Endosomal Escape: Emerging Strategies for Nucleic Acid Therapeutics

A forward-looking perspective on confining escape enhancement to the target cell and the intracellular trigger

For every oligonucleotide modality now in the clinic, the rate-limiting step is not cell entry. Uptake is comparatively easy: phosphorothioate ASOs bind cell-surface and serum proteins and are taken up gymnotically, GalNAc conjugates are cleared into hepatocytes within minutes by the asialoglycoprotein receptor, and lipid nanoparticles (LNPs) are internalized by clathrin-mediated endocytosis and macropinocytosis. The problem is that endocytosis delivers the cargo into a membrane-bounded vesicle, and the pharmacologically relevant targets – RNase H1, the RISC complex, the spliceosome – are in the cytosol and nucleus, on the other side of that membrane (Fig. 1). The great majority of what a cell internalizes is recycled, exocytosed, or trafficked to the lysosome and degraded. Only a small, poorly controlled fraction ever reaches the compartment where it can act.

The most-cited number for how small that fraction is comes from Gilleron and colleagues¹, who conjugated colloidal gold to siRNA so individual molecules could be counted by electron microscopy. They estimated that escape of LNP-delivered siRNA into the cytosol occurs at 1-2% efficiency, and only during a narrow time window while the LNP resides in a compartment with mixed early- and late-endosomal character. Two companion observations sharpen the picture. Sahay et al. found that roughly 70% of internalized siRNA is exocytosed back out of the cell through the endocytic recycling pathway, so escape competes against active egress, not merely against degradation.² Wittrup imaged the release in live cells and saw cytosolic entry occur in rare bursts from maturing endosomes about 5-15 minutes after endocytosis, each event immediately flagged by cytosolic galectin-8 recognizing the damaged vesicle.³

Endosomal escape is the rate-limiting step shared by every endocytic route of nucleic acid therapeutic (NAT) delivery. Historically, endosomal escape has been viewed primarily as a problem of membrane disruption. Increasingly, however, delivery systems aim to combine cell-selective uptake, favorable intracellular routing, and locally activated escape mechanisms into a concrete design principle: enhancement must be made cell-restricted (active only in the target cell) or stimulus-gated (active only under an intracellular trigger), and not systemic. Several emerging strategies embody one or both constraints.

Lipid nanoparticles: pH-gated escape, with a tropism caveat

Even optimized lipid nanoparticles (LNPs) release only ~1-2% of internalized small interfering RNA (siRNA) into the cytosol, and only transiently, from a narrow window as the endosome matures.³ Yet LNPs remain the clinically validated benchmark, and their escape mechanism is itself a form of stimulus-gating. The cationic ionizable lipid (CIL) – an amine-bearing lipid held near-neutral in blood – protonates only as the endosome acidifies, whereupon it ion-pairs with anionic endosomal phospholipids and drives the lamellar-to-inverted-hexagonal (H_II) transition that destabilizes the bilayer.⁴ Because the membrane-active species exists only at endosomal pH, the perturbation is intrinsically conditional. Lipid-tail architecture tunes it: a single one-carbon tail branch conferred a ~10-fold gain in mRNA delivery, attributed explicitly to stronger ionization at the late-endosomal pH of ~5.0.⁵ The branched, multi-tailed vaccine lipids SM-102 and ALC-0315 subsequently supplanted the linear-tailed DLin-MC3-DMA. A major practical limitation of systemically administered LNPs is their tendency to accumulate in the liver through apolipoprotein-E-mediated uptake pathways, even as endosomal escape remains a significant intracellular bottleneck.⁶

Through tuning the lipid formulations, however, the LNP can be directed towards other tissues – the approach termed selective organ targeting (SORT) – by favoring binding to serum proteins whose cognate receptors are expressed in that tissue.^7-9 This is a laborious process requiring extensive screening. Escape efficiency and endosomal toxicity are, moreover, coupled: across ionizable-lipid libraries the most escape-competent lipids tend to be those that most perturb the endosomal membrane and provoke inflammation, yet the coupling is breakable – lipids such as 4A3-SC8 deliver mRNA efficiently while triggering little inflammatory activation, evidence that potent escape need not demand indiscriminate membrane damage.¹⁰

