Over the past decade, the main thread of gene therapy has been "rewriting hardware"—from gene supplementation to CRISPR gene editing. Their shared logic is direct modification of the DNA itself: either supplying a functional gene or cutting and rewriting genomic sequences. However, this path carries inherent risks, including genotoxicity, off-target effects, and irreversibility.
In recent years, a new class of technologies has been rapidly advancing from concept to clinic—epigenome editing. Its hallmark is that it does not alter the DNA sequence but instead rewrites the "software" of gene expression. In 2022, the first epigenome editing therapy entered the clinic; in 2024, Nature reported a proof-of-concept study showing "a single injection lowers cholesterol with efficacy sustained for nearly one year." This marked the crossing of the threshold from laboratory to human application for this paradigm.
01
How Epigenetic Marks Enable "Heritable Silencing"
1.1 Programmable DNA-Binding Domain + Epigenetic Effector
Epigenome editors consist of two modular components: a targeting module that directs the effector to the target locus, and an effector module that writes the epigenetic mark. Early targeting modules relied on zinc finger proteins (ZF) and TALEs; the introduction of CRISPR-dCas9 (catalytically dead Cas9) in 2013 greatly enhanced programmability and has become the mainstream platform. Effector modules include the catalytic domain of DNA methyltransferase (DNMT3A/DNMT3L), the KRAB domain (which recruits KAP1/SETDB1 to establish H3K9me3 heterochromatin), and demethylases such as TET.
Based on functional direction, epigenome editing can be divided into three categories: epigenetic silencing (writing repressive marks to turn off disease-causing genes, the most clinically advanced), epigenetic activation (CRISPRa, upregulating underexpressed genes), and epigenetic erasure/reversal (active demethylation to reverse aberrant silencing). This article focuses on epigenetic silencing, which has the fastest clinical progress.
1.2 Single Mark vs. Tripartite Combination: The Source of Durability
The self-maintenance of DNA methylation is the molecular basis for durable silencing. The 5mC marks written by DNMT3A/3L can be faithfully copied to the daughter strand by the maintenance methyltransferase DNMT1 during each round of DNA replication, enabling stable transmission across cell divisions. When the tripartite combination of DNMT3A, DNMT3L, and KRAB (ETR) is delivered together, it can achieve near-permanent silencing across a variety of genes. This is the key design laid down by the San Raffaele team in Italy, which they have advanced toward clinical translation.
Core Mechanism · Hit-and-Run
The epigenome editor is transiently expressed (e.g., via LNP-delivered mRNA), writes self-maintaining methylation marks, and is then cleared, while the silenced state is long-term maintained by the cell's endogenous machinery—a single administration achieves durable silencing with no residual exogenous editor in the body and without altering a single base. This breaks the false dichotomy that "durability requires permanent DNA rewriting" or "without rewriting, lifetime repeated dosing is necessary."
1.3 Reversibility: A Double-Edged Sword
Unlike the irreversibility of DNA editing, epigenetic marks can in principle be actively reversed, providing a safety margin of "pause and withdrawal." However, reversibility is also a double-edged sword: if the silenced state is spontaneously reversed by the cell or if epigenetic "escape" occurs, therapeutic durability will be compromised. The true duration and stability of durability remain core questions to be validated before epigenome editing can become a medicine.
Hit-and-Run Mechanism Flowchart
1.4 Unique Differentiation Positioning
The distinctive value of epigenome editing lies in fulfilling three criteria simultaneously: durable (comparable to DNA editing) + reversible (comparable to siRNA) + no DNA cutting (avoiding genotoxicity). It occupies the white space between siRNA/ASO (reversible but requiring chronic repeated dosing) and CRISPR editing (durable but permanent and cleaving), making it particularly well-suited for major diseases such as hepatitis B and hypercholesterolemia.
02
Delivery and Manufacturing
2.1 Delivery Strategy Tightly Coupled with Route
The liver (readily accessible via LNP) is the most mature battleground for epigenetic silencing. Hepatitis B and hypercholesterolemia naturally fit the profile of "liver-targeted organ + need for durable silencing." Extrabepatic delivery (CNS, muscle, eye) represents the next bottleneck shared by epigenome editing and the entire in vivo gene therapy field.
