recognition (Renaud et al., 2016). For genome editing in mammalian cells, these indels will then knockout a gene through a frame shift mutation. In contrast, if donor DNA is present, the DSB can be corrected with HDR, although at a lower frequency than NHEJ. CRISPR-Cas9 was later adapted for genetic manipulation in whole animals, particularly to generate knockout and knockin mice.

Site specific genome editing in mice using CRISPR was first achieved by disrupting an EGFP transgene (Shen et al., 2013), which moved the capabilities of CRISPR beyond just genome editing in vitro towards making actual in vivo animal models. Wang et al. (2013) advanced CRISPR technology even further by targeting first a single, then multiple genes in a mouse. A double-gene mutant mouse was possible by injecting two sgRNAs, where transcribed Cas9 endonucleases are directed to cleave both targeted alleles. So, along with the ease of switching sgRNA to target new alleles, CRISPR-Cas9 can also provide a tool for multiple genome editing when provided more than one sgRNA. This means of generating mutations in multiple genes in mice was difficult to achieve with other genome editing tools. The combinatorial effects of mutating more than one gene could then be analyzed in a mouse, which provides a better means to mimic many human polygenic diseases. The pups produced by zygotic microinjection of Cas9 and sgRNA developed normally, which suggests a low level of toxicity for Cas9 (Wang et al., 2013). CRIPSR mediated gene knockout mice were also shown to faithfully replicate the phenotypes seen using previous gene targeting techniques in ES cells. CRISPR generated mutations are additionally heritable with transmission to the next generation (Li et al., 2013). Moving beyond just knockout mice, Wang et al. also microinjected a single-stranded donor DNA (ssDNA) oligo in conjunction with Cas9 and sgRNA to promote HDR and develop knockin mice with multiple point mutations. After proving HDR mediated genome editing in mice, Yang et al. (2013) further showed that CRISPR could be programed to insert protein tags, fluorescent reporters, and loxP sites into targeted alleles. With point mutations or short epitope tags, only a single stranded donor oligo was required, but a circular plasmid is needed for insertion of a larger fluorescent reporter. Along with protein tags, conditional knockout alleles were additionally shown to be possible using CRISPR via injection of two sgRNAs along with the accompanying oligos to insert loxP sites to flank a targeted exon. NHEJ, however, can occur using two sgRNAs to result in a large deletion spanning both DSB sites.

With the advent of CRISPR technology for genome editing in mice, new efforts have been made to improve the efficiency and specificity of gene targeting. CRISPR has been adapted to generate large genomic deletions that are needed to knockout gene clusters or an allele containing multiple splice variants (Zhang et al., 2015). Along with large deletions, large genomic insertions have additionally shown to be possible using CRISPR (Zhang et al., 2015). Modifications to the sgRNA design have been tested in order to improve genome editing efficiency, including adjustment of the nucleotide length, extension of the hairpin, additional tracrRNA, etc. (Fujii et al., 2013; Chen et al., 2013; Hsu et al., 2013). Injection of dual sgRNAs has additionally been suggested as means to increase the likelihood of the gene targeting, particularly if bi-allelic gene modification is desired (Zhou et al., 2014). A problem with CRISPR mediated genome editing, however, is the possibility of off-target DNA damage, particularly since Cas9 can tolerate up to five mismatches in the targeting sequence of the sgRNA, especially within the 5’ region (Fu et al., 2014). Off-target mutations have been mostly described for CRISPR genome editing in human cancer cells but may be rare in a mouse zygote (Fu et al., 2013; Yang et al., 2013). Nevertheless, off-target mutations are still a concern when analyzing genetically engineered mice. An alternative to using the wild-type Cas9 involves a D10A mutation in the RuvC nuclease domain to create a Cas9 nickase (Cas9n), which has been used to enhance genome specificity through RNA-guided double nicking strategy (Cong et al., 2013). A single Cas9n cut results in only a single strand break that is easily repaired; so, to create an actual DSB, two sgRNAs are needed. Both pairs of sgRNAs need to work in conjunction to produce a DSB where one sgRNA directs Cas9n to create a single stranded nick at a designated site on one DNA strand while the second sgRNA leads to a nearby nick on the opposite strand (Ran et al., 2013a; Shen et al., 2013; Fujii et al., 2014).

The following protocol chapter includes a basic template for generating genetically engineered mice with CRISPR technology, but also provides a current overview of variations on CRISPR mediated genome editing cited in the literature (Fig. 2). Modifications to CRISPR include different styles of guide RNAs, various methods of Cas9 delivery, and chemical means of enhancing genome editing. The techniques used in CRISPR genome editing essentially overlap with methods needed for making transgenic mice, including zygote collection, microinjection of nucleic acid into the mouse zygote, and implantation of manipulated embryos into pseudo-pregnant females. Anyone trained in generating transgenic mice (insertion of exogenous DNA) can also make other genetically engineered mice, such as knockouts (gene disruption) and knockins (gene modification), by using this protocol (Harms et al., 2014). This unit will provide insights and protocols into developing knockout and knockin mouse models using CRISPR-Cas9 technology and will provide an update on the current methodologies needed to generate genetically engineered mice. Basic protocol 1 describes different methods of single guide RNA (sgRNA) synthesis, which is the basis of generating genetically engineered mice through CRISPR. Basic protocol 2 details validation of sgRNA efficiency. Basic protocol 3 explains microinjections of Cas9 and sgRNA into the mouse zygote for derivation of knockout mice by NHEJ. Basic protocol 4 provides advice on donor DNA design needed for generation of knockin mice by HDR. Basic protocol 5 describes the genotyping methods to identify genetically engineered mice.

Figure 2. Generating a Knockout or Knockin Mouse through CRISPR.

Flowchart illustrating from left to right the steps needed to conduct genome editing using CRISPR.

BASIC PROTOCOL 1

SINGLE GUIDE RNA (sgRNA) CONSTRUCTION

A crucial element determining the efficiency of genome editing is the design of the sgRNA. Optimizing the design and specificity of the sgRNAs can also reduce the chance of off-target integration. Cas9, which is derived from Streptococcus pyogenes, generally requires a target sequence of 5’-N₂₀NGG-3’, where NGG is the PAM needed for proper site recognition. Alternate PAM sites including NAG have also been identified, although with lower efficiency (Sander and Joung, 2014). The 20nt target recognition sequence needs to be unique and a whole genome-wide homology search should be conducted to ensure specificity (Wiles et al., 2015). Different configurations of chimeric sgRNA designs have been suggested in the literature, but most are about 100 nucleotides in length. Numerous CRISPR design tools are available on-line, many of which have been reviewed in Wiles et al. (2015). Based on our experience, we will primarily discuss aspects of selecting sgRNAs using the CRISPR design tool (http://crispr.mit.edu/) from the laboratory of Dr. Feng Zhang at MIT (Hsu et al., 2013). The MIT CRISPR software was one of the first programs available for sgRNA design and has been widely used and cited in the literature (Harms et al., 2014), but newer tools are also available that may better predict off-targets sites (Haeussler et al., 2016). Once a site has been designated for targeted genomic editing, 250–500 bp of mouse genomic sequence is entered for sgRNA design. As with conventional gene targeting, the location of genome editing must be carefully considered. To generate a complete knockout, for example, an essential exon should be selected for insertion of indels that will not be bypassed during splicing. After entering the genomic sequence, a list of sgRNAs will be provided along with an efficiency score and possible off-target sites for each sgRNA. The sgRNAs are ranked and scored according to ‘on target’ activity. Typically, the sgRNA requires perfect base-paring 10–12bp directly proximal to the PAM (the ‘seed’ sequence), but tolerates more distal mismatches (Sander and Joung, 2014; Hsu et al., 2013). The orientation of the target sequence to match either the sense or anti-sense strand will not matter as long as the selected 20 bp sequence is followed by the required 3’ PAM site. G-C content should be between 40–60% in total with more G-C content situated within the seed sequence if applicable (Low et al., 2016). Repetitive sequence and repeats of the same nucleotide in the target sequence should also be avoided. DNA accessibility, however, can additionally impact sgRNA efficiency. The rules determining sgRNA efficiency while avoiding off-target effects still need to be worked out further. Therefore, about 3–6 sgRNAs with the best efficiency scores should be selected for further analysis. Running an additional genome-wide BLAST search on the designated sgRNAs ensures that the designated sequence is unique within the given genome.

Method 1

For sgRNA design, the MIT CRISPR software has been widely used and cited, but other design tools are also available (see below). The following protocol for sgRNA synthesis requires two oligos to construct a PCR-based template for in vitro transcription (IVT) (Qin et al., 2016). The resulting sgRNA contains a conventional sgRNA backbone (Cong et al., 2013).

