- Why Standard Cas9 Needs Better Alternatives
- The Base Editing Concept
- Cytosine Base Editors (CBEs): C→T
- Adenine Base Editors (ABEs): A→G
- The Editing Window: Precision and Its Limits
- Base Editing Off-Target Effects
- Prime Editing: Search and Replace for DNA
- pegRNA Design: The Key to Prime Editing
- CRISPRi and CRISPRa: Editing Without Touching DNA
- The Story: The Precise Calligrapher
Section 1 — Why Standard Cas9 Needs Better Alternatives
Standard CRISPR-Cas9 is transformative. But when it comes to correcting specific point mutations that cause genetic disease, it has a fundamental problem: it breaks the DNA first and then hopes the cell repairs it correctly. This is like fixing a typo in a manuscript by cutting out the sentence with scissors and gluing in a corrected version — it works, but the gluing is unreliable, and sometimes the repair introduces new errors.
For disease-causing point mutations — single base changes like the A→T mutation in sickle cell disease or the thousands of other pathogenic variants catalogued in ClinVar — you do not want to create a DSB and hope for HDR-mediated correction. HDR efficiency is low (1–10%), requires a donor template, only works in dividing cells, and is outcompeted by NHEJ which produces indels instead. For single-base correction in non-dividing cells (neurons, muscle, heart cells), standard Cas9 plus HDR essentially does not work.
What was needed was a way to change one specific DNA base to another without breaking the helix — more like using a molecular Tipp-Ex and pen than a pair of scissors. That is what base editors and prime editors provide. They were both developed primarily by David Liu’s lab at the Broad Institute, and they represent the most significant advance in gene editing precision since CRISPR itself.
| Tool | DSB? | What It Can Edit | Limitations |
|---|---|---|---|
| Cas9 + NHEJ | Yes | Gene knockouts (random indels) | Cannot make precise edits; unpredictable outcomes |
| Cas9 + HDR | Yes | Any sequence change (with donor template) | 1–10% efficiency; dividing cells only; NHEJ competes |
| CBE | No (nick only) | C•G → T•A transitions | Only 4 of 12 mutation types; limited editing window |
| ABE | No (nick only) | A•T → G•C transitions | Only 4 of 12 mutation types; limited editing window |
| Prime Editor | No (nick only) | All 12 substitution types + small indels | Lower efficiency than CBE/ABE; complex pegRNA design; large cargo |
Section 2 — The Base Editing Concept: Chemistry Instead of Scissors
Base editing uses a fundamentally different approach to DNA modification: instead of cutting the DNA and relying on repair, it uses a chemical enzyme to directly convert one DNA base to another at the target site. The base editor does not break the phosphodiester backbone of either DNA strand — it performs an in situ chemical reaction that transforms one base into a different one while everything else remains intact.
A base editor is a fusion protein with three functional components working in concert. First, a catalytically impaired Cas9 (either nCas9 with only the RuvC domain active, or dCas9 with neither nuclease active) that finds the target sequence and opens the DNA locally but does not create a DSB. Second, a deaminase enzyme that chemically modifies the exposed single-stranded DNA in the R-loop bubble. Third, optionally, a uracil glycosylase inhibitor (UGI) that prevents the cell from repairing the deaminated base before it is permanently incorporated.
The key to why this works is the R-loop. When nCas9 binds its target sequence and the guide RNA forms its heteroduplex with the target strand, the non-target strand is displaced as single-stranded DNA. Single-stranded DNA is the substrate for deaminase enzymes. The deaminase domain of the base editor acts on this exposed single-stranded region, converting the target base while it is temporarily accessible — and then nCas9 releases, the DNA re-anneals, and the edit is locked in when the cell replicates the DNA.
Section 3 — Cytosine Base Editors (CBEs): Converting C to T
Cytosine Base Editors (CBEs) were the first base editors developed, introduced by David Liu’s lab in 2016. They convert cytosine (C) to thymine (T) — or equivalently, a C•G base pair to a T•A base pair. This single conversion can correct a significant subset of all known pathogenic point mutations, because C→T (and its reverse complement G→A) transitions are among the most common mutation types in human disease.
