Gene editing has revolutionized biomedical research. But the gap between a powerful editing tool and a therapeutic is not the editor itself—it is the delivery vehicle. For researchers developing the next generation of genetic medicines, the question is no longer what to edit, but how to get the editing machinery to the right cell type, in the right tissue, at the right time, without triggering toxicity or immunogenicity.
The past 18 months have witnessed dramatic advances in delivery technologies. From engineered lipid nanoparticles (LNPs) that extravasate beyond the liver, to cell-type-specific adeno-associated virus (AAV) capsids, to virus-like particles (VLPs) that combine the best of viral and non-viral systems—the delivery toolbox is expanding rapidly.
This blog provides a research-focused overview of the current state of in vivo gene editing delivery, highlighting key challenges, recent breakthroughs, and what the data tell us about the path forward.
A sobering reality from the last decade of gene editing research: The most elegant editor is useless if it cannot reach its target.
In vivo delivery requires solving a multi-layered problem:
| Challenge | Description |
|---|---|
| Cargo size | Base editors (~4–5 kb) and prime editors (~6–7 kb plus pegRNA) exceed AAV packaging limits (~4.7 kb) |
| Tropism | Most AAV serotypes and LNPs naturally home to the liver; reaching brain, muscle, or lung requires engineering |
| Immunogenicity | Pre-existing immunity to Cas9 (from S. pyogenes or S. aureus) and AAV capsids limits patient eligibility |
| Durability vs. reversibility | Permanent editing is desirable for genetic diseases; transient expression is safer for acquired conditions |
| Off-target editing | Prolonged editor expression increases risk of off-target mutations in non-dividing cells |
Researchers are now addressing these challenges with creative engineering—not just of the editors themselves, but of the particles that carry them.
LNPs rose to prominence as the delivery platform for mRNA vaccines (COVID-19). The same technology can encapsulate mRNA encoding Cas9 or base editors, along with synthetic guide RNAs. Once inside the cell, the mRNA is translated transiently, producing editor protein for 48–72 hours before being degraded.
Advantages of LNPs:
No size limitation (can package large prime editor constructs)
Transient expression (reduces off-target risk)
Non-immunogenic to pre-existing anti-AAV antibodies
Scalable and manufacturable
The Liver Problem: Standard LNPs accumulate in the liver due to apolipoprotein E (ApoE) binding. This is excellent for treating genetic liver diseases (e.g., familial hypercholesterolemia, transthyretin amyloidosis). However, for diseases of the lung, muscle, or central nervous system, liver tropism is a liability.
Recent Breakthroughs:
Selective Organ Targeting (SORT): By adding a fifth "SORT" lipid (charged or zwitterionic) to the standard four-component LNP formulation, researchers have achieved selective delivery to the lung (positive charge), spleen (negative charge), or liver (neutral). This simple modular approach, published in Nature Nanotechnology, has been validated across multiple mouse strains.
Muscle and Heart Targeting: New LNP formulations incorporating cationic lipids and optimized PEG-lipid ratios have demonstrated efficient delivery of base editors to cardiac and skeletal muscle following systemic injection, enabling correction of Duchenne muscular dystrophy mutations in mouse models.
AAVs remain the most clinically validated delivery vehicle for in vivo gene therapy, with approvals for spinal muscular atrophy (Zolgensma) and retinal dystrophy (Luxturna). For gene editing, AAVs are typically used to deliver the editor, while the guide RNA is either packaged in a separate AAV or delivered as a synthetic oligonucleotide.
The Cargo Size Constraint: Standard AAV capsids package ~4.7 kb of DNA. This comfortably fits SpCas9 (~4.2 kb) but does not fit:
SpCas9 + a long promoter
Most base editors (Cas9 nickase + deaminase = ~5–5.5 kb)
Prime editors (Cas9 nickase + reverse transcriptase = ~6.5 kb)
Solutions:
Dual AAV strategies: Split the editor across two AAVs, relying on homologous recombination or intein-mediated protein splicing to reconstitute the full editor in trans. Several groups have reported successful dual-AAV delivery of prime editors in mouse liver and brain, though efficiency remains lower than single-AAV systems.