Materials

MIT on-line CRISPR tool (http://crispr.mit.edu/)

2 oligo primers for PCR-based generation of a sgRNA template for IVT (IDT or other provider for high quality oligos)

PCR kit - Accuprime Taq DNA polymerase system (Thermo Fisher Scientific, cat. no. 12339016)

PCR cleaning kit - QIAquick PCR purification kit (Qiagen, cat. no. 28104)

In vitro transcription (IVT) kit - MEGAshortscript T7 Transcription kit (Thermo Fisher Scientific, cat. no. AM1354)

In vitro transcription cleaning kit - MEGAclear Transcription Clean-up kit (Thermo Fisher Scientific, cat. no. AM1908)

PCR thermal cycler

0.2 ml thin-walled PCR tubes

2% TAE agarose gel and equipment for gel electrophoresis

Eppendorf Thermomixer R Mixer

Micro-volume spectrophotometer (DeNovix Spectrophotometer DS-11 or Thermo Fisher Scientific Nanodrop)

SgRNA synthesis using a PCR template for IVT

1. Specify a genomic region for introduction of a desired DSB. Regions with high GC content may not be suitable for both designing sgRNAs and selection of genotyping primer (Low et al., 2016; Qin et al., 2016). The site of the targeted DSB will depend on the type of mutation required, but listed are some suggestions:

   1. For a gene knockout, select an essential region of the coding sequence to be disrupted by NHEJ, typically an exon that encodes a critical protein domain. For a complete knockout, one should ensure that the site of DSB in the coding sequence is incorporated into all transcripts where the exon containing an indel mutation is not skipped over following splicing. Placement of the DSB should additionally not interfere with the splice donor and acceptor sites to ensure its incorporation into the transcript (Qin et al., 2016).

   2. With conditional gene knockouts, loxP sites should instead be placed within intronic non-coding sequence, so as not to disturb gene function in cells where targeted Cre recombination does not occur.

   3. For a knockin mouse, the DSB should be engineered as close as possible (<10–20 bp) to the planned insertion site of the desired mutation. This increases the chances for HDR. Point mutations, for example, require that the sgRNAs to target sequence near the selected amino acid substitution. With C-terminal protein tags, the DSB will clearly need to be located around the stop codon. The stop codon should be lost with insertion of coding sequence for the desired epitope tag. Alternatively, an IRES fluorescent protein tracer can be placed after the stop codon.

2. After selecting a site for targeted mutation, submit around 250–500 bp of the surrounding mouse genomic sequence into the MIT CRISPR design tool.

3. A list of potential sgRNAs are provided that are scored for the on-target potential and the likelihood of off-target binding. Select the 3 best sgRNA sequences with minimal off-target probability.

4. Order 2 oligo DNAs for each sgRNA synthesis. Oligo #1 is based on the selected sequence from the MIT CRISPR design tool. In addition, Oligo #1 also contains the T7 promoter for IVT. Oligo #2 is a common sequence that encodes the sgRNA backbone elements.

   OLIGO #1:

   5’gaaattaatacgactcactataggNNNNNNNNNNNNNNNNNNNNgttttagagc tagaaatagc 3’

   N is your specific 20 nucleotide guide RNA sequence upstream of NGG PAM sequence. Do not include NGG PAM to make sgRNA.

   OLIGO #2:

   5’aaaagcaccgactcggtgccactttttcaagttgataacggactagccttattttaacttgctatttctagctctaaaac 3’

   (common for all guide RNAs).

   Note: T7 requires two guanine residues at the transcription initiation site (in bold in OLIGO #1) before the target sequence (N). If the target sequence itself begins with 1 or 2 G’s, OLIGO #1 can be modified to remove these additional 1 or 2 guanine bases to improve later sgRNA efficiency.

5. Reconstitute the oligonucleotides to 10 μM (10 pmol/μl) with nuclease-free water

6. Perform PCR to anneal two oligo DNAs using the Accuprime kit (Fig. 3A). Set up the 50-μl PCR reaction mix as indicated below:

   • 43 μl nuclease free H₂O

   • 5 μl 10X Accprime PCR buffer I

   • 1 μl sgRNA forward primer

   • 1 μl sgRNA reverse primer

   • 0.1 μl Accuprime Taq DNA polymerase)

   PCR - 35 cycles:

   • 94°C 30 sec
   • 55°C 30 sec
   • 68°C 30 sec

   In this case, the sgRNA forward and sgRNA reverse primers are both the template and the primers.

7. Analyze about 1 μl of the PCR product on a 2% agarose gel to assess the yield of the PCR reaction. The product size should be about 124 nt in length.

8. Purify the remaining PCR product using the QIAquick kit, according to the manufacturer’s protocol. Elute the PCR template in 10 μl to ensure a concentrated amount of DNA for IVT (usually recover ~40 ng/μl).

9. Use the purified PCR template for IVT (MEGAShortScript T7) to derive the targeting sgRNA. Mix 8 μl of the purified PCR DNA (about 300 ng - 40ng/μl X 8 μl at a minimum) with 12 μl of the following IVT Master Mix:

   • 2 μl T7 10X Reaction buffer

   • 2 μl ATP solution (75 mM final)

   • 2 μl CTP solution (75 mM final)

   • 2 μl GTP solution (75 mM final)

   • 2 μl UTP solution (75 mM final)

   • 2 μl T7 Enzyme mix

   Incubate at 37°C for 0.5–4 h in a thermomixer.

   Optional: Add 1 μl of DNase and incubate at 37°C for 15 min to degrade the template DNA

10. Clean IVT product using MEGAclear kit (Fig. 3B) according to the manufacturer’s instructions as follows:

    Add 80 μl of the Elution Solution to the 20 μl IVT reaction and gently mix.

    Add 350 μl of binding solution concentrate and mix by pipetting.

    Add 250 μl of 100% ethanol and mix gently by pipetting.

    Add the mixture onto the filter cartridge and centrifuge (RCF of 10,000–15,000 × g).

    Discard flow-through.

    Wash twice with 500 μl of washing solution and centrifuge.

    Add 30 μl elution solution to column and incubate at 70°C for 10 min and centrifuge to collect elute.

11. Measure the concentration and purity of the single guide RNA (sgRNA) using a micro-volume spectrophotometer. After IVT and cleanup, the concentration should ideally be around 1 μg/μl. The 260/230 and 260/280 ratios should be around 2.0 for pure sgRNA. Store the sgRNA at −80 °C.

Figure 3. Preparation of the sgRNA (single guide RNA) for Gene Editing.

A) A gene-specific and a common oligo are needed to generate a template for in vitro transcription (IVT). The gene-specific oligo contains the 20 bp protospacer sequence (N’s) while the common oligo encodes the sgRNA backbone. The two oligos are annealed as shown to produce a DNA template for subsequent IVT. B) Gel picture depicting three different sgRNAs before and after RNA clean up. The IVT reaction uses the T7 promoter within the template DNA to generate a 124bp product.

Method 2

Alternatively, oligos can be ordered and subcloned into pX330, a sgRNA expression vector from the Feng Zhang lab available from Addgene (Cong et al., 2013). The sgRNAs can be expressed by a U6 promoter within the pX330 vector. The plasmid pX330 can also be used to co-express a human codon-optimized SpCas9. Once the sgRNA sequence is subcloned into pX330, the vector can either be directly microinjected into zygotes or be used for sgRNA synthesis using IVT. The cloning protocol is available on the Addgene website, but listed below is a brief synopsis of the procedure.

Materials

pX330-U6 Chimeric_BB-CBh-hSpCas9 (Addgene, cat. no. 42230)

2 Phosphorylated Oligo DNAs (IDT or other high-quality oligo provider)

BbsI (New England Biolabs, cat. no. R0539S)

Quick Ligase (New England Biolabs, cat. no. M2200S)

sgRNA synthesis by cloning into the pX330 CRISPR vector

1. Identify a region to target for a DSB and select sgRNAs for CRISPR genome editing using the MIT CRISPR software as listed above

2. After identifying 3 potential sgRNAs, two phosphorylated oligos need to be ordered to generate each sgRNA:

   5′ –CACCGNNNNNNNNNNNNNNNNNNN –3′

   5′ -AAACNNNNNNNNNNNNNNNNNNNC −3′

   When annealed, the oligos can be subcloned into pX330 into a BbsI site:

   5′ –CACCGNNNNNNNNNNNNNNNNNNN – 3′

   3′ –CNNNNNNNNNNNNNNNNNNNCAAA –5′

3. Digest pX330 with BbsI and gel purify the plasmid.

4. Ligate the annealed oligos into the BbsI digested pX330 and transform competent bacteria.

5. For IVT, a T7 promoter needs to be added using PCR amplification using the primers below (Yang et al., 2014). The PCR template is then used for IVT.

   T7-sgRNA F

   5′-ttaatacgactcactataggNNNNNNNNNNNNNNNNNNN-3′

   T7-sgRNA R

   5′-AAAAGCACCGACTCGGTGCC-3′

   The U6 and T7 promoters basically need a G or GG at the end, but this requirement may be bypassed by permitting mismatches, adding an additional G (21 bp guide sequence), or shortening the target recognition requirements (Ran et al., 2013b, Sander and Joung, 2014).

Method 3

The use of synthesized guide RNAs to conduct genome editing in mice has been validated using Easi-CRISPR (Efficient additions with ssDNA inserts-CRISPR), an adaption of traditional CRISPR techniques that may enhance the efficiency of getting a knockin allele (~ 25% to 67%) (Quadros et al., 2017). With Easi-CRISPR, chemically synthesized guide RNAs are acquired from IDT in the form of separate crRNA and tracrRNA. In this method, a PCR template for IVT is not necessary. Synthesis of the tracrRNA (cat. no. 1072532, 5 nmol scale) is invariant, while the crRNA is constructed using the variable 20 nt protospacer region.

Materials

Synthesized guide sgRNAs can also be purchased that are ready to use from IDT and other providers.

