Table 2 A quick guide to CRISPR for beginners ^a

1.

Prerequisites

• Delivery method, reaching the germline or other cells of interest: microinjection, transfection, electroporation

• Genomic sequence of target genes

• Robust phenotypic assays to determine the effect of gene targeting

2.

Experimental strategy

• Decide on the targeting approach (knock-in or knock-out), depending on whether you want to disrupt gene function, engineer a specific mutation, generate a reporter, etcetera.

• When testing CRISPR for the first time, choose a simple knock-out approach, selecting targets that produce phenotypes that are easy to score, such as pigmentation genes or a GFP transgene [70, 98, 122], or genes with a known, robust and specific phenotype.

For knock-ins

• To knock-in large constructs, use HDR templates in which the knock-in construct is flanked by homology arms - typically >1 kb in length - matching the sequences on either side of the double-strand break [46–49]; shorter homology arms give lower efficiencies [50]. Provide the template as a circular plasmid.

• To introduce small changes, use synthetic single-stranded oligos (ssODNs) bearing 10 to 40 nucleotides of homologous sequence at each end as templates for HDR [48, 50–52].

• The sequence targeted by CRISPR should be mutated in the repair template to protect the template and targeted alleles from cleavage.

• Alternatively, a homology-independent knock-in approach (see Figure 1) may be used to knock-in large DNA fragments [42, 45] or short double-stranded oligos (dsODNs) [44]. Using this approach, the insertion may be imprecise [45] or directed by complementary overhangs [42, 44].

• Select an approach that will minimize lethality due to NHEJ-mediated indels in somatic tissues, for example, by restricting CRISPR/Cas activity to the germline [48, 123], targeting constructs to introns, or adopting a strategy that improves the relative efficiency of knock-ins [42, 45, 50].

3.

Design of guide RNAs - finding target sequences

• Use the most reliable genomic sequence available for the target gene. Consider that the targeted site may bear nucleotide polymorphisms; if this is likely to be an issue, obtain sequences from the strain used for gene targeting and/or test multiple guide RNAs.

• Use online software to search for potential target sites (see Online Resources for CRISPR, below). The software search a given sequence for sites with a suitable PAM motif (NGG for S. pyogenes Cas9) and additional sequence constraints depending on the mode of guide RNA production (GGN₁₈NGG for in vitro T7-synthesis of guide RNAs, GN₁₉NGG for U6-mediated expression). The latter requirements can be relaxed, as extra Gs may be added to the 5′ end of the guide RNA without significantly compromising targeting efficiency [51, 88].

• When working with a sequenced genome, the software can also detect potential unintended targets and help select guide RNAs with fewer off-targets.

• Although the presence of the PAM sequence at the genomic target site is essential, it should not be included in the guide RNA (see Figure 2). For an N₂₀NGG target site, only the N₂₀ sequence is incorporated at the 5′end of the guide RNA.

• Design and test multiple guide RNAs, if possible, to control for off-target effects and because some guide RNAs fail (due to polymorphisms, RNA secondary structure or for unexpected reasons).

• Strategies to reduce off-target effects may require special design of guide RNAs: paired nickase approaches require pairs of target sequences offset by no more that 30 nucleotides on opposite DNA strands [44, 109, 117, 119, 124, 125]; truncated guide RNAs bear targeting sequences that are shorter than 20 nucleotides [88].

4.

Providing guide RNAs and Cas9

• Guide RNAs are easily generated by cloning pairs of synthetic oligos, corresponding to the two strands of the target sequence (determined above), into vectors carrying the invariable portion of the guide RNA (available at http://www.addgene.org/CRISPR). Cloning is facilitated by a restriction site on the vector - usually BbsI or BsaI, which does not constrain the cloned sequence - and compatible overhangs in the annealed oligos.

• The guide RNAs can be expressed either by in vitro transcription via the bacteriophage T3, T7 or SP6 promoters, or by in vivo expression via the U6 promoter. For initial experiments in species where U6 promoters and terminators are untested, choose in vitro synthesis of the guide RNA. Vectors and protocols can be found at http://www.addgene.org/CRISPR.

• Cas9 can be expressed from a helper plasmid carrying the coding sequence of Cas9 under the control of an appropriate promoter. Alternatively, if a promoter is unavailable for the species of interest, Cas9 can be provided in the form of in vitro transcribed capped mRNA or as purified recombinant protein [104, 105]. For initial experiments performed by microinjection, the use of recombinant Cas9 protein overcomes uncertainties with untested promoters and mRNA translation.

