Key takeaways

• Guide RNA design drives everything downstream — on-target efficiency and off-target risk are largely fixed once you pick the spacer.
• SpCas9 requires an NGG PAM immediately 3′ of the 20-nucleotide target; if no suitable PAM exists, change nuclease rather than compromise the guide.
• Delivering Cas9 and guide as a preassembled ribonucleoprotein gives fast, transient activity and the lowest off-target burden, because the nuclease is cleared rather than continuously expressed.
• Repair pathway determines outcome: non-homologous end joining produces indels and knockouts, homology-directed repair with a donor template produces precise knock-ins but is far less efficient.
• Always run a non-targeting guide and a Cas9-only control — without them you cannot separate an editing phenotype from a delivery phenotype.
• Verify at the DNA level before drawing conclusions at the protein level; sequencing with TIDE or ICE deconvolution quantifies editing in a bulk population without cloning.
• A knockout is not confirmed until the protein is shown to be absent, since in-frame indels and alternative start codons can leave a functional product.

Contents

• 1. Designing the guide RNA
• 2. Preparing guide RNA and Cas9
• 3. Delivering components into cells
• 4. Repair pathway and editing outcome
• 5. Selection and enrichment
• 6. Screening and verifying edits
• 7. Analysing the consequences of the edit
• Assessing off-target activity
• Controls and documentation
• Workflow summary
• Troubleshooting
• Reagents for validation
• Frequently asked questions

Reagents for validating an edit

Most CRISPR reagents are made in-house or ordered as oligos. What is usually missing is the validation layer — confirming the nuclease arrived, and measuring what the edit did to the cell.

SpCas9

Anti-CRISPR-Cas9 SpCas9 Monoclonal Antibody

MonoclonalWB / IF

Confirms SpCas9 is expressed before you spend weeks screening clones that never received it.

View product →

SaCas9

Anti-CRISPR-Cas9 SaCas9 Monoclonal Antibody

MonoclonalWB / IF

The equivalent check for SaCas9, the compact ortholog used where AAV packaging limits apply.

View product →

RNA prep

GenieZol Total RNA Extraction Reagent

RNA prepAll cell types

Total RNA for the qPCR step that quantifies transcript loss after a knockout.

View product →

Viability

MTT Cell Viability Assay

ColorimetricAll cell types

Establishes whether the edited population is viable before phenotypic conclusions are drawn.

View product →

Caspase-3/7

Caspase-3/7 Activity Assay Kit

FluorometricAll cell types

Detects apoptosis induced by knocking out an essential or survival gene.

View product →

PCNA

Human PCNA ELISA Kit

Sandwich ELISAHuman

Proliferation marker for assessing cell-cycle consequences of an edit.

View product →

1. Designing the guide RNA

The guide RNA directs Cas9 to a chosen genomic site through a 20-nucleotide spacer complementary to the target. This is the highest-leverage decision in the whole workflow: on-target cutting efficiency and off-target risk are both largely determined here, and neither can be rescued later.

What the target site must satisfy

• A valid PAM. SpCas9 requires a protospacer adjacent motif of NGG immediately 3′ of the target. No PAM, no cut — so PAM availability constrains where you can edit, not merely where it is convenient.
• Specificity. Sites differing from the target by only one to three mismatches, especially distal from the PAM, are plausible off-target substrates. Design tools score this; take the score seriously.
• Position within the gene. For a knockout, target an early constitutive exon — ideally within the first third of the coding sequence and present in every transcript isoform. Cutting a downstream or alternatively spliced exon frequently yields a truncated but functional protein.
• Sequence composition. Very high or very low GC content reduces efficiency, and long homopolymer runs, particularly four or more Ts, can terminate transcription of a plasmid-expressed guide prematurely.

Design tools

Benchling, CHOPCHOP and CRISPOR all generate candidate guides with on-target efficiency predictions and genome-wide off-target listings. Use at least two, take the guides they agree on, and order three to four per gene rather than one — guide performance varies substantially and unpredictably between adjacent sites, and testing several in parallel is far cheaper than repeating the experiment.