Easi-CRISPR method using synthetic guide RNAs

Resuspend each component in injection buffer to generate a 1 μg/μl solution, mix 5 μl of the crRNA solution with 10 μl of the tracrRNA solution, then anneal by heating at 95°C for 5 minutes, followed by cooling to 25°C at 5°C per minute as detailed in Quadros et al. (2017). Note, if using more than one targeting crRNA for genome editing, the crRNA/tracrRNA complexes should be formed separately.

Additional CRISPR sgRNA Design Software

ZiFiT - http://zifit.partners.org/ZiFiT/ (Fu 2014 et al., 2014)

E-CRISP - http://www.e-crisp.org/E-CRISP/ (Heigwer et al., 2014)

CHOPCHOP - http://chopchop.cbu.uib.no/ (Montague et al., 2014)

CRISPRdirect - http://crispr.dbcls.jp/ (Naito et al., 2015)

Breaking Cas- http://bioinfogp.cnb.csic.es/tools/breakingcas (Oliveros et al., 2016)

Note

Depending on the design of the genome editing experiment, the following suggestions for selecting sgRNA need to be considered:

• For knockin mice, a sgRNA should be chosen that cuts closest (<10–20 bp) to where the inserted mutation (point mutation, loxP site, protein tag, etc.) is to be introduced.

• If using Cas9n, two sgRNAs are required to make a DSB. For this double-nicking strategy, the sgRNAs should be designed to create a 5’overhang with a gap of only 0–20 bp for efficient NHEJ or HDR (Ran et al., 2013b). Shen et al. (2013) also suggest that the sgRNAs should be in a tail-to-tail orientation. The MIT CRISPR tool listed above is able to perform nickase analysis to design sgRNAs.

• Two sgRNAs instead of one should be used when intending to make large deletions. A large deletion may be preferable to ensure a definitive knockout over an indel mutation that generates a frame shift.

• For knocking in a large mutation, use two sgRNAs for high knock in efficiency. The two sgRNAs are designed to generate a DSBs within the interior genomic sequence near to each flanking homology arms.

Improvements to the structure and chemistry of the synthetic guide RNAs are being examined to enhance base pairing and prevent nuclease digestion (Kelley et al., 2016). Chemical modification to the single guide RNA or dual RNAs (crRNA and the tracrRNA) has been reported to provide nuclease resistance and enhance genome editing efficiency in human cells (Hendel et al., 2015; Rahdar et al., 2015), but such modifications of the guide RNAs, however, may not necessarily provide any further advantages for genetically engineered mice, as extended Cas9 activity could lead to more off-target damage. Future testing of these modified guide RNAs is required before adaptation to mouse genome engineering.

BASIC PROTOCOL 2

DETERMINATION OF sgRNA EFFICIENCY BY BLASTOCYST TEST

Factors such as DNA accessibility can affect the on-target efficiency of the sgRNA. One option is to test the efficiency of the sgRNAs first before proceeding to make the mutant mice. If an adequate supply of donor female mice is available for additional zygote collection, the efficiencies of the sgRNAs can first be tested in microinjected blastocysts that are grown in culture for 4–5 days. Testing of the blastocysts will best predict the sgRNA efficiency before actual implantation into pseudopregnant females and testing of newborn pups.

Materials

Blastocyst Lysate Buffer (Scavizzi et al., 2015):

50mM KCl (Filtered)

10mM Tris-HCl (pH 8.3)

2.5mM MgCl₂

0.1mg/ml gelatin (from porcine skin, type A; Sigma)

0.45% NP40

0.45% Tween20

0.1mg/ml Proteinase K

0.2ml PCR tubes (Fisher Scientific)

KSOM medium (Millipore Sigma, cat. no. MR-121-D)

M2 medium (Millipore Sigma, cat. no. MR-015-D)

Culturing of microinjected embryos to determine sgRNA effectiveness

1. Culture microinjected embryos in KSOM medium for 4–5 days to develop until they become hatched blastocysts at 37°C, 5% CO_2.

2. Using a pulled glass pipette, aspirate each blastocyst in a minimal volume of M2 medium and place into the bottom of 0.2ml PCR tube. These samples can be frozen in −20°C for later use.

3. Set up Thermocycler program for 56°C for 30 min, then 95°C for 10 min for inactivating the Proteinase K.

4. Add 10μl of blastocyst lysis buffer to each PCR tube, then place in a Thermocycler and start the program.

5. When the program is over, samples are vortexed for 5–10s, then centrifuged for 1min at maximum speed after cooling down.

6. Perform regular PCR reaction with the pair of primers flanking asymmetrically DSB region using 5 μl of the lysates as template. Details concerning genotyping using an enzyme mismatch cleavage assay (T7E1 assay or Surveyor^®) are listed below.

Note

1. When testing genome editing efficiencies, it is critical to achieve a sufficient amount of the correct PCR product that is clear and specific. It is, therefore, highly recommended to test the PCR conditions using wild-type blastocysts before performing the assay.

2. Hatched blastocysts from 4–5 days of culture increase the yield of PCR product compared to blastocysts with intact zona pellucida from 3–4 days of culturing.

3. The mouse blastocyst test can serve as a guideline for judging the efficiencies of each sgRNA in a knockin experiment (Fig. 4). Therefore, it is recommended that sgRNAs with ≥ 50% of the NHEJ editing efficiency be used for subsequent generation of knockin mice through HDR.

Figure 4.

Correlation graph between HDR efficiency (%) and NHEJ efficiency (%) showing significant coefficient between the two values (R² > 0.99). The NHEJ efficiency was calculated by screening 15–19 blastocysts that are injected with three sgRNAs. The HDR efficiency was calculated by screening 37–81 live pups that are injected with each corresponding sgRNAs and donor DNA (a single strand oligonucleotide). Note that for the sgRNA with under 20% NHEJ efficiency, there were 0 HDR events out of 81 live pups. Based on this correlation, it is suggested that sgRNAs with over 50% of NHEJ efficiency should be used to achieve at least 5% HDR efficiency. These data are based on four different sgRNAs for two independent genes.

Although not equivalent to the actual conditions in the mouse zygote, validation of the sgRNAs can be tested in tissue culture if donor mice for mouse blastocyst testing are not readily available (Ran et al., 2013a). ES cells, for example, (Wang et al., 2013; Yang et al., 2013) can be employed to determine genome editing efficiency before injection, particularly since ES cells have better HDR efficiencies than somatic cells (Yu et al., 2015). The efficiency of sgRNA can also be readily examined using the pCAG-EGXXFP plasmid (available at Addgene), which employs a fluorescent means to determine if a DSB is generated (Mashiko et al., 2013). In this assay, a 500 bp target genomic sequence is subcloned into pCAG-EGXXFP. Then, the resulting modified target plasmid is transfected in HEK293T cells along with pX330 (a plasmid used to express both Cas9 as well as the selected sgRNA – also available at Addgene). If the targeted sequence is cleaved, HDR will proceed to align the EGFP fragments and restore fluorescence, where the efficiency of the sgRNA being graded by fluorescence intensity. One can also simply test the efficiency of the guide RNAs by running an in vitro digestion assay. With this cloning-free CRISPR/Cas system, the genomic sequence to be targeted is amplified by PCR and used as a template to test guide RNA efficiency using recombinant Cas9 protein (Aida et al., 2015).

BASIC PROTOCOL 3

GENERATION OF KNOCK OUT MICE BY NHEJ

CRISPR mediated gene knockout mice containing indel mutations can be easily derived in one step via microinjection of the mouse zygote (Wang et al., 2013). All the steps required for the collection of mouse zygotes from donor female mice has been previously described in Cho et al. (2009). The same procedure for harvesting zygotes to make transgenic mice can be applied for genome editing with CRISPR as well. Our protocol chapter on the “Generation of Transgenic Mice” provides detailed instructions on the superovulation of the donor female mice as well as the harvesting of the resulting donor zygotes. We typically harvest around 250 zygotes from 15 superovulated female mice. For donor mice, we typically use a FVB/N strain, particularly because the fertilized zygote contains a large pronucleus that better withstands microinjection. CRISPR genome editing, however, can be conducted in other mouse strains, which is an improvement over the limited number of germline efficient ES cells available with previous gene targeting. One-step generation of mutant mice using CRIPSR additionally produces genetically engineered mice on a single strain, rather than the mixed background typically generated by injecting ES cells into donor blastocysts. Modification to our protocols on superovulation and zygote collection, however, might be required when using different strains (Qin et al., 2016). Delivery of the sgRNA and Cas9 mRNA into the mouse zygote can be performed either through cytoplasmic injection, often using a piezo-based micromanipulator, or by pronuclear injection, typically with a microinjector (Yang et al., 2014). In general, no phenotypic differences have been detected with either method of delivery (Horii et al., 2014). The concentrations of sgRNA and Cas9 mRNA, however, need to be lowered with pronuclear injection because of potential toxicity (Yang et al., 2014). Cytoplasmic injection of the donor zygote has been suggested to result in better genome engineering efficiency and better blastocyst viability, particularly if just making knockout mice (Horii et al., 2014). Both CRISPR Cas9 genome editing and transgenic mice production, however, are possible if skilled in standard pronuclear injection. Pronuclear injection, in particular, brings the CRISPR reagents directly to the site of activity, which is advantageous when introducing a donor plasmid or injecting Cas9 protein (even with a NLS) (Aida et al.,2015; Singh et al., 2015; Low et al., 2016). As detailed below, we microinject the pronucleus using a constant flow rate to essentially deliver the CRISPR reagents into both the cytoplasm and pronucleus. When microinjecting, the expansion of the pronucleus through delivery of the CRISPR reagents also nicely provides a visible guide to assess the quality of the injection needle (Cho et al., 2009). For all intents and purposes, we will refer to pronuclear injection procedure described in detail in Cho et al. (2009) for making CRISPR mediated genetically engineered mice. This protocol, therefore, provides a single method for making all types of genetically engineered mice, include transgenic mice, knockouts, and knockins.