The CBE Mechanism: Deamination to Uracil to Thymine
The chemical reaction at the heart of CBE is cytosine deamination: the deaminase enzyme removes the amino group (-NH&sub2;) from cytosine, converting it to uracil (U). Uracil is not normally found in DNA — it is an RNA base — but when it appears in DNA it is read by the cell’s replication machinery as thymine. So after the next round of DNA replication, the original C•G pair has become a U•G mismatch, which resolves to T•A. The C has been permanently converted to T.
removes –NH₂
repair; replication
The UGI role: Without UGI (Uracil DNA Glycosylase Inhibitor), the cell would quickly repair the U back to C, removing the edit. The UGI domain of the CBE fusion protein blocks the cellular repair enzyme UDG, preventing this correction and giving the edit time to be fixed by DNA replication. Multiple UGI units (tandem UGI) are used in the most efficient CBE designs (BE4max, ABEmax).
Which Diseases Can CBEs Correct?
CBEs can correct any pathogenic mutation involving a C→T change (or G→A on the opposite strand). This covers a substantial fraction of known disease variants. Examples include: the LDLR mutations causing familial hypercholesterolaemia (some variants), premature stop codons that could be converted to sense codons, and splice site mutations. The Liu lab’s ABEmax CBE variant has been used in preclinical models to correct Hutchinson-Gilford Progeria syndrome (a C→T mutation in LMNA) and has entered clinical trials for certain heart conditions.
Section 4 — Adenine Base Editors (ABEs): Converting A to G
Adenine Base Editors (ABEs) convert adenine (A) to guanine (G) — or equivalently, A•T to G•C base pairs. Published by the Liu lab in 2017, ABEs were a more remarkable engineering feat than CBEs for a fundamental reason: no naturally occurring DNA deaminase that converts adenine in DNA was known to exist. The team had to evolve one.
They started with a bacterial tRNA adenosine deaminase (TadA) that naturally converts adenosine in RNA to inosine (which is read as guanosine). Through seven rounds of directed evolution — iteratively mutating the enzyme and selecting variants with increased activity on DNA substrates — they created a TadA variant that efficiently deaminates adenine in single-stranded DNA. This engineered enzyme, combined with nCas9, became the first ABE.
The ABE Mechanism: Deamination to Inosine to Guanine
The ABE chemistry mirrors CBE but with adenine as the substrate. The evolved TadA enzyme deaminates adenine to inosine (I). Inosine is read by DNA polymerase as guanine. After the next round of DNA replication, the A•T pair has become I•T then G•C. The A has been permanently converted to G.
deaminates A
I as G
Why ABEs have lower bystander editing than CBEs: Unlike C, there is no endogenous DNA adenine deaminase activity in human cells to interfere. And because inosine repair is less active than uracil repair, ABEs do not need a glycosylase inhibitor. The result is that ABEs tend to show cleaner editing with less bystander activity than comparable CBEs.
Clinical Applications of ABEs
ABEs are arguably even more therapeutically valuable than CBEs because A→G (and T→C) transitions account for a large fraction of the ClinVar pathogenic variants database. Key applications include: correcting the sickle cell HBB E6V mutation (T→A transversion, which base editors cannot correct directly, but adjacent ABE edits can compensate), correcting TTR mutations causing ATTR amyloidosis, and correcting mutations in PCSK9 to permanently lower cholesterol. Beam Therapeutics, founded by David Liu, has multiple ABE-based drugs in clinical trials.
Section 5 — The Editing Window: Precision and Its Limits
Both CBEs and ABEs do not edit every base in the 20-nucleotide protospacer equally. The deaminase domain is physically attached to one end of the nCas9 protein and can only reach bases that are positioned within a certain distance of its active site. This creates an editing window: a defined range of positions within the spacer where the target base can be edited.
For most CBEs and ABEs, the editing window spans approximately positions 4–8 from the PAM-distal end of the spacer (counting position 1 as the most PAM-distal nucleotide). This is roughly the centre of the 20-nucleotide protospacer. Bases outside this window are largely unaffected by the deaminase, even if they are the target base type.
Practical implication: If your target C or A is at position 1, 2, 3, or 9–20, it may not be in the window. You must either choose a different guide RNA that repositions the target base into the window, or use a widened-window base editor variant.