2. Preparing guide RNA and Cas9

Once designed, the guide and nuclease must be produced in a compatible format. The choice determines how long the nuclease persists, which in turn drives off-target accumulation.

Guide RNA formats

• Synthetic single guide RNA. Ordered as a chemically synthesised oligo, often with modified ends for nuclease resistance. Fastest route and works well for transfection or electroporation.
• Two-part crRNA and tracrRNA. Annealed before use. Cheaper for screening many guides, since only the short crRNA changes between targets.
• Plasmid-expressed guide. The spacer is cloned into a U6-driven expression vector. Necessary where continuous expression or viral packaging is required, and the only option for pooled library screens.

Cas9 formats

• Purified Cas9 protein, preassembled with the guide into a ribonucleoprotein complex. Active immediately, degraded within roughly a day, and consequently the cleanest option for off-target burden.
• Cas9 mRNA, giving transient expression without any risk of genomic integration.
• Cas9 plasmid or virus, giving sustained expression. Useful for hard-to-transfect cells and stable lines, at the cost of prolonged nuclease exposure.

Whichever format you use, confirm the nuclease is actually present before screening clones. A western blot against Cas9 distinguishes a failed delivery from a failed edit — two problems with very different remedies that look identical in the sequencing data.

3. Delivering components into cells

Delivery is where most CRISPR experiments actually fail, and the right method depends far more on the cell type than on the edit.

As a general rule, use ribonucleoprotein delivery by electroporation wherever the cell type permits. It gives a sharp pulse of activity, avoids integrating foreign DNA, and produces the most interpretable result. Reserve viral delivery for cases where nothing else works or where stable expression is genuinely required.

4. Repair pathway and editing outcome

Cas9 creates a double-strand break; the cell decides what happens next, and that decision determines your outcome. This step is absent from most workflow summaries and is the usual reason a knock-in attempt yields only knockouts.

Non-homologous end joining is the dominant pathway. It religates the break imprecisely, introducing small insertions or deletions. Where an indel shifts the reading frame, the result is a premature stop codon and a functional knockout. This is efficient and requires no donor, which is why gene disruption is the easiest CRISPR application.

Homology-directed repair uses a supplied template to copy in a defined sequence, enabling point mutations, tags and reporter insertions. It is much less efficient, operates mainly in dividing cells during S and G2 phase, and needs a donor — a single-stranded oligodeoxynucleotide for small edits, a plasmid or viral template with longer homology arms for larger insertions. Expect to screen many more clones than for a knockout.

If a precise sequence change is the goal, decide the pathway before designing the guide, because the cut must sit close to the intended edit — typically within tens of base pairs — for homology-directed repair to work at usable frequency.

5. Selection and enrichment

After delivery the population is mixed: unedited cells, monoallelic edits, biallelic edits and a range of indels. Enrichment raises the proportion worth screening.

Enrichment approaches

• Antibiotic selection, where the construct carries a resistance marker. Simple, but selects for delivery rather than for editing — a resistant cell is not necessarily an edited one.
• Fluorescence-activated cell sorting, where a fluorescent reporter is co-delivered. Sorting the brightest fraction enriches for cells that received most cargo, which correlates with editing.
• Single-cell cloning by limiting dilution or single-cell sorting into plates. Essential for a clonal line, since a bulk edited population is genotypically heterogeneous and unusable as an isogenic model.

For a defined cell line, sort single cells, expand, and genotype each clone. Do not assume a well is clonal because one cell was sorted into it; confirm by sequencing, and keep an early-passage frozen stock of every validated clone before extended culture introduces drift.

6. Screening and verifying edits

Verification proceeds from DNA outwards. Confirming the genomic change first prevents a great deal of wasted effort chasing phenotypes that were never caused by an edit.