Materials

Injection/dilution buffer (10mM Tris and 0.1mM EDTA in RNAase free H₂O) For a 50 ml volume, add 500 μl 1M Tris pH 7.4, 10 μl 0.5M EDTA, 49.5 ml RNAase free H₂O). It is convenient to filter this solution using a vacuum filtration device (Steriflip, Millipore-Sigma, cat. no. SCGP00525)

0.2μm syringe filter (Nalgene cat. no. 171–0020)

Purified sgRNA from above

Cas9 mRNA (Trilink, cat. no. L-7206, or IVT), Cas9 protein (Thermo Fisher Scientific, cat. no. B25640), or pX330 (to express sgRNAs and Cas9)

RNASIN, Ribonuclease inhibitor - 10,000U (Promega, cat. no. N2115)

RNase AWAY (Thermo Fisher Scientific, cat. no. 10328011)

M2 medium (Sigma Chemical, cat. no. M7167)

Mineral oil (Sigma Chemical, cat. no. M3516–1L)

Washed, ready-to-inject zygotes in M2 medium (Basic Protocol 2 of Cho et al., 2009)

M16 medium (Sigma Chemical, cat. no. M7292)

Microinjection system including the following components (Displayed in Cho et al., 2009): Inverted microscope (Zeiss, Axiovert 135M) ideally using DIC optics for glass microinjection platform

Anti-vibration table (Kinetic systems)

Automated DNA injector (Eppendorf, Femtojet)

Capillary holder for holding pipet (Mitutoyo)

Micromanipulator (Narishige, model MO-202U)

Compressed nitrogen gas cylinder used for anti-vibration table

Microloaders (Eppendorf, cat. no. 930001007)

One-chambered glass slide to be used for the microinjection platform (Nalge Nunc International, cat. no. 177372)

Mouth-controlled pipet assembly for handling zygotes

CO2 incubator (Thermo Electron Corporation, model 370)

Microinjection needles and holding pipets

Note

Making holding pipets with a Microforge (Technical Products International Corporation, MF-1) and microinjection needles using a needle/pipet puller (Kopf Instruments, model 720) and glass tubing (Sutter Instrument, cat. no. BF100–78-10) has been previously described (Cho et al., 2009). Prior to microinjection, carefully break the tip of the injection needle against the holding pipet to obtain a good DNA flow.

Microinjection of Cas9 and sgRNAs into mouse embryo

1. Spray the work area with RNase AWAY solution.

2. Prepare 10mM Tris / 0.1mM EDTA injection/dilution buffer solution.

3. Filter the buffer solution (at least 500 μl) through a 0.2μm syringe filter.

4. Add RNASIN RNase inhibitor: 2 μl 10,000U RNASIN+ 398 μl buffer solution to a final concentration of 0.2U.

5. Dilute Cas9 mRNA (in vitro transcribed or commercially purchased) to 10 ng/μl with injection buffer or dilute Cas9 protein to 30 nM (commercially purchased) and add to the tube in step 4.

6. Dilute sgRNA to 5 ng/μl and add to the tube in step 5. IMPORTANT: Occasionally a sticky gel-like material can clog the injection needle in subsequent steps. To reduce clogging, we often centrifuge this final injection mixture in a table top centrifuge at 13,000 rpm for about 2 minutes.

7. The injection solution should be kept cold, at about 4°C (to further inhibit RNase activity). Load the DNA solution into the back of the injection needle using a microloader pipet. Take the top of the injection solution to avoid the “gel-like” precipitate from addition of sgRNA.

8. Microinject the CRISPR reagents into the larger pronucleus of 0.5 dpc (days post coitum) zygotes as described in Cho et al., 2009. For microinjection, we use a constant flow rate of 50–70 hPa (hectopascal) to deliver the CRISPR components. The constant flow rate also has the advantage of delivering CRISPR components, especially the IVT transcribed sgRNA to the cytoplasm, in addition to the pronucleus.

9. Inject about only 40 eggs at a time to essentially avoid keeping zygotes outside the incubator for too long.

10. After injecting each embryo, move the microscope stage down and release the zygote at the bottom of the M2 medium in order to differentiate microinjected zygotes from uninjected ones during microinjection.

11. Transfer microinjected zygotes into a pre-warmed M16 culture medium dish under mineral oil and maintain the zygotes at 37°C in a 5% CO2 incubator until ready for implantation into recipient mice.

12. Transfer the embryos into the oviduct of pseudo-pregnant females (described in Cho et al., 2009)

13. Pups are generally delivered about 20 days after microinjections.

14. From 10 day postnatal pups, collect either a tail snip or ear biopsy for genotyping purposes.

15. Indels generated by NHEJ are genotyped using an enzyme mismatch cleavage assay, which is detailed below.

16. Select 3 founder mice for further analysis. Multiple founders are preferred to reduce the possibility of generating mosaic mice.

Forms of Cas9

The Cas9 endonuclease typically produces a DSB 3 nt upstream of the PAM site (Yu et al., 2015). Cas9 creates a blunt ended cut in the DNA that is then repaired by the cell through NHEJ (Zetsche et al., 2015). NHEJ eventually results in indel mutations that prevent further target recognition. The most common mutations generated with error-prone NHEJ are short deletions, but insertions, point mutations, and large deletions can also occur (Shin et al., 2017).

1. Cas9 mRNA: Cas9 mRNA that is ready for microinjection is available from Trilink or IDT. Alternatively, Cas9 mRNA can be derived using the expression plasmid pX330 through IVT. If using pX330, the following oligonucleotides are needed to make a PCR template for IVT (Wang et al., 2013):

   Cas9-f

   TAATACGACTCACTATAGGGAGAATGGACTATAAGGACCACGAC

   Cas9-r

   GCGAGCTCTAGGAATTCTTAC

   Another Cas9 expression vector is pCAG-T3-hCAS-pA (also from Addgene) (Fujii et al., 2013). This vector has a CMV/T3 promoter that can allow for mammalian expression, while also permitting IVT of Cas9 mRNA.

2. Cas9 Protein: Microinjection of Cas9 plasmid or mRNA requires translation in the cytoplasm and importation back to the nucleus, so injection of Cas9 protein may be preferred because it functions immediately at its intended cellular site and degrades quickly to reduce chances of off-target DNA damage and mosaicism.

3. Cas9 Plasmid: The most commonly reported plasmid for Cas9 expression is pX330, which is available at Addgene (Cong et al., 2013). Injection of a single plasmid for expression of both sgRNA and Cas9 can help to minimize the steps needed to make mutant mice (Mashiko et al., 2013). The pX330 CRISPR vector contains two expression cassettes, a CBh (altered chicken beta actin) promoter to express a human codon optimized SpCas9, and a U6 promoter for the sgRNA. Subcloning of a selected sgRNA is mentioned above. A modified version of pX330 (Addgene) was developed that allowed for multiplex genome editing in HEK293T cells, but may be useful in creating genetically engineered mice as well (Sakuma et al., 2014). This technique basically employs Golden Gate cloning for assembly of a single plasmid vector of multiple sgRNA expression cassettes. Such vector assembly may be useful if more than one sgRNA is required, as is the case with conditional knockouts, the dual nickcase strategies, large deletions, etc. Although the extended expression of Cas9 via a plasmid may increase the chance of genome editing, it may also increase the possibility of off-target mutations (Singh et al., 2015). Injection of a plasmid can also lead to random insertion of the Cas9 transgene, but at a low frequency (Mashiko et al., 2013).

4. Cas9 Ribonucleoproteins (Cas9 and guide RNA complex): Preassembly of Cas9 into ribonucleoprotein particles has been suggested as a means to improve genome editing efficiency while reducing off-target effects, as Cas9 is injected into the pronucleus in its fully active state. Using cultured cells and mouse single cell embryos, Kouranova et al. (2016) suggested that direct injection of Cas9 preformed into ribonucleoprotein particles (RNP) is the most efficient means of genome editing, where recombinant Cas9 is incubated with the sgRNA before microinjection at about a 1:5 molar ratio. Others, however, recommend pronuclear injection of a Cas9 protein complexed with dual RNAs of both the targeting crRNA and the tracrRNA instead of a single chimeric guide RNA (Aida et al., 2015; Quadros et al., 2017). The Easi-CRISPR method mentioned above (Quadros et al. 2017) suggest the following procedure to produce of an active Cas9 RNP: For RNP mediated genome editing, combine the crRNA/tracrRNA preparation (derived from Basic Protocol 1 – Method 3) with Cas9 (Alt-R^® S.p. Cas9 Nuclease 3NLS – IDT, cat. no. 1074181). Both the crRNA/tracrRNA and Cas9 are then diluted to a final concentration of 20 ng/ μl in a total volume of 100 μl.