The editing window can be widened or shifted by engineering changes to the linker between the deaminase and nCas9. Variants like BE4max, ABE8e, and ABE8.20 have optimised linkers that improve efficiency and expand the window. For cases where no suitable window position exists, prime editing provides the alternative.
Section 6 — Base Editing Off-Target Effects: A Different Risk Profile
Base editors have a fundamentally different off-target profile from standard Cas9. Because they do not create DSBs, they do not produce large deletions, chromosomal translocations, or the p53-driven selection issues described in Cluster 6. However, they introduce their own off-target concerns that are unique to base editing.
DNA Off-Target Editing
Like Cas9, base editors can bind near-match sequences off-target and convert bases there. Because nCas9 still requires the PAM sequence and 20-nucleotide guide RNA matching for site selection, DNA off-target base editing generally follows the same rules as Cas9 off-target cutting — seed region matches are most dangerous. However, base editors may produce off-target base conversions at sites where Cas9 would not produce detectable cleavage, because the deaminase requires less dwell time than the full cutting reaction.
RNA Off-Target Editing: A Unique Problem
Perhaps the most surprising off-target finding for base editors is RNA editing. The deaminase domain of CBEs (cytidine deaminases like APOBEC family members) can also edit cytosines in single-stranded RNA — including cellular mRNAs. Studies have detected thousands of A-to-I or C-to-U edits in the transcriptome of cells expressing base editors, even at concentrations used for therapeutic editing. This RNA editing is separate from DNA editing and occurs because the deaminase domain acts on any available single-stranded nucleic acid, not just the genomic DNA at the Cas9-bound site.
This was not a concern that was anticipated before base editors were widely used, and it is an active area of investigation. The biological consequences of transient, widespread RNA editing are unclear — mRNAs are short-lived and the edits likely disappear when base editor expression ceases — but it is a regulatory concern for therapeutic applications. Engineered deaminase variants with reduced RNA editing activity (SECURE-CBE, rAPOBEC1 variants) have been developed to address this.
Section 7 — Prime Editing: Search and Replace for Any Sequence
Base editing is powerful but constrained: it can only make four types of transition mutations (C→T, T→C, A→G, G→A), only at positions within the editing window, and cannot make transversions (C→A, C→G, etc.), insertions, or deletions. For the full range of pathogenic mutations — including transversions like the sickle cell A→T mutation, small deletions, and precise insertions — a different approach is needed.
Prime editing, published by the Liu lab in 2019, solves this comprehensively. It can make all 12 types of point mutations, small insertions, and small deletions — all without a DSB and without a separate donor DNA template. The information specifying the desired edit is encoded directly in a specially designed RNA molecule called the pegRNA (prime editing guide RNA). Prime editing is often described as a “search and replace” for the genome.
The Three Components of a Prime Editor
The RuvC-active, HNH-dead nickase. Finds the target sequence guided by the pegRNA and nicks the non-template strand. Does not create a DSB.
An engineered MMLV reverse transcriptase fused to nCas9. Uses the 3’ end of the pegRNA as a template to synthesise new DNA encoding the desired edit.
A specially designed guide RNA with an extension at its 3’ end. Contains: (1) the spacer for target recognition, (2) the primer binding site (PBS) that anneals to the nicked strand, (3) the RT template encoding the desired edit.
How Prime Editing Works: The 5-Step Mechanism
Section 8 — pegRNA Design: The Key to Prime Editing Success
Prime editing is only as good as its pegRNA. Designing an effective pegRNA is more complex than designing a standard guide RNA, because you must specify not just the 20-nucleotide spacer but also the primer binding site (PBS) and the RT template. Each of these elements requires careful design, and the rules are less established than for standard guide RNA design.
The Four Components of a pegRNA
PE2, PE3, PE5: The Prime Editing System Generations
Section 9 — CRISPRi and CRISPRa: Editing Gene Expression Without Touching DNA
The most conservative use of the CRISPR machinery is one that does not change the DNA sequence at all. CRISPRi (interference) and CRISPRa (activation) use dCas9 — the catalytically dead Cas9 that binds DNA without cutting — fused to transcriptional regulators that turn genes off or on at will. No DNA modification, no repair, no indels. Just programmable gene expression control.
