Editing pools versus single-cell clones

A bulk well the morning after editing is a crowd of genotypes sharing a medium. A single-cell clone is an attempt to make one genome grow until you can name it. Experiments go wrong when a pool measurement is written up as a clone, or when a clone that survived is treated as a fair sample of the pool. This overview is the decision between those objects. Cutting and repair are in how CRISPR-Cas9 editing works in research. The traps inside the word clone are in selecting edited clones without fooling yourself.

Culture vessels, guides and genotyping reagents are classes in the molecular biology catalogue. The biological sentence you need should be on the quote request before anyone plates a 96-well dish.

What the pool contains

After a typical nuclease delivery, the well holds unedited cells, cells with one allele scarred, cells with every copy scarred in different ways, cells that took the reagent and were not cut, and cells that are dying because the edit or the pulse was costly. A marker sort removes some of the never-transfected cells and leaves the genetic mixture. Deep amplicon sequencing is the method that counts that mixture. A Sanger trace of the bulk PCR collapses it into overlapping peaks and should not be forced into one genotype call.

Pools are the right object for several real questions. A screen that asks which guides deplete cells is a population statistic. An editing-rate comparison between two guides is a population statistic. A phenotype that is strong, cell-autonomous and visible above a background of unedited cells can be piloted in a pool, with the allele frequency measured in the same flask. Essential-gene dropout is inferred from who is missing, which only makes sense if you still have the guide counts and the controls a screen requires.

Pools are the wrong object when the sentence is "this protein is absent" or "this allele causes this phenotype". Unedited neighbours supply the protein. A subpopulation can supply the phenotype. Averaging them produces a half-effect that no single genotype possesses. Indels, frameshifts and protein loss is what that half-effect does to a blot.

What a clone can support, and what it costs

A clone expanded from one cell can, after sequencing of every copy you can account for, support a sentence about one genome. You can ask whether the protein is gone in that genome, whether a phenotype travels with the allele, and whether a cDNA puts the phenotype back. You pay in time, in incubator space, and in bias. Cells that grow poorly under the edit are under-represented. Cells that were doublets and included an unedited neighbour may grow especially well. The clones you meet are not a random draw from the pool unless you show they are.

Limiting dilution and single-cell sorting are the usual deposition methods. Both leave empty wells and both admit doubles. Record which you used. Outgrowth conditions are ordinary culture practice, described in mammalian cell culture for research labs. A line that will not grow alone cannot be forced into this design by plating harder. Change the question or change the model.

Genotype early, freeze early, and keep an unedited clone from the same plating as a sibling control. The sibling saw the same sort, the same medium and the same weeks in a well. It is a better control than the parental flask that never left the incubator. Sequence against the parental alleles on a named assembly from Ensembl. A variant that predates the edit is not a clone-specific mutation.

Branch before you dilute

Measure the pool. If the intended allele is rare, cloning is a needle search. Change the guide, the delivery or the repair plan, and measure again. Delivering editing reagents to cells is the exposure side of that branch. If the pool shows abundant indels and every completed clone is in-frame or wild type, frameshifted cells are likely dying. More plates will not conjure them. You need a conditional strategy, a different cell type, or a claim about lethality instead of a claim about a living knockout.

If the pool trace is already a single clean mutant sequence, you may still not have a clone. A dominant scar in a mixture can look clean on Sanger while a minority allele hides. Deep sequencing or a dilution to true single cells distinguishes a pure edit from a loud one. Do not skip that step because the chromatogram was pretty.

If the first clones are mixed at the cut, the deposition put down two cells or the PCR is contaminated. Do not sequence your way out of a doublet. Reclone from a fresh single-cell deposit. The assay order in checking whether a genome edit worked applies to each candidate, not only to the prettiest one.

Staying mixed on purpose

Some materials should stay pools. Primary cells with a short life, many organoid systems, and in vivo edits do not offer a single-cell expansion that means what a HeLa clone means. The honest design is allele frequency plus phenotype in the same population, with enough biological replicates to see the relationship. Short-read counting of amplicons is outlined as a technology class by Illumina. Phrase the result as a distribution.

A screen hit that you intend to study as a mechanism should leave the pool. Validate with a second guide, then with clones or with a rescue, or the hit remains a pool statistic. Staying in the pool because cloning is slow is a schedule decision. It is not a scientific substitute for the clonal sentence.

Off-target interpretation changes with the object. A rare off-target clone can dominate a phenotype in a pool and then vanish when you genotype "the" cells you kept, if you kept a different subpopulation. Off-target checks and their limits should be aimed at the same object as the phenotype, pool or clone, not at whichever was convenient to sequence.

Phenotypes that vanish, and phenotypes that appear only after cloning

If the pool showed a phenotype and every well-genotyped clone with the intended allele does not, the allele is not sufficient for the phenotype in this system. Causes that deserve a test are an off-target site present in the pool and absent from the clones, a stress response to transfection, and a subpopulation you failed to expand. A second guide that recreates both the allele and the phenotype is the cleanest rescue of the claim. More clones of the first allele, all negative, make the negative more solid.

The inverse also happens. A clone shows a phenotype the pool barely had, because the clone carries a passenger mutation or because the phenotype required a pure genotype to rise above noise. Independent clones separate those. One spectacular clone is an anecdote with a freezer location.

Do not "fix" a vanished phenotype by keeping the drug selection on so the editor remains, then attributing a new phenotype to the original allele. Continued cutting changes the genome again. Remove the nuclease, confirm the alleles are stable, and then measure.

Mixtures are still the regulated material

A pool of edited cells is not a lower biosafety case than a clone. It may be a less characterised one, which is a reason for care rather than for informality. Institutional approval covers the editing, the cell type and the delivery. This overview does not extend that approval, does not clear a diagnostic use, and does not describe human embryo or germline editing. Discard unused wells, including ones that look empty, under the same rules as the parent culture. The WHO laboratory biosafety manual is a general laboratory reference for that practice.

If the phenotype is unexpected growth or an unexpected toxin-like effect, stop and reread the risk assessment before you expand the clone because it is interesting.

Contamination, heat and a plate that only grew at the edge

Clonal plates spend weeks in an incubator. A power cut, a hot room that pushes the incubator past its set point, or a dried edge well will select for the cells that happened to be in a kinder well. That selection looks like a genotype cluster if you only sequence the survivors. After an environmental failure, start the deposition again. Cross-contamination between neighbouring wells during feeding is how a genuine clone becomes a mixture at passage three. Change tips. Date the lids.

Mycoplasma and other chronic contaminants are a culture problem, not an editing result. A phenotype that appears in every well including the unedited sibling is a reason to test the culture, not a reason to redraw the guide. Humidity on labels makes well A1 become well H1 in the notebook. The genotype table should be written from the plate map on the day of deposition.

What the enquiry should distinguish

Say whether you are genotyping a pool, a set of clones, or both. Include the cell type, whether it can grow from single cells, the locus, and the allele fraction you need to resolve. A pool with a five-percent edit and a clone with two alleles are different sequencing requests. Discuss them through the CRISPR validation sequencing enquiry reference as an enquiry reference. Do not read that page as a statement that cloning or sequencing is performed as an operated service. Culture context sits on the molecular biology pathway. Ask whether a quotation is possible for the reagents the chosen object requires. A request for knockout clones, when the cells will not grow alone, needs the population design instead.