Generation of Founder Mice by CRISPR

After genome editing with Cas9, the genetically modified zygotes are incubated at 37°C until ready for implantation. Pseudo-pregnant females need to be prepared ahead of time for implantation of the zygotes through mating of wild-type female mice to vasectomized males. The surgery involved in transferring the embryos into the oviduct of pseudo-pregnant females has been described already by Cho et al. (2009) in the same protocol chapter mentioned above about the “Generation of Transgenic Mice”. Alternatively, the microinjected embryos can be cultured overnight to develop to the two-cell stage and transferred in pseudo-pregnant females the following day. Pups should be born after about 20 days after injections. A tail snip or an ear punch should be collected in about 10 days postnatal for genotyping. Aspects of genotyping are listed below. Although rare, off-target mutations can occur with genome editing with CRISPR in mice. If off-target damage is a concern, an enzyme mismatch cleavage assay (detailed below) can be conducted to identify possible unintended indels. Most CRISPR design tools should be able to identify possible off-target sites. Actual off-target sites, however, are often overlooked by most current software and some mutations such as large deletions would not even be detectable with PCR (Tsai et al., 2015; Haeussler et al., 2016). Depending on the genetic linkage, however, most off-target mutations can be lost when mating founders to new wild-type mice of the same inbred strain. Breeding with wild-type mice will introduce “clean” DNA onto your genetically engineered mouse model. The phenotype of your mouse model should still be verified using more than one founder to ensure the validity of your findings and completely rule out any possible off-target effects. Mosaicism can also occur using CRISPR, but different NHEJ mutations can be segregated as well by breeding founders to wild-type mice. For all these reasons, founder lines should definitely be backcrossed at least once with a wild-type mouse. A minimum of three founder lines should be selected for further analysis. Males are preferable to females because males allow for accelerated breeding. Ideally, all mouse lines of specific interest should be cryopreserved. CRISPR provides a simple one-step means of generating knockout mice, but a majority of the ~ 25,000 mouse genes have been knocked out already through either conventional gene targeting or by gene trap vectors (Hall et al., 2009). In 2007, when the Nobel Prize in Physiology or Medicine was awarded to Mario R. Capecchi, Sir Martin J. Evans and Oliver Smithies for their work on gene targeting in embryonic stem cells, about 11,000 genes were already knocked out in mice. Currently, about 18,000 knockout alleles are available through the International Knockout Mouse Consortium (Singh et al., 2015). Nonetheless, in certain instances, it may be advantageous to use CRISPR to redevelop a knockout mouse. For example, multiple backcrosses (at least five) are needed to get a knockout allele on a desired mouse strain with CRISPR. The sharing of mice can be also problematic because of mouse pathogens. Lastly, some knockout alleles are only available within ES cells, but germline efficiency can be a concern. CRISPR genome editing can, therefore, provide an expedient means of circumventing these problems to generate knockout mice. Careful analysis of CRISPR-generated knockout, however, needs to be conducted to ensure both deletion of the targeted protein from just a single indel mutation and faithful replication of the expected phenotype (Singh et al., 2015).

Minimizing Off-Target binding

Various strategies have been employed to minimize the degree of off-target DNA damage, which can be a potential drawback to using CRISPR genome editing, particularly for genetically engineered mice (Slaymaker et al., 2016). With genome editing in human cells, Cas9 has been shown to accommodate up to five mismatches from the intended targets (Fu et al., 2014). Truncated guide RNAs of 17 to 18 nucleotides have been suggested as a means of reducing off-target mutagenesis, as the 5’-end nucleotides may actually allow for mismatches rather than enforce specificity (Fu et al., 2014). The CRISPR design tool ZiFiT Targeter program (http://zifit.partners.org/) was subsequently modified to allow for selection of truncated guides. Injection of the recombinant Cas9 protein has been suggested as another means of reducing off-target mutations because Cas9 will degrade shortly upon entry into the pronucleus and limit the time of actual targeted nuclease activity, thereby minimizing the chance for unintended DNA damage. Limiting the Cas9 endonuclease activity, however, may have a negative impact on the efficiency of genome editing. As previously mentioned, a modified Cas9n double-nicking strategy can be employed to increase specificity and prevent off-target damage as well. The dual-nickase strategy, however, requires two sgRNAs to work effectively together, which can lead to lower genome editing efficiency. Lastly, new Cas9 variants have been developed to improve specificity and decrease the level of off-target damage. Slaymaker et al. (2016) produced a Cas9 with “enhanced specificity” (eSpCas9) by introducing mutations to attenuate its helicase activity and make mismatches less favorable. Kleinstiver et al. (2016) mutated four sites within Cas9 that limited the energy needed for recognition to produce a more specific “high-fidelity” Cas9 (SpCas-HF). Thus, SpCas-HF, eSpCas9, future specific Cas9 mutant variants, or even other types Cas proteins from different bacteria may eventually become standard genome editing tools to make genetically engineered mice while alleviating concern about off-target mutations.

Electroporation: an alternative to microinjection

Direct pronuclear injection of DNA into the mouse zygote, a preferred method for developing transgenic mice, has essentially been adapted for use in genome editing through CRISPR, but microinjection can be a technically difficult to master and time consuming. Electroporation has become an alternative to microinjection (Hashimoto et al., 2015; Qin et al., 2015). Electroporation can be a simple high throughput means of genome editing that is less invasive than microinjection. Both CRISPR generated knockout and knockin mice have been created using this technique, and the birth rate of the embryos is generally higher with zygote electroporation. The zona pellucida, however, can act as a barrier that prevents electroporation, so treatment of zygotes with acidic Tyrode’s solution has been suggested as a way to weaken this glycoprotein layer, and, therefore, allow for efficient nucleic acid delivery (Qin et al., 2015). Zygote electroporation has also been applied to deliver both Cas9 protein and whole Cas9 ribonucleoprotein complexes as well (Wang et al., 2016; Chen et al., 2016). With the weakened zona, the zygotes often become sticky and are difficult to remove, so some suggest washing the embryos with KSOM+BSA (Chen et al., 2016).

BASIC PROTOCOL 4

GENERATION OF KNOCK IN MICE BY HDR

The generation of knockin mice with CRISPR requires the additional injection of donor DNA for targeted insertion of a desired sequence into a designated genomic location. Essentially, the creation of a DSB facilitates homologous recombination if donor DNA is present, but HDR still occurs at a lower frequency than NHEJ. Moving beyond just making knockout mice, there is great interest in using CRISPR genome editing to introduce subtle mutations (Menke, 2013; Singh et al., 2015). Most human monogenetic disorders, for example, involve single nucleotide substitutions or small insertion/deletions within a gene. Single amino acid substitutions can affect enzyme activity or transcription factor signaling, while nucleotide changes in the promoter of a gene can have consequences on gene expression (Inui et al., 2014). CRISPR provides an efficient means of knocking in point mutations to mimic human diseases, particularly if there is adequate homology between the mouse and human gene. A mutation in the mouse genome can essentially be generated using CRISPR by additional injection of a mutant donor oligo. Along with subtle mutations, the addition of small protein tags such as HA, Myc, FLAG, etc. can also be readily introduced using short donor oligos. The insertion of an epitope tag can be useful when trying to track the expression of a protein of interest, particularly if good antibodies are not available. Fluorescent tags, however, typically require co-injection of a plasmid rather than short oligos due to the size of the fluorescent marker. Lastly, CRIPSR allows for insertion of loxP sites to make conditional knockout mice that are useful for the identification of tissue specific gene functions. Conditional knockout alleles can be produced by injection of the whole floxed exon sequence in a plasmid or by adding two short loxP containing oligos along with the accompanying sgRNAs to properly flank an essential exon (Yang et al., 2013).

Materials

Cas9 and sgRNA preparation is same as above in Basic Protocol 3.

Donor DNA:

1. A short oligos (100–200 bp) can be used for point mutations, loxP sequences, the insertion of short protein tags, etc. The homology arms should be about 30–60 bp in length (Jacobi et al, 2017).

2. Plasmids have traditionally been used for larger DNA inserts such as fluorescent proteins, minigenes, floxed exon/s, etc. Donor plasmids require longer homology arms of about 0.5–3kb, depending on the size of the insert.

3. New methods in single-strand DNA synthesis have allowed for long single-stranded DNAs (ssDNA) to be used as donor DNA (see below). Repair templates derived from long ssDNA have been shown to have equal or higher HDR efficiencies than plasmid DNA (Miura et al., 2015; Paix et al., 2017). For single-stranded donor DNA, the 3’ homology should include sequence directly downstream of the DSB (Paix et al., 2017).

4. Double stranded PCR fragments with short homology arms (~35 bases) can also be employed to insert fluorescent tags (~ less than 1kb) into targeted alleles (Paix et al., 2017). PCR fragments have an added advantage of ease of synthesis, especially for large fragments.

Introduction of donor DNA for generation of knockin mice using CRISPR

1. Add a final 5 ng/μl donor DNA to the final Cas9, sgRNA, RNASIN buffer solution #6 above.

2. Microinject the whole mixed CRISPR reagent into pronucleus as above.

3. Transfer the embryos into pseudo-pregnant mice as above.

Note

HDR occurs at a low efficiency, even with a DSB, so more NHEJ generated indel mutations are likely to be found amongst the pups from a microinjection than knockins. Importantly, the donor DNA used for microinjection needs to be pure of any embryotoxic chemicals. If using plasmid DNA, for example, be sure to purify the plasmid using an endotoxin free kit. Listed above is a standard concentration of donor DNA needed for generation of knockin mice. While increasing the concentration of donor DNA may, in certain cases, improve knockin efficiency, it can inhibit HDR at high amounts (Raveux et al., 2017). Injecting high concentrations of DNA into the mouse zygote can additionally be toxic.