CRISPRi: Programmable Gene Silencing
CRISPRi uses dCas9 fused to a transcriptional repressor domain, most commonly the KRAB domain (Kruppel-associated box) from mammalian zinc finger proteins. The KRAB domain recruits co-repressors (TRIM28/KAP1) that establish and spread repressive histone marks (H3K9me3) across the targeted promoter region, silencing gene expression. KRAB-dCas9 silencing is highly effective — often achieving 90%+ knockdown — and because it is epigenetic rather than genetic, it is completely reversible when the dCas9 is removed.
A powerful application is genome-wide CRISPRi screens. A library of guide RNAs targeting the promoters of every gene in the genome is delivered to cells, along with KRAB-dCas9. Each cell receives one guide and silences one gene. By measuring which cells gain or lose fitness in a given condition (cancer drug treatment, viral infection, nutrient deprivation), researchers can systematically identify which genes are essential for that phenotype. This approach has dramatically accelerated the discovery of drug targets and genetic vulnerabilities.
CRISPRa: Programmable Gene Activation
CRISPRa uses dCas9 fused to transcriptional activator domains that recruit RNA polymerase and the transcription initiation complex to the target promoter, dramatically amplifying gene expression. The most effective CRISPRa systems use multiple activation domains in combination: VP64 (four tandem copies of the VP16 activator), p65, and Rta, connected by flexible linkers (the VPR system), or the SAM (Synergistic Activation Mediator) system that recruits multiple activators via an extended guide RNA scaffold.
CRISPRa is particularly valuable for studying gene function by overexpression (complementing the knockdown approach of CRISPRi), for identifying gain-of-function phenotypes, and for potential therapeutic applications where upregulating a protective gene could compensate for disease. For example, upregulating fetal haemoglobin production via CRISPRa targeting the HBG1/2 promoters has been explored as an alternative to the BCL11A disruption approach used in Casgevy.
The Precise Calligrapher, the Word Processor, and the Dimmer Switch
Our saga of the molecular surgeon has reached its most elegant chapter. So far, the surgery has involved opening doors (Cas9 binding), cutting walls (the DSB), and calling in repair crews whose work is somewhat unpredictable (NHEJ). Now meet three new specialists who have joined the team — specialists who can do their work without breaking anything.
The first is the Calligrapher. She carries a special chemical pen and is guided to a specific location in a manuscript by the now-familiar GPS system (the guide RNA). When she arrives, she does not tear out the page. She does not even cross out the word. She simply changes a single letter — carefully applying her chemical pen to convert a C into a T, or an A into a G — right there on the page. The surrounding text is untouched. The binding of the manuscript is untouched. She has changed one character in 3.2 billion, and left. This is base editing.
The Calligrapher has two pen types. The red pen (CBE) changes C to T. The blue pen (ABE) changes A to G. Between them, they can correct four of the twelve possible single-letter mutations. But there is a constraint: each pen has a reach. She can only correct letters that fall within a certain physical distance of where her hand rests (the editing window, approximately positions 4–8). A letter two sentences away — outside the window — is beyond her reach. She cannot move her hand without repositioning the entire GPS system (choosing a different guide RNA).
The second specialist is the Word Processor. He represents prime editing. He carries something extraordinary: not just a pen, but a complete replacement text for the exact phrase that needs to change. This text is encoded in his instructions (the pegRNA RT template). He arrives at the target location, carefully marks the old text with a nick — just enough to lift the corner of the page — and then reads his instruction sheet and types in the replacement text using a molecular typewriter (the reverse transcriptase). The new text can be any sequence: a substitution, an insertion, a deletion, or a transversion that the Calligrapher’s pens cannot make. He gently reseals the page corner, and the replacement text is now permanent.
The Word Processor is more versatile than the Calligrapher — he can fix any error type — but he is also slower and sometimes less reliable. His replacement text occasionally fails to integrate properly, especially in difficult cells. And his instructions (the pegRNA) are more complex to write than the Calligrapher’s simple directional pen. In a clinical setting, you reach for the Calligrapher if her pen can fix the specific letter you need to change. You reach for the Word Processor when no pen can do the job.
