In a review published in BMC Methods, researchers Sumiyo Morita, Takuro Horii, and Izuho Hatada examine recent technological advancements in epigenome editing and evaluate their in vivo applications for disease modeling and therapeutic intervention. Unlike conventional genome editing, which introduces double-strand breaks to permanently alter the genomic DNA sequence, epigenome editing alters chromatin marks—specifically DNA methylation and histone modifications—to modulate transcription without damaging or mutating host DNA. Epigenetic mechanisms regulate gene expression across cell division during development and differentiation, and aberrations in these pathways are linked to cancer and lifestyle-related disorders. While pharmaceutical therapies targeting epigenetic machinery, such as DNA methyltransferase inhibitors like azacitidine and decitabine or histone deacetylase inhibitors, are clinically approved, their systemic activity indiscriminately alters global methylation and acetylation. Consequently, these broad-spectrum chemical agents carry considerable risk of side effects, highlighting the necessity for modular platforms that direct site-specific chromatin modifications to reversibly correct localized epigenetic anomalies.
The architecture of epigenome editing relies on linking a programmable DNA sequence-binding module to an effector module containing an enzyme that modifies chromatin. Investigators utilize major platforms developed for genome editing, including Zinc-Finger Nucleases (ZFNs), Transcriptional Activator-Like Effector Nucleases (TALENs), and CRISPR/Cas9. ZFNs assemble multiple zinc finger domains that each recognize three nucleotide bases, fused to the FokI cleavage domain; however, engineering linked zinc fingers presents challenges because inter-domain interactions can disrupt binding specificity and affinity. In contrast, TALENs employ DNA-binding domains composed of around fifteen TALE repeats coupled to FokI, which are easier to handle because adjacent repeats do not interfere with reciprocal base recognition. CRISPR/Cas9 pairs Cas9 with a single guide RNA that defines genomic targets through base pairing, providing unmatched flexibility. To adapt CRISPR/Cas9 for epigenome editing, researchers engineered catalytically inactive dead Cas9 (dCas9) harboring D10A and H840A mutations, which abolish cleavage activity while preserving sequence-specific binding.
Effector modules supply the enzymatic machinery required to install or remove chemical marks at targeted loci, producing distinct transcriptional outcomes. DNA methyltransferases such as DNMT3 catalyze methyl group transfer to the C5 position of cytosine, driving transcriptional repression, whereas Ten-eleven translocation enzymes including TET1 and TET3 remove methyl groups to promote activation. Beyond DNA modifiers, diverse histone-modifying enzymes targeting acetylation, methylation, phosphorylation, and ubiquitination serve as effectors across nine recognized modification classes. For instance, methylation of histone H3 at lysine 4 and acetylation of H3K27 promote activation, whereas methylation at H3K9 or H3K27 and H2AK119 monoubiquitination cause silencing. Histone citrullination mediated by peptidylarginine deiminase (PPAD) fused to dCas9 directs locus-specific citrullination to activate or repress transcription depending on genomic context. Furthermore, transcriptional regulators such as VP64 recruit co-activators to stimulate expression, while the Krüppel-associated box recruits co-repressors to form heterochromatin, supplemented by hypercompact activators of 64 to 98 amino acids identified from viral interferon regulatory factor 2 core domains.
Early epigenome editors fusing dCas9 directly to effector domains, such as the catalytic domain of TET1 (TET1CD), often exhibited weak demethylation even when using multiple guide RNAs simultaneously. To enhance efficacy, researchers engineered multi-effector recruitment systems. The VPR architecture fuses VP64, p65, and Rta to dCas9, achieving robust and reproducible transcriptional activation across diverse target genes and cell types. The Synergistic Activation Mediator (SAM) system utilizes dCas9-VP64 alongside an sgRNA bearing two hairpin sequences that recruit MS2 coat protein fused to p65-HSF1. The SunTag platform uses dCas9 fused to a repeating GCN4 peptide array that recruits single-chain antibodies carrying effectors such as VP64 or TET1CD, concentrating catalytic activity at target loci. Moreover, the CRISPRoff system couples dCas9 to Dnmt3a, Dnmt3L, and a KRAB domain, establishing DNA methylation and repressive histone marks that elicit heritable gene silencing lasting up to approximately one year in vivo when delivered via lipid nanoparticles.
Beyond multi-effector complexes, recent strategies integrate systematic mapping of histone marks and biophysical phase separation principles. Using a comprehensive epigenome editing toolkit, Policarpi and colleagues installed nine major histone modifications at single-cell resolution, revealing that promoter-targeted H3K4me3 remodels chromatin to trigger transcription, whereas co-targeting H3K27me3 and H2AK119ub maximizes gene silencing penetrance. Concurrently, researchers have leveraged liquid-liquid phase separation (LLPS), where biomolecules assemble into membrane-less organelles such as stress granules and nucleoli to form transcriptional condensates. The DropCRISPRa platform merges the SunTag system with intrinsically disordered regions from FET proteins, such as Fus or TAF15, and activation domains including VP64, p65, or VPR. This arrangement forms phase-separated condensates at target sites that concentrate endogenous transcriptional regulators, significantly enhancing transcription, though excessive phase separation can conversely impede expression. Temporal regulation has also advanced through systems inducible by abscisic acid, small molecules binding epigenetic modifiers, and phototropic stimulation, offering reversible control over targeted modifications.