Beyond formulation-based approaches such as SORT nanoparticles, active targeting has emerged as a complementary strategy for extrahepatic delivery. Antibodies, single-domain antibodies (VHHs), peptides, aptamers, carbohydrates, and other receptor ligands can be incorporated into LNPs to increase uptake into defined cell populations while reducing off-target exposure.¹¹ Increasingly, tissue targeting and endosomal escape are viewed as interconnected design variables. Delivery to the correct cell provides limited therapeutic benefit if cargo remains trapped within endosomes, whereas highly efficient escape mechanisms provide little advantage when particles accumulate primarily in off-target tissues. Thus, contemporary delivery systems increasingly seek to optimize biodistribution, intracellular trafficking, and escape as a unified problem rather than as independent objectives.

Trafficking as an escape determinant

Increasing evidence suggests that productive NAT delivery may depend as much on intracellular trafficking as on membrane-disruptive potency itself. Productive release frequently correlates with trafficking through late endosomes and the intraluminal vesicles (ILVs) they accumulate, suggesting that access to specific intracellular compartments may influence delivery efficiency independently of membrane activity.¹² Particularly intriguing is retrofusion, the process by which these ILVs re-fuse with the limiting (outer) membrane of the multivesicular body (MVB, the ILV-laden late endosome), potentially returning luminal contents to the cytosol.^13,14 Although the quantitative contribution of retrofusion to productive NAT delivery remains unresolved, increasing recognition of trafficking-dependent escape has motivated efforts to engineer carriers that influence endosomal maturation, sorting, recycling, and compartmental residence rather than simply increasing membrane disruption. In this view, intracellular trafficking itself becomes a target for delivery optimization.¹⁴

Cell-restricted and stimulus-gated escape: emerging routes

The following strategies build the escape-enhancing activity into the carrier itself, so that it acts only where it is wanted – within a target cell or compartment – or only when a local cue is present, rather than constitutively throughout the body.

Intrinsic-escape carriers. The endosomal escape vehicle (EEV) platform uses proprietary cyclic cell-penetrating peptides (CPPs) whose non-lytic escape is engaged only after the conjugate has been taken up by the target cell; the cyclic scaffold was designed specifically to overcome the poor uptake, poor escape, and toxicity of canonical linear CPPs. EEV-conjugated splice-modulating oligonucleotides restored dystrophin in striated muscle in a Duchenne muscular dystrophy mouse model and knocked down aberrant transcripts in facioscapulohumeral muscular dystrophy (FSHD) patient-derived myocytes.¹⁵

Masked, endosome-activated fusogens. A second approach follows how enveloped viruses breach membranes. The archetype is the influenza-derived fusogenic peptide INF7, an analog of the HA2 N-terminal fusion sequence. At neutral pH its glutamate-rich sequence is disordered and inert; as the endosome acidifies, the carboxylates protonate, the peptide collapses into an amphipathic α-helix, and it inserts into and fuses with the endosomal membrane. Shown below (Fig. 2) is the ribbon structure of the INF7 20-residue peptide, GLFEAIEGFIENGWEGMIDG, based upon the NMR coordinates of the wild-type influenza hemagglutinin fusion peptide HA2 at pH 5.

As a carrier-agnostic escape module it has been incorporated into polyplexes, lipoplexes, and lipid nanoparticles alike. Such viral-derived lytic peptides carry an inherent toxicity liability, so the low-toxicity exemplar is biological rather than viral: surfactant protein B (SP-B), an endogenous cationic amphiphilic lung-surfactant protein, promotes cytosolic siRNA delivery by endosomal membrane fusion rather than rupture. SP-B-derived peptides – a minimal N-terminal heptapeptide plus amphipathic helix – reproduce this activity, and peptidolipid-coated nanogels built from them matched state-of-the-art LNPs for inhaled-siRNA potency while provoking less inflammatory activation of bronchial epithelium and better resisting nebulization stress.¹⁶ Fusion-based escape, which merges membranes rather than tearing them, marks the low-toxicity end of the design space.

