Editing in bacteria versus mammalian cells

A protocol written for HEK293 cells, pasted under an E. coli strain list, fails in a specific way. The nuclease can still be Cas9. The cell that receives the break cannot do the same things with it. Editing in bacteria versus mammalian cells is a choice of repair logic, delivery, and biosafety, made before anyone orders a kit. How the protein finds DNA is in how CRISPR-Cas9 editing works in research. How either host's alleles should be read is in checking whether a genome edit worked.

A break is only a reagent

Cas9, pointed by a guide, cuts both strands next to a motif it accepts. What happens next is the host's repair, or the host's death. Mammalian cell lines commonly join the ends by non-homologous end joining and produce a mixture of small indels. Homology-directed repair can write a donor sequence into the break, mostly when the cell is in the part of the cycle where that machinery is available. Many experiments see end joining dominate and homology-directed outcomes as a smaller fraction. Design and verification have to expect that split.

A large set of bacteria used for cloning do not offer that end-joining route. Mycobacterium and some other species encode a Ku and ligase system that can join ends, so "bacteria" is not one repair genotype. For E. coli and many relatives, an unrepaired chromosomal break is lethal. Editing schemes therefore supply homology, often as a recombineering oligo or a double-stranded cassette, and let the nuclease remove cells that kept the original target. That is counterselection. The colony you pick has survived. Survival is not yet a sequenced allele.

Bacterial practice, at decision level

Recombineering systems, such as phage-derived proteins that promote annealing of a donor, are a method class. They are not a volume table, and the plasmid or strain you have carries its own instructions. The donor needs homology to the flanks of the change and should be designed on the accession you actually have. GenBank is a public place those accessions live. A guide designed on a different strain's PAM can miss, and then nothing is counterselected.

Delivery is usually transformation of plasmid DNA, sometimes with the donor as an oligonucleotide in the same step. Follow the competent-cell or electroporation instructions for that strain. Heat shock and electroporation are different physical insults. Do not borrow a mammalian lipofection recipe for a bacterial pellet.

After recovery and plating on the antibiotic the method specifies, screen colonies. Colony PCR across the locus, with primers outside the homology arms, asks whether the new junction is in the chromosome rather than only on a donor plasmid. Then sequence the product. A PCR size shift can be a deletion, an insertion, or a mixed colony. The culture habits around plates and antibiotics are in bacterial culture and transformation and in antibiotic selection and blue-white screening. Selection proves the marker. Sequencing proves the allele.

If no colonies appear, branch. A guide that cuts and a donor that cannot recombine yields an empty plate, which looks like a failed transformation. Include a control transformation without the cutting plasmid, or with a guide that has no target, so you can see that cells were competent. If that control grows and the editing plate does not, the counterselection may be working and the donor may be wrong. If neither plate grows, fix competence and DNA quality before you redesign the guide.

Mammalian practice, at the same decision level

In cultured mammalian cells the common product of a nuclease plasmid, an RNA, or a ribonucleoprotein is a population of indels at the target, plus untouched alleles. Clonal isolation, or deep reads of the pool, is how you stop calling that population a single genotype. Homology-directed knock-in needs a donor and a cell that uses it. Synchronising or timing delivery into S or G2 is a strategy some laboratories test. It is not a promise, and it belongs to the protocol of the cell line, not to a universal clock.

Delivery choices change the experiment. A plasmid that stays in the cell keeps expressing nuclease. A ribonucleoprotein is a shorter exposure. A viral vector may be required to reach some primary cells and brings a separate biosafety question. Lipofection mixtures are not electroporation pulses. Match the method to the cells. Background on handling those cultures is in mammalian cell culture for research labs.

Readout starts with primers outside the scar, an unedited sister culture, and sequence, not with a fluorescent marker on the delivery plasmid. The marker reports that something entered many of the cells. It does not report repair. If the knock-in is the claim, sequence the junctions. If only end joining was intended, sequence enough alleles to know the frame.

When a mammalian "no edit" result appears, branch differently from an empty bacterial plate. Cells usually survive an uncut well. Check transfection with the marker or with a known-cutting guide in the same cell batch. If the known guide cuts and yours does not, the failure is the guide or the chromatin, not the cuvette. If nothing cuts, fix delivery before you order more spacers.

Biosafety does not copy across the bench

A cloning strain with a chromosomal edit is a recombinant bacterium. A human cell line with a ribonucleoprotein may still fall under recombinant-DNA rules once a donor or a plasmid is involved, and a lentiviral preparation almost certainly changes the review. The WHO laboratory biosafety manual is general background. Your institutional biosafety committee decides the room, the waste, and whether the strain or the cells may be handled at all. Antibiotic markers that are routine in cloning are still a resistance gene. Do not treat "laboratory strain" as a synonym for harmless, and do not treat a mammalian knockout as automatically the same category as the parent line if the edit changes a hazard the approval cared about.

Work with pathogens, toxin genes, or human clinical material is outside the comparison this page can authorise. If the host is already under a higher containment decision, the editing method does not lower it.

Rooms, heat, and incubators

Bacterial plates dry out in a warm room and in an incubator that runs hot after a crowded week. A dried plate looks like counterselection. Log the incubator temperature when a transformation that used to work goes empty. Mammalian CO2 incubators that restart after a power cut need a recorded recovery of temperature and gas before a transfection is trusted. The two failures look similar in a notebook ("no edit") and have different causes.

Competent cells and mammalian recovery media follow the storage on their labels. A frost-free freezer cycle is a bad place for either. Write the strain genotype or the cell-line source next to the guide. A PAM present in the reference accession and absent in your strain is a design error, not a delivery error.

What the enquiry should distinguish

On the quote request, say bacterium or mammalian cell, the strain or line name, plasmid versus ribonucleoprotein versus another class, and whether you need a donor. Colony-PCR primers and mammalian locus primers are different oligonucleotides. Browse classes in the molecular biology catalogue. If sequenced verification is part of the plan, the CRISPR validation sequencing enquiry reference is a discussion prompt and an independent method reference. Ask whether a quotation is possible. The molecular biology pathway is the research context, not a statement of containment.

Do not send a mammalian microlitre table and ask for it to be scaled to a bacterial culture, or the reverse. Send the host first.