Designing gene-specific CRISPR guides

1. Go to http://www.ensembl.org/ and search for your gene of interest in your organism of interest. As an example, consider the mouse activation-induced cytidine deaminase, AICDA or AID.

While the same information will be found in NCBI and UCSC databases, the Ensembl database is graphically more intuitive and pleasant. If you are otherwise familiar with genome databases, skip down to step 5.

2. Click to the relevant gene page. Note that there may be multiple transcripts for a gene, but usually there will only be one annotated protein-coding transcript with other transcripts being spurious. In the case of AID, there are two protein-coding transcripts, which suggests that targeting exon 4 may be necessary.

3. Import your desired gene into Benchling by clicking the "+" button on the left sidebar, then Sequence > Import DNA > Search External Databases. Input the accession number for your gene of interest, e.g. ENSMUSG00000040627 for Aicda, and search. It does not matter too much what mouse genome you choose, but CH₁₂ cells are derived from B₁₀ mice, and therefore either mm9 or mm10 genomes based off of B₆ mice would probably be closest. It is important, however, to choose which transcript you want to import (see step 2).

4. Select an early coding exon (it should have a translated peptide track), common to all isoforms if applicable, for sgRNA analysis. Click the bulleye's eye on the right sidebar and click "Design and Analyze Guides". The default settings for sgRNA analysis should be fine, so click finish and press the "+" to add your selected exon to analysis. Under the Genome Region column, there should be a "No region set" link-- click it. Find the best genome match, then set it so that the analysis disregards your locus of interest for the purpose of off-target analysis. If you do not do this step, your sgRNA will likely score lower on off-target analysis than necessary.

5. Pick a CRISPR guide that, ideally, has good off-target and on-target scores.

As a general guideline, pick 4-6 CRISPR guides for a gene (or DNA sequence) of interest. Not all sgRNA will work, possibly in part due to "DNA accessibility" issues. Therefore, space out CRISPR guides and consider targeting more than one exon if necessary. Additionally, if there are no sgRNA that have both good off-target and on-target scores, consider weighing what is more important to your experiment, e.g. off-target mutations may not be as big of an issue as generating any sort of on-target mutant. More generally, the gold standard of demonstrating the direct relationship between a specific mutatation and phenotype is a rescue experiment.

Make sure your guides do not have a BbsI site; see step 16. Finally, if there is a suitable guide that overlaps a restriction site for which we have an enzyme, prioritise it (see step 29).

6. With your 20 nt CRISPR guide, add 5' overhangs for cloning. For the "sense" CRISPR strand, add "CACC" to the 5' end. If your target does not start with a guanidine, add a 5' G to make a 20+1 nt guide RNA; a 5' G improves U₆-mediated transcription, and a 21 nt guide will function like a 20 nt guide. For the "antisense" strand, reverse complement the "sense" CRISPR strand and add "AAAC" to the 5' end. It is worth noting that the Joung lab has demonstrated that truncated sgRNA can retain activity while decreasing off-target activity, so that may be something to look into.

For the AID example, look at guides #1 and #2. Note that #2 does not start with G, so add a 5' G!

CRITICALDO NOT INCLUDE THE PAM MOTIF INTO THE GUIDE. If your guide is 23 or 24 nt, look back because you dun screwed up.

Designing primers for mutant screening

7. In parallel with CRISPR guide design, one will need to design primers to amplify the genomic region that will contain the expected indel. There are many primer design softwares out there, but Primer3 is always a good choice. Whatever your method, check your primers with Primer-BLAST against your organism's genome. Obviously, do not use primers that have many predicted amplification products within range of your desired product.

8. CRITICALIt is highly recommended to do a temperature gradient PCR with your screening primers before you start screening clones, as one will need to move quickly with cells in culture. A gradient between 55 and 70 degrees should identify an optimal annealing temperature with minimal off-target amplification; also, on-target amplification can sometimes be better with different Taq buffers (between our homebrew Standard, ThermoPol, and LongAmp buffers, at least one usually works). For really difficult templates, consider using Phusion rather than Taq polymerase. Some loci, for whatever reason, will prove impossible to amplify; if you think this is the situation, design new sgRNA at a different locus.

Cloning CRISPR guides into pX330

72 h

9. Dissolve your CRISPR guide oligos to 100µM in water.

10. [Optional] The T4 PNK reaction adds a 5' phosphate to the oligos for cloning. However, since the cut vector will have 5' phosphates, the T4 PNK reaction is not strictly necessary. I've only included this step as a holdover from the Zhang protocol.
    30 m

CRITICALDo not use T4 PNK buffer as this does not have ATP; T4 ligase buffer is essentially the same composition but with ATP. Use the newest tube of T4 ligase buffer as DTT tends to oxidise over time and with freeze/thaws. Incubate reaction at 37 Celsius @ 30 minutes. If you would like to skip over to the annealing step (step 15), you still need to add the oligos into a tube with some salt in it (e.g. ligase or PCR buffer).