5.

Rapid assays of CRISPR activity and genotyping

• The melting curve and surveyor or T7E1 endonuclease assays are invaluable for a rapid assessment of CRISPR activity in new species, for routine testing of new guide RNAs prior to more laborious experiments, and for genotyping animals at specific target sites. These assays detect indels and other point mutations generated by NHEJ. They rely on PCR and require only a small amount of starting material.

• Genomic DNA is extracted from embryos or tissues to be tested and PCR is performed using primers that flank the target site. Untreated genomic DNA gives a PCR product with perfectly annealed strands (unless there are natural polymorphisms within the fragment), whereas mutagenized genomic DNA also yields some heteroduplex DNA, consisting of strands that differ by small indels and point mutations. The following assays are used to detect of these mismatches.

  - Surveyor/T7E1 endonuclease assays are based on cleavage of the heteroduplexes by a mismatch-specific endonuclease - either Surveyor or T7 endonuclease 1 [126, 127]. Cleavage products, indicating the presence of mispaired DNA, are detected by electrophoresis on an agarose gel. This is a sensitive detection method, best performed on 400 to 800 bp amplicons with target sites positioned near the middle.

  - The melting curve assay [128] relies on the fact that heteroduplex DNA has a lower melting temperature than the corresponding homoduplex fragments. That temperature difference, which is in the order of 1 to 2°C for 100 to 200 bp fragments, can be detected by performing melting curves in real-time PCR instruments with high temperature resolution.

• More specific PCR-based assays can be devised for knock-in approaches, employing pairs of primers that span the genomic locus and knock-in fragment.

• The PCR products can be cloned and sequenced to examine the nature and spectrum of corresponding mutations.

6.

Scoring phenotypes

• The effects of CRISPR targeting can be assessed in the animals where CRISPR was delivered (G0) or in their progeny. It is important to keep in mind that G0s are mosaics where only some cells are likely to carry alleles targeted by CRISPR; in the best cases a significant proportion of the animal shows bi-allelic targeting and a corresponding phenotype. The degree and distribution of targeted cell clones however are difficult to determine, unless a cell-autonomous marker is used (for example, targeting of some pigment genes, knock-in of GFP).

• If the germline of G0s has been hit, targeted alleles will be recovered in the next generation (G1). In contrast to G0s, G1 individuals are non-mosaic and may inherit one targeted allele (per locus) from the CRISPR-targeted parent. Animals may be genotyped by PCR (see above) and crossed to produce homozygotes and to maintain mutant lines.

• Choosing reliable, specific phenotypic assays and appropriate controls is crucial. Phenotypes may be subtle or show incomplete penetrance.

7.

Off-target effects and controls

• Unintended targets (off-targets) may be anywhere in the genome and are difficult to predict. Two strategies can help to overcome problems with off-target effects: appropriate experimental design allowing us to detect and account for off-target effects and approaches that improve the specificity of CRISPR.

• In most cases it is possible to control for off-targets by using different guide RNAs to achieve targeting; guide RNAs targeting different sequences are very unlikely to share the same off-targets. Alleles generated using different guide RNAs may be brought together by crossing, in heteroallelic combinations that are likely to complement off-target mutations.

• The specificity of CRISPR can be significantly improved by using paired nickases [44, 109, 117, 119, 124, 125] or truncated guide RNAs [88] (see main text).

• Off-target mutations will segregate away from targeted alleles in genetic crosses, unless they are linked on the chromosome.

8.

Online resources for CRISPR

• General

  http://www.genome-engineering.org/crispr

  http://www.addgene.org/CRISPR

  http://www.flycrispr.molbio.wisc.edu

  http://www.crisprflydesign.org

  http://groups.google.com/forum/#!forum/crispr

• Software for designing guide RNAs

  http://crispr.mit.edu

  http://www.addgene.org/CRISPR/reference/#gRNA

  http://tools.flycrispr.molbio.wisc.edu/targetFinder

  http://www.e-crisp.org/E-CRISP/designcrispr.html

  http://www.rgenome.net/cas-offinder

• CRISPR technology is recent and rapidly evolving. Online resources are likely to change, as improvements and new tools are introduced.