With a short oligo as the donor DNA, desalted oligos are normally used for CRISPR, which are free of any contaminants used during synthesis. The donor DNA can complement either strand, sense or antisense. Longer homology arms beyond 60 bp generally will actually reduce HDR efficiency. Modifying the donor DNA to have a defined asymmetry to the PAM site may improve HDR (Richardson et al. 2016). Overall, when designing a short oligo DNA, be sure the mutations are introduced within 10–30 bp of the DSB.

For a plasmid donor DNA, homology arms shorter than 500 bp will result in lower HDR efficiency (Raveux et al., 2017). Inserts of 1–2 kb have been generated using CRISPR, but the efficiency of HDR generally decreases as the size of the insert size increases beyond this length. Plasmid donor DNA can randomly integrate into the genomic DNA, particularly at high concentrations. For CRISPR mediated HDR, the plasmid donor DNA does not need to be linearized, which may help prevent random integrations (Yang et al. 2013). Alternatively, long SS DNA templates can be used, which are both less toxic and less prone to integrate than double stranded DNA.

Smart Donor DNA design

If possible, the PAM should be mutated to prevent cleavage of the donor DNA or knocked in sequence. The mutations to the PAM site should not, however, alter the amino acid coding sequence. Although Cas9 best recognizes NGG, NAG is also possible as a PAM site. If the PAM cannot be mutated, silent mutations can be made within the 20 bp target recognition sequence. Mutations to the intronic sequence are preferable to any change in the coding sequence, if possible. Depending on the type of knockin, a nucleotide mutation to add a restriction enzyme site can be useful for subsequent genotyping purposes using Restriction Fragment Length Polymorphism (RFLP) (Cong et al., 2013). Unlike ES cell based homologous recombination, a donor plasmid to be used for genome editing does not need to be linearized since this might result in random integration. In addition, positive and negative selection markers are not required within the donor DNA plasmid as needed with conventional gene targeting in ES cells. As mentioned above, the genome editing strategy should have a DSB produced as close as possible to where the desired inserted sequence is to be introduced. The knockin sequence ideally should be incorporated within 10 nucleotides of the DSB, but definitely not beyond 100 bp, as the efficiency of HDR decreases with distance from the Cas9 generated DNA cut. It can also be useful to introduce DNA near to the designated DSB site particularly if this change disrupts the PAM and blocks further Cas9 target recognition. In contrast to traditional gene targeting that relies on rare homologous recombination events, the donor DNA used with CRISPR can have shorter homology arms as the DSB promotes HDR to mend the genetic damage. Still, some simple considerations should be followed when designing the donor DNA. The mouse strain used for genome editing must be considered when constructing the donor DNA. Because of SNPs and other polymorphisms between mouse strains, isogenic DNA should be used if possible as a template when constructing donor DNA. Recombination can also be more efficient if using isogenic DNA. As mentioned, we typically generate both transgenic and CRISPR engineered mice on an FVB/N background because the large visible pronucleus in this strain is more capable of withstanding microinjection. In contrast, a C57BL/6 background may be preferred since it is the most commonly used inbred laboratory strain and most CRISPR design tools use C57BL/6 as a reference genome. Most genomic sequence information to construct donor DNA is available through the Mouse Genome Sequencing Consortium (MGSC). The location of any targeted insertion needs to carefully be examined so as to not interfere with any regulatory elements. If using PCR to generate homology arms, one should use a high-fidelity Taq. Even with CRISPR, large insertions greater than 5 kb are difficult to generate in mice. If trying to generate a floxed conditional knockout allele, the loxP sites need to be placed in an intronic sequence so as to not interfere with protein expression. Ensure that the loxP sites in the targeting construct are the same orientation or the floxed sequence will be inverted rather than deleted from the gene. Further details about Cre mediated gene recombination and loxP sites have been previously described in Hall et al. (2009), including the sequence for loxP.

In general, single stranded oligos (ssODNs) exhibit higher rates of recombination into a targeted allele than double stranded DNA (Quadros et al., 2017). The use of single stranded DNA repair templates also limits cytoxic responses and decreases the potential of random integration. Most oligos are, however, limited in length to 200 bases long. New techniques, however, have been developed that allow for longer ssDNA, where a template DNA is first transcribed into RNA, then reverse transcribed to generate ssDNA (Miura et al., 2015). Long ssDNA of up to 2000 bases are also available from IDT technologies (Megamer^™ Single-Stranded DNA fragments). Alternatively, TaKaRa-Clontech (cat. no. 632644) has a Guide-it^™ Long ssDNA Production System for generation of ssDNA donor templates between 0.5–5 kb in length. This kit uses PCR followed by Strandase digestion to make ssDNA donor templates. Artificial microRNA and floxed exon sequences have been reported to be efficiently recombined into targeted alleles using long ssDNA. Along with long ssDNA, modification of the donor oligonucleotides has also been suggested to help improve HDR. Modifications to the phosphate backbone, such as 5’ and 3’ end phosphorothioate-modification (substitution of sulfur for oxygen) not only helps to prevent nuclease digestion, but is also able to improve HDR mediated knockin insertion in mice without toxicity (Renaud et al., 2016).

Enhancing HDR efficiency

After Cas9 produces a blunt ended DSB in the DNA, it will be repaired either through the NHEJ or the HDR pathway. In terms of genome editing, the HDR pathway is inherently a less efficient process than NHEJ, so a CRISPR mediated knockout allele is more often generated than the desired knockin. Typically, when a DSB is generated, it is rapidly repaired through NHEJ with simple ligation of the DNA ends. Unlike HDR, NHEJ does not requires a repair template. Compared to NHEJ, HDR is a more complicated and, consequently, slower means for DSB repair. HDR first involves resection of the DNA to generate a 3’overhang to serve as a scaffold for DNA repair proteins. Next, these repair proteins search for a matching homologous template in order to properly correct the CRISPR generated DSB (Brandsma and Gent, 2012). Although NHEJ occurs throughout the cell cycle, HDR repair pathway is principally limited to the S and G2 phases. Due to the speed and simplicity of NHEJ as compared to HDR pathway, indels essentially occur more frequently than the preferred knockin mutation (Yang et al. 2014).

New techniques of enhancing HDR are being developed to help expedite the development of knockin mice. Pronuclear injection of Cas9 complexed with dual crRNA and tracrRNA, instead of a chimeric single guide RNA, has been suggested as a means of facilitating the development of knockin mice (Aida et al., 2015). Chemical enhancers of HDR along with inhibitors of NHEJ are also being assessed for their ability to promote donor DNA insertion rather than the generation of an indel mutation. Inhibition of DNA ligase IV with SCR7, for example, has been proposed to suppress NHEJ and encourage HDR in mouse zygotes (Maruyama et al., 2015; Chu et al, 2015; Singh et al., 2015), but many gene targeting core facilities find that perturbation of DNA repair processes may be too toxic (Quadros et al., 2017). In contrast, Song et al. (2016) suggest that an enhancer of HDR, RS-1 (RAD51-stimulatory compound 1), improved the probability of getting a knockin allele over a NHEJ mutation over SCR7, which has minimal effects. To uncover additional compounds that facilitate HDR, Yu et al. (2015) used a high throughput screen that measures the insertion of a fluorescent reporter and have identified other small molecules that promote HDR over NHEJ. Lastly, genome editing with Cpf1 (as mentioned above) may promote HDR better than Cas9 (Zetsche et al., 2015). With Cas9, the blunt end DSB near the PAM site essentially disrupts target recognition if NHEJ occurs, but the staggered cut by Cpf1 occurs far from target site, which may allow for a subsequent chance at HDR. More testing needs to be conducted to determine if any of these approaches will become a gold standard for making knockin mice.

Basic Protocol 5

GENOTYPING TO IDENTIFY GENETICALLY ENGINEERED MICE

After implantation of CRISPR mutated embryos into pseudo-pregnant females, the mouse pups are born typically 20 days later and, for genotyping purposes, a tail snip or ear punch biopsy should be collected when the pups are 10 days of age. In terms of the expected genotypes, an overview of targeting efficiencies has already been reported in Table 1 of Yang et al. (2014). As mentioned, about three founders should be selected, usually males. A PCR strategy for identifying the specific NHEJ or HDR mediated mutation should be planned before pups are derived. In most case, primers need to be designed that flank the location of the DSB. Indels are then typically detected using a T7E1 or Surveyor^® assay (Fig. 5), while knocked in point mutations can be identified by RFLP analysis if a restriction enzyme site was either introduced or ablated (as described above) (Fig. 6). We typically design primers using the Primer-BLAST software program (https://www.ncbi.nlm.nih.gov/tools/primer-blast/), but other primer designing tools are available. Primers should be designed to generate a PCR product of about 400–800 bp in length, particularly if T7E1 or RFLP analysis is required (Ran et al., 2013b). PCR products should be DNA sequenced and aligned to the wild-type sequence to determine the type of indel generated. If just generating knockout mice with NHEJ, biallelic mutations are possible that should probably be segregated and analyzed separately. With the founder mice, mosaicism can occur in about 11–35% of the pups, depending on the gene (Singh et al., 2015), probably because of a delay in Cas9 activity. The different mutations can be segregated by breeding to wild-type mice, but mosaicism may confound initial PCR genotyping of founder mice.