2.2 CMC: From "Being Able to Edit" to "Being Able to Release and Manufacture"
The three delivery routes correspond to three distinct manufacturing logics: LNP–mRNA type (Omega, Tune, Chroma, Weiguang, etc., for liver routes) can directly leverage the mature processes already de-risked by mRNA vaccines and siRNA-LNP, representing the lowest CMC risk and fastest scale-up path; AAV type (Epic Bio, Modalis, Sangamo, etc., for extrahepatic routes) is constrained by both payload size and production capacity; VLP/RNP type has lower integration risk but still immature manufacturing processes, making it the least mature among the three routes.
Epigenome editing also has its unique CMC challenges: the editor molecule is relatively large (the ETR triple fusion is one of the "heaviest" payloads in gene therapy), multi-component stoichiometric control (the editor-to-gRNA ratio directly affects efficiency and off-target effects), and reconstruction of potency assays (potency is defined as "target-site methylation rate/silencing rate" rather than concentration). The LNP–liver route not only leads in clinical progress but also has the greatest advantage in CMC and scale-up.
03
Preclinical and Clinical Evidence
Technology Milestone Timeline
3.1 Proof of Concept: One Injection Lowers Cholesterol for Nearly One Year
In February 2024, the San Raffaele Research Institute team in Italy published a study in Nature: delivering an epigenetic silencer targeting PCSK9 via a hit-and-run approach achieved durable reduction in circulating cholesterol levels in mice for nearly one year, without altering the genomic sequence—this was a landmark in vivo validation of the pharmacological characteristics of epigenetic silencing.
3.2 First Global Entry into Clinic: Omega OTX-2002
In 2022, Omega Therapeutics' OTX-2002 (LNP-delivered, targeting downregulation of MYC expression in hepatocellular carcinoma) became the world's first epigenome editing therapy to enter the clinic (MYCHELANGELO-I, currently Phase I/II). As the earliest "first mover," Omega also experienced early-stage financing and commercial pressures typical of emerging modalities—a risk footnote worth remembering when assessing the field.
3.3 Flagship Indication for Hepatitis B: Tune and Chroma/nChroma
Chronic hepatitis B is the flagship indication for epigenetic silencing. Tune Therapeutics' TUNE-401 (LNP-delivered, silencing both HBV cccDNA and integrated DNA) has entered Phase I clinical trials (RHYTHM study), co-founded by Charles Gersbach and Fyodor Urnov; Chroma Medicine / nChroma Bio's CRMA-1001 has advanced to Phase I/II, with pipelines also covering PCSK9, ANGPTL3, CFB, and others.
3.4 Neuromuscular: Epicrispr EPI-321
Facioscapulohumeral muscular dystrophy (FSHD, driven by aberrant DUX4 expression) is considered one of the ideal lead indications for epigenome editing. Epicrispr Biotechnologies' EPI-321 (silencing DUX4), developed on the GEMS platform, has entered human clinical trials.
04
Global and Chinese R&D Landscape
4.1 Overall Landscape: 80+ Assets, Clinical Pipeline Beginning to Take Shape
According to PharmCube data, as of August 2026, more than 80 R&D assets globally are annotated as "epigenetic editing therapies," covering over 15 companies and academic institutions, with approximately 7 assets having entered the clinic. The field is transitioning from "platform narrative" to "clinical readouts." Based on functional direction, it can be divided into epigenetic silencing, epigenetic activation, and the adjacent branches of epigenetic reprogramming/anti-aging. The table below summarizes representative assets across these three categories.
4.2 Competitive Landscape: Crowded Targets, Delivery Determines the Winner
PCSK9 (hypercholesterolemia) is the most crowded red ocean: Omega, Chroma, Tune, Scribe, Epic Bio, Yijielike, Weiguang Gene, Southern University of Science and Technology, and others—seven or eight players are clustered around the same target, with the marketed PCSK9 siRNA (inclisiran) already ahead. Hepatitis B (HBV/cccDNA) represents a tougher battlefield, with clinical value and the Chinese patient base far greater than lipid-lowering; neuromuscular (FSHD, DMD) is the exclusive home turf of activation-type editors; neurodegenerative directions hold long-term promise but are constrained by CNS delivery. As editors become increasingly homogenized, the real differentiators are delivery capability and who will first produce human proof-of-durability data.
Target Competition Matrix
4.3 Financing and BD
The capital landscape shows a pattern of "top-tier USD funds + pharma companies both entering the arena." On the flip side, it is equally clear that Omega Therapeutics—the first to enter the clinic globally—has faced dual pressures in financing and operations since its 2022 enrollment. "Entering the clinic ≠ deliverable value," and the cash flow and data risks of early-stage modalities coexist.