The third specialist does something entirely different. She is the Dimmer Switch Operator. She carries no pen and no typewriter — she does not change a single letter in the manuscript. Instead, she goes to the switch panel on the wall next to a specific chapter (the gene promoter) and either dims the lights in that room (CRISPRi, silencing the gene) or turns them up brighter (CRISPRa, activating the gene). The manuscript is completely untouched. But the room is darker or brighter, and the effect is real and immediate. This represents dCas9 transcriptional regulation.
The Dimmer Switch Operator has a unique property: her changes are reversible. When she leaves the building, the switch returns to its default position. No permanent record, no heritable change, no repair required. For research — asking “what happens when this gene is turned off?” across an entire genome, one gene per cell — she is invaluable. For therapy in conditions where you need permanent correction, you need one of the other specialists. But for conditions where temporary modulation is all that is required, or where the therapy itself does not need to be permanent, the Dimmer Switch Operator may be the safest and most elegant choice of all.
The progression of this series — from scissors (Cas9) to precise pens (base editors) to a word processor (prime editor) to a dimmer switch (dCas9 regulators) — mirrors the actual historical progression of the field. Each tool was developed to address a limitation of the one before it. And each adds to a growing arsenal of molecular precision that, collectively, puts correction of almost any genetic disease within conceptual reach. The question is no longer whether we can edit the genome precisely. The question is whether we can get these tools to the right place, in the right cells, at the right time — which is where Clusters 8, 9, and 10 of this series take us.
References & Further Reading
- Komor et al. (2016) — Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage. Nature 533:420. — The original CBE paper. The first base editor.
- Gaudelli et al. (2017) — Programmable base editing of A•T to G•C in genomic DNA without DNA cleavage. Nature 551:464. — The original ABE paper. Required evolving an enzyme that did not exist in nature.
- Anzalone et al. (2019) — Search-and-replace genome editing without double-strand breaks or donor DNA. Nature 576:149. — The original prime editing paper.
- Nelson et al. (2022) — Engineered pegRNAs improve prime editing efficiency. Nature Biotechnology 40:402. — The epegRNA paper showing how stabilised pegRNAs improve efficiency 2–4 fold.
- Villiger et al. (2021) — In vivo cytosine base editing of hepatocytes without detectable off-target mutations in cynomolgus monkeys. Nature Medicine 27:1178. — Landmark in vivo base editing safety data in primates.
- Gilbert et al. (2013) — Reprogramming Cas9 to silence rather than cut DNA. Cell 154:442. — The original CRISPRi paper using dCas9-KRAB.
- PrimeDesign tool — primedesign.stanford.edu — Free web tool for pegRNA design. Required for any prime editing experiment.
- Base editors change single letters without breaking the DNA backbone. nCas9 opens the helix, deaminase converts the target base, and replication locks in the edit. No DSB, no large deletions, no translocations, no p53 selection.
- CBEs convert C→T; ABEs convert A→G. Together they cover four of the twelve possible point mutation types (the transitions). CBEs use APOBEC deaminases; ABEs use an evolved TadA enzyme that does not exist in nature.
- The editing window defines where editing can occur. Positions 4–8 from the PAM-distal end are the typical activity window. Target bases outside the window are not accessible to the deaminase. Guide RNA selection must position the target base within the window.
- CBEs have RNA off-target activity. The deaminase can edit cytosines in cellular mRNAs, not just the genomic DNA target. This is transient but must be characterised. SECURE-CBE variants reduce RNA editing.
- Prime editing handles all 12 mutation types plus small indels. The pegRNA encodes both the targeting information and the desired edit sequence. The RT domain synthesises the edit as new DNA. No DSB, no donor template.
- PE5 + epegRNA is the current gold standard. Second nick (PE3 strategy) + MLH1 suppression (PE5) + engineered pegRNA scaffold (epegRNA) together achieve 20–70% efficiency in dividing cells.
- CRISPRi and CRISPRa control gene expression without any DNA change. KRAB-dCas9 silences genes epigenetically (reversible). VP64/VPR/SAM activates genes by recruiting transcription machinery. Both are powerful research tools; CRISPRa has therapeutic potential for compensatory upregulation strategies.