Figure 5. Genotyping for CRISPR Generated Indel Mutations.

A) Schematic of T7E1 enzyme mismatch assay. First, the genetic locus with an indel mutation (X) is amplified by PCR. Flanking primers should generate a product where the indel is asymmetrically situated in the PCR product. The PCR is then denatured and reannealed. Any mismatch is then cleaved by the T7E1 enzyme (WT – wildtype; HT – heterozygous) to create two smaller fragments on an agarose gel. B) Gel picture of the T7E1 enzyme mismatch assay using mouse tail DNA.

Figure 6. Genotyping for CRISPR Generated Point Mutations.

A) Schematic demonstrating a PCR based method for detecting the insertion of a point mutation using RFLP (restriction fragment length polymorphisms). A non-interfering restriction enzyme (RE) site is introduced into the targeted allele along with the designated mutation using CRISPR/Cas9 and the accompanying donor DNA. The site of the introduced point mutation and RE site is then amplified using flanking PCR primers. The RE site should ideally be located asymmetrically in the PCR product to produce two distinct smaller cleavage fragments on an agarose gel (WT – wildtype; HT – heterozygous; KI - knockin). B) Conversely, a restriction enzyme site can also be eliminated by HDR, as shown in the gel picture. An EcoRI site was lost upon insertion of the donor DNA in order to allow for genotyping of the CRISPR generated knockin mouse where KI PCR band is not cut by the RE.

Identifying knockout mice by enzyme mismatch assays

Indels are typically detected using a T7E1 assay or Surveyor^® enzyme mismatch cleavage assay (Fig. 5). A PCR strategy should be designed to amplify the sequence around the DSB. Enzyme mismatch cleavage assays essentially involves PCR of the DSB site, denaturation of the PCR products, subsequent rehybridization, and, lastly, treatment with an enzyme that can then digest mismatched DNA. The T7 endonuclease 1 is a bacteriophage enzyme that recognizes and cleaves heteroduplexed DNA. The Surveyor^® assay, in contrast, uses a mismatch-specific nuclease that is a member of the CEL family from celery. PCR primers for an enzyme mismatch assay should be designed to flank the DSB asymmetrically to aid in the detection of a heteroduplex (e.g. Forward primer in the 200 bp 5’ upstream from DSB region is paired with Reverse primer 400 bp 3’ downstream from the same region) (Fig. 5).

Materials

2 primers designed to amplify the DSB.

PCR thermal cycler

0.2 ml thin-walled PCR tubes

2% TAE agarose gel and equipment for gel electrophoresis

T7E1 Mismatching assay kit for Knock out allele detection (NEB, cat. no. M0302)

or

Surveyor^® Mutation Detection Kit (Integrated DNA Technologies, Coralville, Iowa)

T7 Endonuclease mismatch assay

1. Run PCR to amplify DSB region, usually a 50 μl reaction. About 100 ng of genomic DNA is typically used as a template for PCR. The PCR should include not only the test DNA, but also genomic DNA from a normal (wild-type) mouse of the same strain.

2. Melt and anneal the resulting PCR products (~200 ng PCR template) using the following touchdown PCR program

   1. 95°C for 2 min. for denaturation

   2. decrease from 95°C to 85°C at −2°C/sec

   3. decrease from 85°C to 25°C at −0.1°C/sec for annealing

   4. finish at 16°C

3. T7 Endonuclease I Digestion:

   • 10 μl of the PCR product (0.5–1 μg)

   • 2 of 10X Buffer

   • 0.25 of T7E1 enzyme

   • Add ultrapure H₂O up to achieve a final volume of to 20 μl

   Incubate at 37°C for 20 minutes to digest potential regions of mismatch repair.

4. Run a 2% agarose gel to detect mismatched amplicons after T7E1 digestion in step 3. The nuclease treated and non-treated sample are loaded side-by-side on the 2% agarose gel, stained with ethidium bromide then photographed under UV light. Samples with NHEJ mutations will be indicated the appearance of multiple bands (e.g. 200 bp and 400 bp in contrast to 600 bp of the uncut original PCR product). The wild-type control band, in contrast, will not have a mutation and should always generate the full-sized PCR product.

Note

The T7E1 enzyme mismatch assay may fail to detect single nucleotide mutations by NHEJ. Alternatively, the Surveyor^® enzyme mismatch cleavage assay is better at detecting single nucleotide changes but is less sensitive than the T7E1 assay (Vouillot et al. 2015).

Thorough sequencing analysis of the CRISPR-mediated indel should still be conducted to fully understand the nature of the NHEJ mutation. In-frame mutations are still possible with NHEJ that may not knockout the intended gene (Parikh et al., 2015). Unexpected exon skipping is also possible due to splicing. The T7E1 test for mismatches can also be applied to determine if there is any off-target damage.

Off-target mutations may occur during genome editing with CRISPR. The enzyme mismatch cleavage assay (listed above) can be used to identify possible unintended indels. Essentially, the MIT CRISPR software should identify possible off-target sites. A PCR strategy to amplify each location would then need to be developed for each site (genomic sequence and primers can be developed using BLAST). Off-target mutations are then tested for each identified founder.

Identifying Knock in mice by PCR based RFLP analysis

For genotyping knockin mutations, a unique restriction enzyme is generally engineered within the donor DNA to detect proper integration (as mentioned above) (Fig. 6). This allows for RFLP analysis of the PCR produce to determine if a restriction enzyme site was either introduced or ablated. Analogous to the enzyme mismatch cleavage assay, asymmetric placement of the restriction enzyme site within the PCR product allows for ideal detection of the mutated allele on an agarose gel. For insertions of protein tags or fluorescent proteins, a primer can be designed that corresponds to the inserted DNA sequence (internal + external primer PCR) (Harms et al., 2014). A corresponding wild-type primer is also essential to amplify non-disrupted sequence to identify heterozygous versus homozygous mice. For large deletions, primers should be designed further away (over 100 bp) away from the DSB sites because the subsequent repair may delete some of the nearby nucleotides (Williams et al., 2016). Sometimes, if using pairs of sgRNAs for conditional knockouts (loxP sites) or nearby genetic loci, a single large deletion can be generated between the two DSB sites. Such a large deletion might be missed by your PCR strategy (Parikh et al., 2015).

Note

The PCR conditions should be thoroughly tested before generating knockin mice so that a genotyping protocol is in place when the founder mice are born. Use wild-type genomic mouse tail DNA when optimizing the PCR and determine the adequate amount of DNA needed per PCR as too much or too little genomic DNA may result in reduced amplification.

COMMENTARY

Background Information

CRISPR-Cas9 technology provides an efficient means to generate knockout and knockin mice that is both less costly and laborious than regular gene targeting in ES cells. What started out as a genomic peculiarity (Barrangou and Horvath 2017) in bacteria and archaea has turned out to be one of the most preeminent genome editing tools currently available. Essentially, a genomic array of palindromic repeats and intervening spacers termed CRISPR was discovered in different bacteria and archaea. These spacers were later determined to be analogous to phage DNA, which thereby established CRISPR as a probable bacterial adaptive immune system. With identification of Cas9 as an RNA guided nuclease that participates in gene silencing, CRISPR genome editing was then adapted for use in eukaryotic cells to produce targeted DSBs at designated genomic loci (Cong et al., 2013; Mali et al., 2013). The next major step in CRISPR genome editing was to move from in vitro experimentation in tissue culture to generation of in vivo mouse models with targeted mutations (Wang et al., 2013; Yang et al., 2013). In conjunction with established transgenic microinjection techniques, Cas9 and sgRNAs are essentially delivered directly into the mouse zygote and transplanted into surrogate mothers for one-step generation of knockout mice. Insertions of point mutations and fluorescent tags were later conducted in mice using donor DNA as a template to repair the CRISPR generated DSBs. Along with providing a means to study gene function in mice, CRISPR has also been envisioned as a tool to eventually correct genetic disorders in humans. Wu et al. (2013), therefore, used CRISPR mediated HDR to repair a mutation that caused cataracts in mice in order to demonstrate the potential of Cas9 for gene correction. Since then, CRISPR has become an indispensable research tool that has allowed for accelerated development of new mouse models to both understand human diseases and determine gene function.

Besides direct delivery into the mouse embryo, other approaches to conducting Cas9 mediated genome editing in mice have been reported. Of note is a Cre-dependent Cas9 knockin mouse that has been used to conduct genome editing in neurons, immune cells, and the lung epithelia (Platt et al. 2014). A Cre-dependent conditional Cas9 mouse allows limited of genome editing to only a desired cell type. Subsequent injection of AAV (adeno-associated virus) vectors are then needed for delivery of sgRNAs and donor DNA. The delivery of multiple sgRNAs to Cas9 knockin mice readily allows for modeling of multi-genetic diseases in mice, like the formation of lung adenocarcinomas harboring three cancer causing mutations. While sgRNAs and donor DNA can be readily packaged into AAV vectors for delivery into Cas9 knockin mice, the packaging of Cas9 itself has been more problematic due to the size of its gene (~4.2 kb for Cas9 from Streptococcus pyogenes or 3.2 kb for the smaller Streptococcus aureus Cas9) (Komor et al. 2017). Codelivery of two AAV vectors, one encoding Cas9 and another with the sgRNAs, is therefore needed if attempting to develop CRISPR for gene correction, as seen in a mouse model of muscular dystrophy (Long et al. 2016, Nelson et al. 2016, Tabebordbar et al. 2016). Lastly, complicated genome editing can be difficult to achieve with CRISPR, particularly with large deletions and insertions, so an alternative to microinjection of a mouse zygote is to transfect ES cells with Cas9 and the sgRNAs (Oji et al. 2016). Large deletions and knockin mutations with a double stranded donor DNA occur at higher efficiency in ES cells as compared to a fertilized egg, and a large number of ES cells can be screened for the desired mutation instead of using up mice for a rare genome editing event. Germline transmission will still be an issue, but the cell autonomous functions of some genes can be analyzed with chimeric mice that are partly derived from ES cells containing biallelic mutations.

