T7 endonuclease and amplicon sequencing

A band cut out of a PCR product by T7 endonuclease reports that two strands in the tube did not match. It does not name the mismatch. Amplicon sequencing names the bases and estimates how common each allele was. Laboratories get into trouble when they treat the first assay as a cheaper version of the second. This page explains the heteroduplex logic, the homozygous blind spot, and what a deep read adds. The edit those assays are trying to see is introduced in how CRISPR-Cas9 editing works in research. The wider standard for believing an allele is in checking whether a genome edit worked.

Polymerases, mismatch nucleases and library reagents are classes in the molecular biology catalogue. Say which readout you need on the quote request.

Heteroduplexes are the actual substrate

You amplify across the cut from a population that may contain wild-type and edited molecules. You melt the products and let them cool so strands find partners at random. Where a wild-type strand anneals to a mutant strand, the duplex contains a bubble or a mismatch. T7 endonuclease I recognises that distortion and cuts. On a gel you see the full-length amplicon, from homoduplexes and from uncut heteroduplexes, plus shorter fragments whose sizes add up to the amplicon if the cut was at the lesion.

Surveyor nuclease, from the CEL family, is the same idea with a different enzyme. It is often chosen when the lesion might be a single-base substitution, because T7 endonuclease I cuts many indels more readily than it cuts simple mismatches. Neither enzyme is a sequencer. A parental SNP that makes the amplicon heterozygous will produce fragments in an unedited sample. Always run the parent. A cut in the parent means the assay cannot be interpreted for editing until you move the primers or sequence through the SNP.

The fraction cleaved is not the fraction edited. If a fraction f of alleles is one sequence and the rest another, random reassortment puts about 2f(1-f) of molecules into heteroduplexes. The cleaved fraction tracks that quantity only if every heteroduplex is cut once and homoduplexes are spared. Real digests are partial. A square-root conversion is sometimes used to estimate f from the cleaved fraction. It is an estimate under those assumptions. More than two alleles, which is normal in a CRISPR pool, breaks the two-class picture. Use the gel to decide whether a mixture exists, not to publish a precise editing rate.

Primer position decides whether the fragments are visible. Primers far from the cut produce large pieces a standard agarose gel can separate. Primers very close to the cut produce a small fragment that can run off or hide in the primer-dimer region. Gel judgement of that sort is the same craft as agarose gel electrophoresis for DNA. Design the primers in NCBI Primer-BLAST so the amplicon is specific, then confirm the size with a marked ladder.

The uniform mutant that the gel calls wild type

If every allele in the tube is the same indel, melting and annealing recreates that indel as a perfect duplex. T7 endonuclease I has nothing to cut. A clone that is homozygous for a frameshift, or a pool that has been edited to completion with one dominant scar, looks like an unedited sample. This is the central failure of the assay as a clone screen. The fix is either to sequence, or to spike the clone amplicon with wild-type amplicon before the melt so that heteroduplexes are forced. A spike that produces fragments shows the clone differed from wild type. It still does not show how, or whether both copies match.

Two different indels and no remaining wild type will form heteroduplexes with each other and will cut. The lane looks "positive" and the interpretation "mixture" is fair. The interpretation "fifty percent edited" is not available from the band intensities without a model you have not earned.

Incomplete denaturation does the opposite. Strands that never separated never form new heteroduplexes, and a genuinely mixed pool looks negative. Follow the melt and the enzyme timing on the instruction sheet for the nuclease you have. Do not copy a minute and a unit number from a different vendor into the notebook. Over-digestion can nibble matched DNA and create bands that are not alleles. A no-edit parental digest, run in parallel, shows that background.

What a deep amplicon read gives you, and what it invents

Amplicon sequencing counts molecules of each sequence in the PCR product. You see the one-base insertion, the seven-base deletion, the unedited read, and their fractions, within the error of the method. That is the result a gel cannot supply. It is the right assay for a pool, for a heterozygote with two named alleles, and for any off-target site where the expected lesion is small. How those reads are produced as a class is outlined by Illumina and, as a laboratory path, in next-generation sequencing from library to reads.