Epigenome editing has demonstrated significant utility in generating precise animal models to study human diseases. For instance, Horii and colleagues constructed a mouse model of Silver-Russell syndrome, an imprinting disorder caused by loss of DNA methylation at the paternally derived H19-differentially methylated region (H19-DMR). Using dCas9 combined with a modified SunTag platform, the team induced targeted demethylation of this locus via three experimental approaches: tetraploid complementation introducing transfected embryonic stem cells into tetraploid blastocysts; microinjecting editing mRNA into preimplantation embryos; and stable transgenic expression in mice. While the first two approaches generated edited animals, induced epigenetic alterations were erased during germline passage and were not inherited by offspring. Stable transgenic expression circumvented germline loss but increased potential off-target risks. The resulting model mice successfully reproduced clinical symptoms of Silver-Russell syndrome, including fetal and postnatal growth retardation, body asymmetry, disproportionately enlarged heads, abnormal facial features, anorexia, and hypoglycemia, illustrating the capacity of epigenome engineering to model complex epigenetic pathologies.
In therapeutic settings, in vivo epigenome editing provides distinct mechanistic advantages over RNA-targeting platforms such as siRNAs, microRNAs, and antisense oligonucleotides, which primarily suppress gene expression. Epigenome editing permits bidirectional control, enabling both upregulation and targeted repression across disorders driven by haploinsufficiency, imprinting defects, X-linked inheritance, cancer, or aging. To achieve persistent silencing without sustained editor activity, Neumann and colleagues developed the CHARM system, which established durable methylation and transcriptional silencing of the prion protein across mouse neurons. CHARM incorporates a DNA-binding domain, a Dnmt3l recruitment domain, and a histone H3 tail that activates endogenous DNMT3a at CpG islands. Importantly, CHARM features an engineered self-silencing function that terminates expression once the target locus is repressed, limiting risks associated with prolonged editor presence. The system functions across ZFP, TALE, and CRISPR platforms; when paired with compact zinc fingers, up to three targeting units and cell-specific control elements fit within a single delivery vector, an approach accelerated by deep learning models automating ZFP design.
A critical frontier in epigenetic therapeutics is allele-specific editing, which discriminates between maternal and paternal gene copies in diploid cells. Allele-specific targeting exploits natural sequence variations such as single-nucleotide polymorphisms (SNPs) within target sites. Rajaram and colleagues demonstrated that SNPs located within the seed region of the guide RNA or the PAM motif allow dCas9 to direct DNA methylation exclusively to a chosen allele. Researchers have applied this approach using ZFP or TALE fusions linked to KRAB repressor domains to selectively silence mutant huntingtin in Huntington’s disease models while preserving normal expression of the healthy allele. Similar allele-specific interventions show promise for conditions like Machado-Joseph disease and frontotemporal dementia, and provide a direct method for correcting imprinting disorders by manipulating DNA methylation exclusively on the aberrantly marked parental copy.
Clinical implementation of in vivo epigenome editing requires efficient delivery systems, with adeno-associated viral (AAV) vectors serving as a primary delivery platform. However, AAV vectors are subject to a strict packaging capacity limit of 4.7 kilobases. Full-length Streptococcus pyogenes dCas9 spans approximately 1,368 amino acids, leaving insufficient space to package dCas9 coupled with effector modules and regulatory cassettes; exceeding this threshold reduces viral titers and generates heterogeneous truncated genomes. To resolve packaging constraints, researchers have adopted smaller Cas orthologs, such as Staphylococcus aureus dSaCas9 (1,052 amino acids). Thakore and colleagues demonstrated that systemic delivery of a dual-vector AAV system encoding dSaCas9-KRAB and a Pcsk9-targeting guide RNA significantly reduced serum Pcsk9 and cholesterol levels in mice. Evolutionary screening has identified even smaller variants, including CjCas9 (984 amino acids), NmeCas9 (1,082 amino acids), Cas12j (750 amino acids), and dCasMINI (529 amino acids), while ancestral nucleases such as IscB and TnpB (~400 amino acids) offer promising compact alternatives.
Lipid nanoparticles (LNPs) provide an alternative non-viral vehicle offering large cargo capacity and transient delivery, mitigating immunogenicity and off-target effects associated with sustained viral expression. While intravenous LNPs naturally accumulate in the liver, tuning lipid chemistry allows delivery to organs including the brain, lungs, heart, and muscle. Illustrating non-viral therapeutic efficacy, Capelluti and colleagues designed the EvoETR repressor system targeting Pcsk9 in hepatocytes. Testing constructs combining ZFPs, TALEs, or dCas9 with KRAB, cdDNMT3A, and DNMT3L domains demonstrated that a ZFP platform flanked by DNMT3A-DNMT3L and KRAB domains yielded superior silencing, producing a nearly 50 percent reduction in circulating PCSK9 for up to one year following a single LNP administration, with epigenetic repression persisting through liver regeneration. Achieving stable regulation also requires understanding host degradation mechanisms. Mammalian Keap1 acts as an endogenous E3 ubiquitin ligase that degrades Cas9 and dCas9 through an ETGE-like degron; mutating this degron shields dCas9 from degradation, prolonging protein half-life and substantially enhancing editing efficacy in mammalian cells.
Subject of Research: Technological advancements and in vivo therapeutic applications of epigenome editing
Article Title: Epigenome editing strategies: technological progress and in vivo applications
Article References: Morita, S., Horii, T., & Hatada, I. (2026). Epigenome editing strategies: technological progress and in vivo applications. BMC Methods, 3(1), Article 19. https://doi.org/10.1186/s44330-026-00069-4
Image Credits: AI Generated
DOI: 10.1186/s44330-026-00069-4
Keywords: epigenome editing, CRISPR-dCas9, DNA methylation, histone modifications, in vivo therapy, viral vectors, lipid nanoparticles, gene regulation, Epigenome, editing, strategies, technological