11. Anneal oligos together by boiling at 95 Celsius @ 5 min, then cool to approximately room temperature. In my experience, any protocol for labannealing will work, likely even immediate cooling on the bench since these are short DNA products.

PAUSEAnnealed phosphorylated oligos can be stored at -20 Celsius.

12. The Zhang protocol recommends combining digestion with BbsI with ligation of annealed-phosphorylated oligos into the vector. In my (Alex) experience, complete digestion of pX330 takes some effort (3 days of overdigestion-- 1µg in 100µL digest, 10U BbsI each day), as the two BbsI cut sites are close together; thus, a second BbsI cut is probably inefficient once the first BbsI cut generates a linear end. The combined digestion/ligation is fairly efficient, but if you anticipate lots of cloning, it will save time and reagents to overdigest your vector and proceed with a normal ligation.

While the Zhang lab protocol suggests using 1µL of a 1:250 dilution of annealed-phosphorylated oligo, undiluted oligo works fine.

13. Incubate digestion/ligation reaction with the following program:
    5 h

Note that while the Zhang lab protocol says 6 cycles (1 hr total), but a longer cycling tends to improve cloning efficiency (from 60% to near 100%). Transform 50µL chemically competent bacteria with 5µL and plate onto LB-ampicillin plates overnight.

14. Screen transformants the next day by colony PCR, using the human U₆ sequencing primer (GATACAAGGCTGTTAGAGAGATAATT, also for pLKO shRNA sequencing) and the relevant antisense (crick) oligo.

Cloning efficiency should be greater than 80%.

15. Pick a clone, make a glycerol stock, miniprep the CRISPR construct and sequence the insert with the U₆ sequencing primer. If you're lazy and/or have lots of different cloning reactions, sequencing isn't strictly necessary since the coupling efficiency in oligo synthesis shouldn't be a huge issue with short oligos, and you're testing multiple sgRNA anyway, right?

Transfecting and cloning CH₁₂F₃-2 cells

16. Culture CH₁₂F₃-2 cells such that they are in the logarithmic growth phase on the day of electroporation. Typically, seeding 1 million cells in 10mL CH₁₂F₃-2 media the day before works well. You will need 1 million cells per electroporation.

17. Resuspend 1 million CH₁₂F₃-2 cells in 600µL ZAP buffer, and combine with 5µg CRISPR construct in 4 mm electroporation cuvette.

18. Electroporate in BioRad GenePulser Xcell with following conditions:

Immediately place electroporated cells on ice and incubate for 5 min, then plate in 10mL CH₁₂F₃-2 media and incubate @ 37 Celsius/5% CO₂.

19. 72 hr after electroporation, serially dilute electroporated cells to 10-50 cells/mL and multi-channel pipette 100µL into 96-well flat bottom plates. Typically, 10 x 96-well plates @ 20 cells/mL yields around 100 clones, perfect for a 96-well PCR block. Sorting single cells should also work, but is more expensive!

An accurate and precise count with a hemocytometer is required; for example, a count of a dozen cells in one field is not accurate. Do not serially dilute at more than 1:100 per dilution.

While in theory 10 x 96 wells @ 0.2 cells/well (20 cells/mL) should result in 192 clones, you will get less likely due to adsorption of cells to plastic, or death due to stresses of living alone; however, you will usually get approximately 100 clones. You can screen less clones, but 96 clones is definitely manageable starting off, and screening two constructs in parallel (2 x 96 clones) is feasible with practice. Sometimes the cloning efficiency does go up or down for whatever reason, so the best concentration may need to be determined empirically on your end.

Note that a 72 hour rest post electroporation may not be optimal for all genes, especially for genes where deficiency causes a growth defect.

20. Incubate clonally plated cells in the long-term 37 Celsius incubator, to reduce water loss, for 7 to 10 days. Clones should be observable by microscope ~5 days, and by eye ~7 days. There is some timing flexibility with step 21 and onwards.

Screening CH₁₂F₃-2 clones

48 h

21. Typically at~9-14 days, clones should be confluent enough to take a fraction to make genomic DNA (gDNA). Since the screening is done with PCR products, you only need 0.5-1k cells to generate gDNA; 0.5 to 0.9 volume of cells from a well should be more than sufficient. Spin down these cells within 0.2mL PCR tubes @ 300 x g for 5 min.
    2 h

In parallel, cells can be transferred to a new vessel for continued growth. Clones can be conveniently labelled by any combination of plate number, row heading, and column number.