At the DNA level

• PCR across the target site is the starting point. For large deletions or insertions the product size shifts visibly on a gel; for small indels it will not, so PCR alone is insufficient.
• Restriction enzyme digestion works where the edit creates or destroys a recognition site. Quick and cheap, but only applicable opportunistically.
• Sanger sequencing with deconvolution is the practical workhorse. Sequencing the bulk PCR product and analysing the trace with TIDE or ICE quantifies the spectrum of indels in a mixed population without cloning — the fastest way to know whether a guide worked at all.
• Amplicon next-generation sequencing gives precise allele frequencies and detects low-frequency events that Sanger traces miss. The right choice for a definitive answer or a therapeutic context.

At the RNA and protein level

A frameshift usually triggers nonsense-mediated decay, so quantitative PCR on extracted total RNA should show reduced transcript. This is supportive rather than conclusive, since transcript can persist.

The decisive test for a knockout is absence of protein, assessed by western blot with an antibody raised against a region downstream of the edit. This matters because in-frame indels, exon skipping and alternative start codons can all leave a partially functional protein behind an apparently convincing genomic edit. A clone is only a knockout when the protein is gone.

7. Analysing the consequences of the edit

With the genotype established, the biology can be addressed. Choose assays matched to the gene’s expected function, and include the phenotypic basics regardless.

• Viability, to establish that the edited population is healthy. Knocking out an essential gene depletes the culture and can masquerade as a failed edit.
• Proliferation, for cell-cycle and growth-regulatory genes.
• Apoptosis, where the target has a role in survival or cell death signalling.
• Pathway-specific readouts, such as cytokine secretion, phosphorylation state or reporter activity, according to the hypothesis.

Where a phenotype is strong, the most rigorous confirmation is rescue: reintroduce the gene as a construct resistant to the original guide and show the phenotype reverses. That distinguishes a genuine consequence of the edit from clonal variation or an off-target effect, and it is the standard reviewers increasingly expect.

Assessing off-target activity

Off-target cutting is the principal technical criticism of any CRISPR result, and the level of assurance required scales with the stakes.

For routine cell-line work, sequencing the top handful of computationally predicted off-target sites is usually adequate, combined with two or more independent guides against the same gene. If distinct guides produce the same phenotype, an off-target explanation becomes improbable, since they have different off-target profiles.

For therapeutic or otherwise high-consequence work, unbiased empirical methods such as GUIDE-seq or CIRCLE-seq identify genuine off-target sites without relying on prediction. Two design choices also reduce the problem at source: deliver as ribonucleoprotein so the nuclease is transient, and consider a high-fidelity Cas9 variant, which retains on-target activity while substantially reducing tolerance of mismatches.

Controls and documentation

Controls to include

• Non-targeting guide RNA — a guide with no genomic target, delivered identically. This is the correct comparator, since it controls for delivery, nuclease activity and the stress of the procedure.
• Cas9 without guide, to separate effects of nuclease expression from effects of cutting.
• Mock-treated cells, carrying the delivery procedure with no cargo.
• Multiple independent guides per gene, which is the single most effective safeguard against over-interpreting an off-target effect.

What to record

Reproducibility depends on detail that feels excessive at the time and is irreplaceable later: exact guide spacer sequences with genomic coordinates and genome build, nuclease format and supplier lot, delivery parameters including electroporation programme and cargo amounts, passage number at editing, the full genotype of every clone with its sequencing traces, and the passage at which each frozen stock was banked.

Workflow summary

Troubleshooting

• No editing detected. Establish first whether delivery or cutting failed — blot for Cas9. If the nuclease is present, the guide is likely poor; test alternatives rather than optimising conditions around a bad guide.
• Editing in bulk but no edited clones. Often means the gene is essential, so edited cells are outcompeted. Consider an inducible system or a conditional approach.
• Only monoallelic edits. Increase cargo dose or extend exposure, and screen more clones. Some loci are persistently refractory on one allele.
• Knock-in fails while knockout works. Expected. Move the cut closer to the intended edit, lengthen homology arms, and enrich for cells in S or G2 phase.
• Genomic edit confirmed but protein still present. Suspect an in-frame indel, exon skipping or a downstream start codon. Sequence the transcript and reassess the guide position.
• Phenotype varies between clones. Likely clonal variation rather than the edit. Compare several independent clones per genotype and confirm by rescue.

Frequently asked questions