5.1 Recombination that is not programmable #
5.1.1 Cre–lox and its relatives #
Cre from phage P1 recombines 34 bp loxP sites: 13 bp inverted repeats around an 8 bp directional spacer. It does not search for an arbitrary sequence. The sites must already have been inserted.
Same orientation on one molecule excises the interval as a circle. Opposite orientation inverts it. Sites on two molecules can integrate, but excision is favoured because it is intramolecular and reversible while Cre remains. A floxed marker can therefore be removed after selection, and a floxed gene deleted only where Cre is expressed. FLP/FRT is the same logic. Conditional alleles exist because a constitutive knockout may only say that the embryo died. High Cre can itself be toxic. Sequences of variant lox sites are not needed to understand the bias.
5.2 Protein-guided nucleases #
5.2.1 Zinc-finger nucleases and TALENs #
Both fuse a DNA-binding array to FokI, which cuts as a dimer. Two monomers must bind with the right spacing and orientation, so one binding event is a weak cutter and the target is long. Zinc fingers recognise about three base pairs and are context-dependent, so many arrays failed. TALE repeats from Xanthomonas are more nearly one repeat per base, and the proteins are large and repetitive. Both leave a break. The cell’s repair makes the edit. CRISPR replaced the protein array with RNA. It did not replace the repair problem.
The dimer is doing specificity work that a single binding domain would not. A monomer bound at a partial site does not cut unless its partner binds nearby and in the right orientation. That is why these reagents were built as pairs, and why each new target was a new protein rather than a new oligonucleotide. Delivery suffered for the same reason: the coding sequences are long and repetitive.
5.3 CRISPR–Cas #
The study of CRISPR–Cas provides fundamental principles governing cellular architecture, molecular interactions, and physiological regulation.
Crispr-Cas9 Mode Of Action – Ru
5.3.1 Guide, PAM, and the break #
CRISPR–Cas is bacterial adaptive immunity: a spacer from an invader is transcribed as crRNA. In Streptococcus pyogenes, crRNA plus tracrRNA guides Cas9. The laboratory single-guide RNA fuses the two. Jinek and colleagues showed the programmable nuclease in 2012; mammalian editing followed in 2013. Doudna and Charpentier received the 2020 Nobel prize.
SpCas9 also requires the PAM NGG on the non-target strand, next to the protospacer. The array stores spacers without cutting itself because of that requirement. Cas12a uses a T-rich PAM, cuts staggered ends, and can process its own array. HNH cuts the strand paired with the guide; RuvC cuts the displaced strand; the break is blunt, about three base pairs from the PAM. D10A and H840A each leave a nickase. Both mutations together leave dead Cas9. No guide sequences are given here.
5.3.2 NHEJ, HDR, base editors, prime editors, and dead Cas #
Non-homologous end joining works in most cells, including non-dividing ones, and often leaves a small indel. That is useful for a frameshift knockout and useless if you needed one specified codon. Scars differ between cells. Homology-directed repair copies a donor, so a designed allele can be written in, mostly in S/G2, and usually less often than NHEJ in mammalian cells. Without a donor, repair from the sister restores the original sequence.
| Wanted outcome | Route | Extra reagent | Limit |
|---|---|---|---|
| Frameshift knockout | NHEJ | None | Mixed indels |
| Precise change | HDR | Homologous donor | Often inefficient |
| C·G to T·A or A·T to G·C | Base editor | Nickase–deaminase | Window and bystander edits |
| Small indel or any substitution, no double-strand break | Prime editor | Nickase–reverse transcriptase and pegRNA | Large reagent; variable efficiency |
| Expression change only | dCas, CRISPRi/a | Repressor such as KRAB, or activator such as VP64 | Lasts while the protein is present |
A base editor deaminates a base in a short window on the displaced strand and nicks so repair keeps the edit. Other editable bases in the window change too. Prime editing copies the new sequence from the extended guide after a nick. Dead Cas blocks or, with a fused domain, represses or activates transcription without changing the sequence. It is not a heritable allele. These notes do not design guides.
5.3.3 Off-target cuts, assessed rather than hoped away #
Mismatches are tolerated better away from the PAM-proximal seed. Chromatin, dose, and how long the nuclease is present decide which similar sites are cut. A predicted list is not a clearance.
Targeted sequencing reads the shortlist. Unbiased methods recover breaks that actually occurred; GUIDE-seq, CIRCLE-seq, and DISCOVER-seq name that family and are not procedures here. The answer belongs to that guide, genome, and delivery. An expressed Cas9 gene cuts for longer than a ribonucleoprotein that is degraded. Paired nickases demand two events. “We used CRISPR” does not name the allele. Sequencing does.
5.4 Delivery, and who inherits the edit #
5.4.1 Named routes, not procedures #
A zygote can be injected. The founder is often mosaic, and breeding plus genotyping, not the injection, makes a clean line. Somatic routes include viral vectors, lipid nanoparticles, and ribonucleoprotein. AAV’s cargo, about 4.7 kb, is tight for SpCas9 plus a guide, and a lentiviral vector integrates. Chapter 9 takes up those vectors as medicines. The exposure also differs by place: the eye is a small compartment, and an intravenous dose aimed at the liver is not.
Plants are reached by Agrobacterium T-DNA, by a gene gun carrying DNA or ribonucleoprotein, or through protoplasts that are later regenerated. If the editor was DNA that does not stay in the genome, later generations can keep the edit and lose the editor. Whether that plant is then regulated as transgenic is a legal question for Chapter 8. None of this is a protocol.
5.4.2 Germline versus somatic #
A somatic edit that misses the germline ends with that person. The uncertainties — off-target alleles, immunity to Cas protein, mosaicism — are serious, and the person can usually be asked.
A germline edit, in a gamete or embryo, is heritable. Descendants cannot consent. Reversal is another breeding programme. An error becomes a whole-body fact about a future person. Efficiency does not close that question. Accidental gonadal exposure can make a “somatic” treatment germline; the distinction is where the edit is, checked by assay. Animal germline edits for research lines are not the human clinical case. Chapter 10 is the case in which the human line was crossed.
5.5 Key terms, summary, and self-check #
Key terms. Cre; loxP; floxed allele; FokI; zinc-finger nuclease; TALEN; PAM; Cas9; Cas12; nickase; NHEJ; HDR; base editor; prime editor; dCas9; CRISPRi; off-target; ribonucleoprotein; somatic edit; germline edit; mosaicism.
Summary. Cre rearranges DNA only at existing lox sites; orientation sets excision, inversion, or integration. ZFNs and TALENs programme a FokI dimer; CRISPR programmes a break with a guide and a PAM. NHEJ makes indels, HDR writes a donor, base and prime editors make small changes without a clean double-strand break, and dead Cas changes expression only. Off-target cuts are measured, not assumed away. Delivery into zygote, soma, or plant is named here and not specified. A germline edit is heritable; a somatic edit is not, unless it reaches the gonad.
Self-check.
1. What does Cre do to two loxP sites in the same orientation, and why is re-integration less favoured?
2. Why do ZFNs and TALENs need two monomers?
3. Which repair makes a frameshift in a non-dividing neuron, and why is the scar unpredictable?
4. What does a base editor avoid, and what error does its window add?
5. Why are the same Cas9 reagents in haematopoietic stem cells and in a human zygote different ethical categories?
—
—
—