The inventions are real. Polymerase errors during the many cycles of a specific PCR become rare indels and substitutions. Index swaps and chimeric amplicons move barcodes. A frequency near that floor is not an allele until the unedited control, sequenced beside it, lacks the same feature. Unique molecular identifiers distinguish molecules that were present in the cells from copies made in the tube. They are worth it when the biological frequency is low. They are optional when you are distinguishing a 40 percent indel from zero in a pool.

A clone sequenced this way should be nearly one or two alleles, not a cloud of one-percent scars. A cloud suggests the culture is mixed, the PCR is contaminated, or the error model was ignored. What to check before a sequencing run is the practical list before you blame the cells. Reads can be deposited in the European Nucleotide Archive when the project requires a public record. Deposition is not validation.

Large deletions that remove a primer site are invisible to both assays. Both will report the remaining allele and forget the one that did not amplify. If a clone shows a single wild-type trace and a missing protein, a larger lesion is still on the list. Move the primers outward or use a method that does not depend on that amplicon.

Choosing the assay for the sample in front of you

A first-pass pool, where you only need to know whether editing happened at all, can start with the gel if the expected lesions are indels and a parent control is on the gel. Move to sequencing before you rank guides by a supposed percentage or before you keep a clone. A base-editing experiment is a particularly bad fit for T7 endonuclease I, because the intended product is often a substitution the enzyme cuts poorly, and a successful uniform edit will not cut at all. Sequence base-edit windows.

If the gel is positive and the sequencing is wild type, believe the sequencing only after you have ruled out that you sequenced the wrong fragment. Gel-purify the uncut and cut species if they are confusing you, or redesign the primers. If the gel is negative and a spike with wild type is also negative, the locus you amplified is probably unedited, provided the PCR actually spans the cut. Say that as a limited negative.

Off-target nomination is the same pair of assays at other coordinates. A T7 screen across twenty sites is twenty semi-quantitative mixtures. It will miss uniform edits and single-base changes. For those sites, amplicon sequencing is the method that matches the claim in off-target checks and their limits.

Enzyme, light and contaminated amplicons

T7 endonuclease I is an enzyme with a sheet. Store it as that sheet says. The gel that reads it out may use a stain and ultraviolet light. Follow the safety note for the stain you use, and do not treat a research amplicon from a hazardous parent cell as harmless because it is short. Institutional biosafety covers the cells the PCR came from. This explainer is not a diagnostic clearance and does not describe embryo or germline editing.

The sharper everyday risk is contamination. Edited amplicons are the template that makes the next unedited control look edited. Physical separation of PCR setup from the gel room is the control, as in PCR controls and contamination control. A no-template lane that shows the band ends the day.

Warm rooms and shared cyclers

A melt step that is not hot enough fails more often when a cycler is overdue for service or has just rebooted after a power cut. Run a known mixed control, even a previously sequenced pool, when you doubt the block. In humid weather, labels on strip tubes smear. Mark the parent and the edited sample on the rack, not only on a lid that will be swapped. An enzyme left at the bench while a power cut resets the cycler is a reason to take a fresh aliquot. A weak digest is then an enzyme problem, not a false story about perfect repair.

What to ask when the gel is not enough

State the amplicon length, the primers, whether the sample is a pool or a clone, and whether you need a mixture screen or named allele fractions. If you are sourcing the nuclease, name T7 endonuclease I or a CEL-family mismatch nuclease as a class and follow that enzyme's own instructions rather than a copied table. Deep reads of the same amplicon can be discussed through the CRISPR validation sequencing enquiry reference. That page is an enquiry reference. It does not mean sequencing is operated as a service. The molecular biology pathway covers the PCR around it. Ask whether a quotation is possible. Do not send a gel photo and ask for a genotype without the trace or the reads.