4.4 Selected Chinese Companies
Yijielike (Shanghai, founded in 2021) is among the first tier in China and globally for clinical epigenome editing: EPI-001 (PCSK9) and EPI-003 (hepatitis B) have both entered Phase I clinical trials, with cumulative financing of approximately $112 million, backed by Qiming Venture Partners, OrbiMed, Longpan Investment, and others. Weiguang Gene (Suzhou) is advancing its proprietary VLP/LNP dual-delivery platform for common disease pipelines including hepatitis B and hypercholesterolemia (preclinical). Additionally, original work from Sun Yat-sen University (HDAC9, rheumatoid arthritis) and Southern University of Science and Technology (PCSK9) has appeared in patents and publications.
China's clinical development is now broadly synchronized with the global first tier. Major diseases such as hepatitis B and hypercholesterolemia align well with China's patient base and reimbursement logic. Against the backdrop of homogenized editors, delivery—particularly proprietary VLP/LNP platforms—represents the most realistic differentiation point for Chinese teams.
05
Summary and Outlook
Epigenome editing is a natural extension of "from sequence rewriting to software rewriting," and represents a new paradigm for chronic disease gene therapy, following gene supplementation and CRISPR editing. It has crossed the threshold of "entering humans," but there is still a considerable distance to "proving itself."
5.1 Key Outstanding Questions
① Human durability remains uncharted — animal models show silencing can last nearly one year, but to date, no asset has produced long-term durability readouts in humans.
② Methylation off-target is "heritable" — methylation erroneously written at off-target sites can be continuously propagated by DNMT1 through cell divisions; off-target effects are not a one-time event but a long-term risk that can be amplified.
③ Extrahepatic delivery paradox — AAV drives persistent expression of the editor, directly counteracting the core advantage of "hit-and-run"; VLP is a potential solution but its engineering is still early-stage.
④ Crowded targets need differentiation — PCSK9 already has seven or eight players clustered, with an approved siRNA ahead; clinical data are needed to demonstrate the unique value of "single administration."
⑤ No precedents for regulation and endpoints — how durable is durable enough to qualify as "functional cure," and how to design payment models—these need to be figured out from scratch with regulators.
5.2 From "Silencing" to the Full Epigenetic Toolbox
Currently, nearly all clinical assets are doing "gene turning off" (silencing), but the programmable potential of epigenetics extends far beyond: from "silencing" to "activation" (CRISPRa targeting haploinsufficiency diseases), from "writing" to "erasing/reversing" (reversing pathological aberrant silencing, pointing to imprinting disorders and epigenetic cancer therapy), from "single-gene" to "multiplex and reprogramming" (Moonwalk enables up to 7-plex editing simultaneously; NewLimit targets anti-aging)—these three directions are opening the next chapter of epigenome editing.
Epigenome editing stands at
"technology validated, clinical data pending"
— the eve of payoff:
in the short term, can it break through the liver boundary;
in the long term, how broad a disease spectrum can this toolbox cover?
The answer will be delivered by clinical readouts over the next few years.
References and Data Sources
PharmCube Innovative Drug Pipeline Database: Epigenetic Editing Therapies—Global and China Pipelines, accessed 2026-08-03.
Cappelluti MA, Lombardo A, Naldini L, et al. Durable and efficient gene silencing in vivo by hit-and-run epigenome editing. Nature, 2024.
Heller EA, et al. Epigenome editing: from bench to bedside. Nature Reviews Drug Discovery (the "14 companies, three clinical programs" figure cited in this review is an earlier snapshot and differs from the PharmCube data as of August 2026; provided for reference only).
4–6. PharmCube pipeline database and financing/BD intelligence: Omega, Tune, Chroma/nChroma, Epic Bio, Modalis, Yijielike, Weiguang Gene, etc.
Nuñez JK, Weissman JS, et al. Genome-wide programmable transcriptional memory by CRISPR-based epigenome editing. Cell, 2021.
【Disclaimer】
This content represents the views of our company based solely on currently available public information and data provided by portfolio companies. It does not constitute investment advice, medical advice, or a guarantee of any particular company's performance. Investment involves risks; readers should make independent judgments based on public information. Unauthorized reproduction or commercial use is prohibited.