Compact Cas homologs: Cas12f (400–500 amino acids) and CasMINI (529 amino acids) are small enough to package with a promoter into a single AAV. Engineered variants have achieved base editing efficiencies approaching those of SpCas9, opening new possibilities for one-vector in vivo editing.
Virus-like particles (VLPs) represent an emerging hybrid platform. VLPs are assembled from retroviral or lentiviral structural proteins (Gag, GagPol) but lack the viral genome. Instead, they are loaded with Cas9 protein (not mRNA) and synthetic guide RNAs as ribonucleoprotein (RNP) complexes.
Why VLPs Are Exciting:
Protein delivery: RNP is active immediately upon entry, with no time for off-target accumulation during transcription/translation.
No DNA payload: No risk of genomic integration (unlike lentiviral vectors).
Size agnostic: RNP loading is not limited by packaging constraints.
Tunable tropism: Pseudotyping with VSV-G, baboon retroviral envelope, or engineered glycoproteins directs VLPs to specific cell types.
Recent studies have demonstrated VLP-mediated base editing in the mouse retina, liver, and brain with efficiencies comparable to AAV but with substantially lower off-target editing. The primary limitation remains manufacturing complexity and lower titers compared to AAV or LNPs.
Systemic delivery inevitably reaches many cell types. The ideal delivery vehicle would edit only the disease-relevant cell type, sparing healthy tissues. Two orthogonal approaches are converging:
A. Capsid Engineering (for AAV):
Directed evolution of AAV capsids using Cre-recombination-based selection has yielded variants with remarkable specificity. For example, AAV9-PHP.eB crosses the blood-brain barrier and transduces neurons efficiently after systemic injection, while AAV2.7m8 is highly specific for retinal pigment epithelium.
B. LNP Surface Functionalization (for LNPs):
Conjugating antibodies, antibody fragments (Fab), or affimers to the LNP surface enables cell-type-specific uptake. A recent landmark study used CD4-targeted LNPs to deliver base editors to CD4+ T cells in vivo, achieving >80% editing of the CCR5 gene (a co-receptor for HIV) after a single intravenous dose in humanized mice.
For researchers designing in vivo editing experiments, not all endpoints are equal. The field is moving beyond simple quantification of editing efficiency in bulk tissue to more rigorous metrics:
| Metric | Why It Matters |
|---|---|
| Editing efficiency in target cell type | Bulk tissue measurement dilutes signal from non-target cells; single-cell resolution (flow cytometry, scRNA-seq) is essential |
| On-target vs. off-target ratio | High on-target with minimal off-target (especially in germline or dividing tissues) |
| Duration of editor expression | mRNA/LNP: 48–72 hours (low off-target risk); AAV: weeks to months (higher risk, but potentially necessary for dividing cells) |
| Immune response | Neutralizing antibodies against Cas9 or capsid limit redosing potential; T-cell responses can eliminate edited cells |
| Phenotypic correction | Editing efficiency is a surrogate; actual functional recovery (e.g., dystrophin restoration, clotting factor activity) is the true endpoint |
Several preclinical advances are poised to enter IND-enabling studies in the next 12–24 months:
Epigenetic editing in vivo: Using dCas9-p300 or dCas9-DNMT3A delivered via AAV or LNP to persistently silence or activate genes without cutting DNA. Early studies in mouse models of PCSK9 (cholesterol), SMN2 (spinal muscular atrophy), and FXN (Friedreich's ataxia) show durable effects after single doses.
Tissue-specific prime editing: Dual-AAV prime editors with optimized split-intein systems have achieved correction of the Pah mutation (phenylketonuria) in mouse liver at 15–20% efficiency, sufficient for biochemical normalization.
In vivo delivery to hematopoietic stem cells (HSCs): The bone marrow microenvironment is notoriously difficult to target. New polymer nanoparticles and HSC-tropic AAV variants (AAV6, AAVHSC16) are showing promise for in situ HSC editing, potentially replacing ex vivo editing for sickle cell disease and beta-thalassemia.
Delivery is no longer an afterthought in gene editing research. It is the central challenge—and the site of the most exciting innovation. LNPs are moving beyond the liver, AAV capsids are evolving toward greater specificity, and VLPs offer a new protein-based delivery paradigm. For researchers at A Helix Bio, understanding the strengths and limitations of each platform is essential for designing translatable in vivo studies.