Beyond the DNA endonuclease capability of Cas9, CRISPR has also been repurposed to provide an RNA guided means of regulating the transcriptional activity of targeted genes. This can be achieved by using a deactivated Cas9 (dCas9) that is fused to transcriptional activators, repressors, and epigenetic modifiers (Komor et al. 2017). Alternatively, target gene activation with wild type Cas9 can also be achieved using modified dead sgRNAs designed with MS2 hairpin aptamers. These dead sgRNAs are shortened to prevent DSB formation while the MS2 aptamers recruit a fusion protein composed of the MS2 bacteriophage coat protein linked with a transcriptional activator (Liao et al 2017). This target gene activation strategy has been shown in mice to drive targeted expression of genes known to help ameliorate diseases such as diabetes, kidney disease, and muscular dystrophy. In contrast to DNA targeting with Cas9, another engineered CRISPR-Cas effector protein, Cas13, allows for RNA targeting in mammalian cells to knockdown RNA expression instead (Abudayyeh et al. 2017). Along with regulating gene expression in vivo, base editing has also been achieved in mice using a Cas9 nickase fused to a cytidine deaminase (Kim et al. 2017). CRISPR mediated base editing could allow for correction of single-nucleotide substitutions in humans without the concern over a potential indel mutation being generated in the human genome by Cas9. In conclusion, new modifications of CRISPR technology are being developed to repurpose the RNA guiding capabilities of Cas9 towards functions other than the creation DSBs.

Critical Parameters

As mentioned in Basic Protocol 2, the sgRNA efficiency is a predominant factor determining the success of your genome editing experiments. The parameters determining the efficiency of the sgRNAs still need to be worked out, however, so three or more sgRNAs should be selected for further testing. Ideally, the efficiency of the sgRNAs should be established in vitro before attempting to generate knockout or knockin mice. Along with sgRNA efficiency, another concern with genome editing is the potential of off-target mutations, particularly when trying to analyze a mouse phenotype. Most CRIPSR sgRNA design software provide a list of potential off-target sites, so, if possible, a PCR based strategy should be performed to detect any indel mutations in these sites using an enzyme mismatch assay. Current CRISPR software programs, however, are still unable to fully predict all of the potential off-target sites, so, unless whole genome sequencing is performed, a hidden CRISPR generated indel may still be present within the founder mouse. Breeding to a “clean” wild-type mouse should be conducted to segregate any unwanted mutations (preferably at least two crosses should be performed to minimize off-target mutations) (Raveux et al., 2017). Additionally, more than one founder line should also be expanded and analyzed to ensure the authenticity of any phenotypic finding. Mosaicism can frequently occur in the founder line due to a delay in Cas9 activity, so the tail DNA genotyping may not accurately reflect the genotype in the germline. Breeding to wild-type mice will generally separate out the mutations, but DNA sequencing of the target site then needs to be conducted on the offspring to fully confirm the exact nature of CRISPR generated mutation within the progeny.

This protocol has attempted to highlight some of the latest advancements in CRISPR genome editing in mice. Areas of improvement for CRISPR include better predictive software for determining sgRNA efficiency and potential off-target mutation sites. Research is also being conducted to identify factors that promote HDR over NHEJ as well, particularly since HDR occurs at a much lower rate than NHEJ. While most genome editing experiments in mice centers on Cas9 from Streptococcus pyogenes, other natural CRISPR systems have yet to become fully evaluated for possible genome editing in mice. For example, although the PAM of NGG for Cas9 is fairly common (about every 8–12 bases within the human genome), other CRISPR effector molecules with different PAM sites may become useful for gene targeting at other sites within the genome, such as AT rich regions. Engineered mutations within CRISPR effector molecules may also be needed to help relax PAM sequence requirements, improve specificity, and reduce off-target DNA damage (Komor et al. 2017; Slaymaker et al. 2016; Kleinstiver et al. 2016). Essentially, CRISPR is still an evolving technology, and, while this protocol attempts to cover most of the currently relevant options needed for generating knockout and knockin mice, there will probably be additional refinements to genome editing suggested in the future.

Troubleshooting

Generally, the sgRNA design is the main reason a CRISPR experiment does not work. An alternative sgRNA should then be selected. Ideally, the efficiencies of the sgRNAs should be tested, as described in Basic Protocol 2, before proceeding to microinjections and transfer of mouse embryos to foster mothers, particularly if attempting an HDR mediated knockin. If other sgRNAs still don’t work, consider the possibility that your genome editing might generate a deleterious mutation that results in embryonic lethality, a point of concern with any gene targeting experiment. In addition, verify that the targeting sequence selected for the sgRNA matches the mouse strain used for zygote microinjection as there could be polymorphisms. For CRISPR mediated HDR, the design of the donor DNA and the complexity of the designated knockin mutation may also affect the chance of getting a knockin allele, even when a selected sgRNA exhibits high NHEJ efficiency. Your genome editing strategy may, therefore, need to be reworked. For example, to generate a conditional knockout allele, a single long SS donor DNA containing a floxed exon/s may work more efficiently than using two sgRNAs to simultaneously insert loxP sites.

When conducting microinjections, know that your RNA CRISPR components can readily degrade, so be sure to spray your work area with RNase AWAY and use RNASIN RNase inhibitor in the microinjection solution. Wear gloves for all laboratory procedures to prevent RNase contamination and use RNase-free filter tips for pipetting. Ensure that all the CRISPR reagents such as donor DNA and sgRNAs are ultraclean and that the injection needle is not clogged with either the injection mixture or cellular debris from the zygotes. Confirm the yield and purity of your sgRNAs and donor DNA using a micro-volume spectrophotometer. During microinjection, the swelling of the pronucleus can act as a guide to measure the quality of the microinjection needle. Also, ensure that the injection pressure is set around 50 hPa with continuous flow (Cho et al., 2009).

Anticipated Results

The efficiency of CRISPR mediated genome editing can, in part, depend on the design of the sgRNAs. As seen in figure 4, a sgRNA with only 20% NHEJ efficiency in cultured injected mouse zygotes failed to produce any knockin pups by HDR. The type of designated mutation and genetic location also affects the efficiency of genome editing. For NHEJ knockout mutations, an overall average founder efficiency of 33% has been derived with various mouse strains (Qin et al., 2016). NHEJ occurs at a higher rate than HDR, however, so new techniques for promoting knockin mutations using CRISPR are being tested. With knockin mutations, large insertions tend to be more problematic to create in mice than smaller insertions equaling less than 1 kb in length. The creation of conditional floxed alleles is also known to occur at low efficiencies, particularly if two sgRNAs need to work effectively in tandem to insert loxP sites (Quadros et al., 2017). The efficiency of deriving the desired knockin mutation through HDR can, therefore, vary greatly depending on the complexity and the length of the insert, where point mutations show higher targeting efficiencies than large inserts and conditional alleles (Yang et al., 2014).

Time Considerations

CRISPR-Cas9 technology has largely replaced gene targeting in ES cells as a means of genome editing in mice, particularly due to the ease of construction and the reduced time required to derive founder mice. Knockin/knockout mice can be generated within 3 months, as compared to 8–10 months with conventional gene targeting techniques. After determining both the location and type of mutation desired, the sgRNAs are designed and synthesized to target a designated genetic locus, along with repair donor DNA if a knockin mutation is desired. The synthesis and purification of all of the necessary reagents for CRISPR mediated genome engineering generally takes between 1–3 weeks (Harms et al., 2014). As listed in Basic Protocol 2, testing of the sgRNA efficiency in mouse zygotes requires around 4–5 days to adequately culture the embryos before DNA extraction and enzyme mismatch analysis. CRISPR genome editing in mice essentially uses the same pronuclear microinjection techniques in transgenic mice derivation to deliver Cas9, the sgRNAs, and possible donor DNA. The collection of donor zygotes and transfer of injected embryos into pseudo-pregnant females has been extensively detailed in other protocols (Cho et al., 2009; Harms et al., 2014; Qin et al., 2016). To gather zygotes for microinjection, donor females are injected with PMSG (pregnant mare’s serum gonadotropin) and HCG (human chorionic gonadotropin), followed by mating with stud males. The microinjected zygotes are then transferred into pseudo-pregnant recipient female mice that have been mated to vasectomized males. A total of 4 days is needed from hormone injection to implantation surgery. About three weeks later, pups are born from the implanted embryos that will need to be genotyped from a tail snip in another two weeks. In general, the amount of time needed to derive of CRIPSR mediated genetically engineered mice, from concept to delivery of founder mice, takes roughly 3 months total (Harms et al., 2014; Qin et al., 2016).