# Epigenome editing

dCas9 fused to a methyltransferase, demethylase or histone-modifying domain, and the heritability problem: which marks survive after the effector is gone, and which relax as soon as it is.

Changing whether a gene is read without changing what it says — and the one question that decides whether the change lasts.

Source: https://en.bioecon.ru/docs/synbio-enabling/genome-engineering/epigenome-editing/
Updated: 2026-09-06



Epigenome editing keeps the DNA sequence intact and changes its regulatory state. The construct is the familiar one with its cutting function removed: a catalytically dead Cas9 retains guide-directed binding but cleaves nothing, and serves purely as a programmable anchor. Fused to it is an effector domain that writes or erases a chemical mark on DNA or on the histones packaging it.

## What the effectors actually do

The best-characterised DNA mark is 5-methylcytosine at CpG dinucleotides. A DNA methyltransferase domain — typically the catalytic region of DNMT3A, often paired with its partner DNMT3L — deposits it across the targeted promoter; a TET dioxygenase domain drives the reverse reaction by oxidising the methyl group, initiating its removal. Histone effectors change the chromatin state instead: the KRAB domain recruits the corepressor complex that deposits repressive H3K9 trimethylation, while acetyltransferase domains such as p300 add the acetylation associated with active enhancers. Simple steric blockade is also available and is the weakest form: dCas9 sitting on a promoter obstructs the transcription machinery and does nothing else.

The advantage over sequence editing is specific and worth stating. There is no break, no repair pathway and no permanent alteration, so an unintended binding event does not leave a scar in the genome. It also works on regulatory elements — enhancers, promoters — where the goal is dosage rather than a coding change, and it can act on many genes in a family through one shared motif.

## Heritability is the honest question

This is where the field must be read carefully. An effector that is present is not evidence of a durable result; the mark it writes must persist through cell divisions after the effector is withdrawn, or the gene simply returns to its prior state. That distinction separates two very different products: a transient modulation lasting days, and a single administration establishing a stable silenced state.

Mechanistically, durability rests on self-reinforcing chemistry. Maintenance methyltransferase activity copies CpG methylation onto the daughter strand at replication, and repressive histone marks recruit the enzymes that write more of the same, so a sufficiently established state can propagate itself. Experimentally, combining DNA methylation with histone repression has produced silencing that persists through many divisions after the editor is gone, while a single effector often relaxes. Results are strongly locus-dependent: CpG island density, the chromatin context and the cell type all change the answer, and a durable result at one promoter does not predict another.

Two limits follow. Some targets are not stably silenceable at all. And durability is exactly what one would want to reverse if the effect proves harmful, which is why demonstrated reversal by a targeted demethylase is part of an honest case rather than an optional extra.

Off-target binding, guide design and delivery follow the general treatment on the [genome editing](../genome-editing/) page; what is specific here is that the wrong outcome is a misregulated gene, not a mutation.