PAUSEClones can be frozen en masse in V- or U-bottom 96-well plates by pelleting and resuspending in 100µL freezing medium, and then storing in -80 Celsius. This is useful if you encounter problems in your PCR.

22. Aspirate media and resuspend cell pellet in 5-10µL proteinase K digest.
    1 h, 30 m

Incubate @ 65 Celsius for 60 min. Like tail digestion @ 55 Celsius, this step can be done overnight as well.

23. Heat inactivate proteinase K @ 95 Celsius for 15 min.
    15 m

PAUSEStore gDNA in 4 degree fridge.

CRITICALProteinase K will cleave Taq if not inactivated. The initial denaturation step of a PCR will not be sufficient, thus inactivation and PCR should occur separately.

24. Use 1-2µL of gDNA in a standard 10-20µL PCR reaction with the relevant screening primers (see steps 7-8). Run 3µL from several wells to confirm PCR product before proceeding on.
    2 h

PAUSEPCR product can be stored @ 4 or -20 Celsius.

25. If your CRISPR cut site overlaps a restriction site, digest 3µL PCR product. Clones where there are no digest products should have a mutation in all alleles of your target gene. In principle, one should also be able to distinguish clones with only one mutant allele by band intensity. The restriction enzyme screening approach is robust and may be preferable to the mismatch cleavage assay (steps 30-31), but both give complementary sets of information. Note that not all genes may have two alleles, especially in cell lines-- for example, U₂OS is known to be near triploid.
    1 h, 30 m

Restriction enzyme (SacII): digest products mean wildtype alleles exist. By intensity, lane #1 is probably "heterozygous", lane #2 is probably wildtype, lane #3 is probably "homozygous mutant".

T7EI mismatch cleavage: cleavage products mean at least one allele is mutated. While less robust than the restriction enzyme example above, band intensity can also suggest full genotype. Note that the information in both assays are complementary.

26. If proceeding with the mismatch cleavage assay, prepare wildtype PCR product in advance (3µL per reaction). First, anneal wildtype and clone PCR product together:
    30 m

Boil @ 95 Celsius and slowly cool to 37 Celsius (e.g. -5 Celsius/min in a thermocycler).

N.B. In principle, mutations in both (or more) alleles of a target are likely different, thus one could technically hybridise mismatch products together without adding WT product and save time and reagents. That said, homozygous mutations do happen often enough to notice. Caveat emptor!

27. Add either Celery Juice Extract (CJE) or T7 endonuclease I (T7EI) to annealed PCR duplexes:
    1 h

Incubate CJE mismatch cleavage reaction @ 45 Celsius for 15 min.

N.B. Stop reaction on ice and add loading dye containing EDTA.

Incubate T7EI mismatch cleavage reaction @ 37 Celsius for 1 hr.

N.B. one can add more T7EI and digest for a shorter period, e.g. 10 unit for 15-30 minutes.

28. Run reactions from step 29 and/or steps 30-31 on a 1.5% agarose gel.
    30 m

Do a celebration dance! Or sulk in failure, depending.

29. For all candidate mutant clones, confirm by protein detection (preferable), quantitative PCR (somewhat less useful as nonsense mediated mRNA decay isn't always apparent), and/or sequencing. In some respects, sequencing may be easier than protein or qPCR detection, especially with TIDE or Indigo, which can deconvolve the signal from sequencing PCR amplicons to help identify mutants. To use TIDE, send PCR amplicons from candidate mismatch-cleavage positive clones for sequencing, along with a wildtype control amplicon, and follow website instructions; Indigo does not necessarily need a control reaction, as it can compare to the reference genomes, but it's useful to send one anyway. The results should identify which clones have biallelic mutations (or multiallelic mutations depending on the cell line and locus), as well as whether such mutations are frameshifted. N.B. Indigo doesn't seem to deal with triallelic genes well from personal experience.

Protocol Modifications

2015-09-24 Cloning instruction modification.

2015-09-10 Added sgRNA designer and CRISPRscan links, removed dead crispr.mit.edu job links.

2016-02-26 Streamlined protocol all over.

2016-07-14 Deprecated steps 5-7, emphasized Benchling tools.

2017-05-05 Streamlined protocol again by keeping most of the instructions within the Benchling ecosystem, and adding TIDE instructions.

2017-07-08 Small clarifications.