Enzyme-activated, in-situ generation. The most explicitly gated strategy makes the membrane-active species a pro-drug that does not exist until the carrier is degraded inside the cell. Rouge and colleagues build nucleic acid nanocapsules (NANs) – small, enzyme-degradable crosslinked DNA micelles carrying DNA-surfactant conjugates (DSCs) – using esterase-labile crosslinkers. Intracellular esterase cleavage releases multi-tailed branched surfactant conjugates in situ; the added hydrophobic tails raise membrane-destabilizing capacity, shown as greater membrane disruption in cell assays and molecular-dynamics simulations and as improved DNAzyme-mediated silencing and mRNA delivery relative to single-tailed conjugates.¹⁷ Because the active, membrane-perturbing species is generated only after intracellular degradation, it cannot weaken endosomes body-wide, and the authors frame the construct explicitly as a “pro-drug” for escape. The work is in vitro and the escape mechanism is not directly resolved, but the branched-tail principle recapitulates the ionizable-lipid lesson that multi-tailed geometry outperforms a linear one.⁵

Ligand-tethered enhancers. The simplest expression of spatial confinement is to conjugate the enhancer to the same targeting ligand that directs the cargo, so that the escape agent is delivered only where the cargo goes. This localizes the membrane perturbation to the target cell without requiring a novel escape chemistry and is the natural complement to receptor-targeted conjugates. For single-domain (VHH) antibody-functionalized LNPs, cytosolic mRNA delivery does not monotonically increase with surface ligand density but follows an inverted-U, peaking at an optimal antibody conjugation density – on the order of 0.1 functional single-domain antibody per 100 nm² for an anti-CD8 system – and declining on either side, which corresponds to approximately 20 antibodies per LNP with a diameter of 80 nm. The descending limb is mechanistically instructive: too high a labeling density on the LNP drives internalization and proteolytic degradation of the target receptor itself, depleting the pool of receptors that the antibody conjugate depends upon for uptake.¹⁸ A ligand-tethered escape enhancer therefore inherits an avidity ceiling, bounded by the density at which productive engagement, rather than receptor depletion, still dominates.

Biomimicry of extracellular vesicles. Nature already escapes the endosome efficiently. Extracellular vesicles (EVs) – cell-derived membrane carriers such as exosomes – have in several studies demonstrated substantially higher estimates of cytosolic delivery efficiency than synthetic LNPs, although reported values vary considerably depending on methodology, cargo identity, and cellular context.¹⁰ Importantly, they do so through endogenous and conditionally activated trafficking and membrane-remodeling pathways rather than through brute-force membrane disruption.

Four mechanisms recur, each triggered by the endosomal environment itself: direct membrane fusion driven by SNARE and Rab proteins and tetraspanins (membrane proteins such as CD63); transient pore formation nucleated by cone-shaped lipids and pH-sensitive insertion peptides; pH buffering that resists acidification; and back-fusion (retrofusion), in which an intraluminal vesicle re-fuses with the limiting membrane – the organelle’s outer, cytosol-facing boundary – discharging its contents into the cytosol and bypassing lysosomal delivery.¹⁰ Retrofusion is not merely postulated: it has been directly visualized and quantified in living cells as a constitutive feature of the multivesicular-body lifecycle, using the same fusion machinery that enveloped viruses exploit to breach the endosome.¹³ For the designer, these are a blueprint for conditional, fusion-based escape rather than lytic – motivating EV-mimetic lipids, fusogenic tetraspanins, and an LNP protein corona engineered toward escape-favoring partners – and they sit at the low-toxicity end of the design space precisely because they merge or locally remodel membranes instead of rupturing them.

Freely co-administered small-molecule enhancers: useful in vitro, unlikely as therapeutics

A distinct class sits apart from the carrier-based strategies above: freely co-administered small-molecule escape enhancers – oligonucleotide-enhancing compounds (OECs) such as Retro-1, together with high-throughput-screen hits and siRNA delivery-enhancer libraries. Added to cultured cells these reliably raise oligonucleotide and siRNA knockdown, and they are genuinely useful research tools – for boosting in vitro potency and for dissecting the trafficking steps that limit delivery.^19-21 Their weakness is the move from petri dish to body. The enhancement is delivery-system specific and has translated only modestly and inconsistently to animal models, and, more fundamentally, a compound that non-selectively renders endolysosomes leaky²² also releases their hydrolytic contents into the cytosol, an intrinsic toxicity ceiling that applies to every untargeted cell exposed.²³ The same permeabilizing activity that is tolerable and even helpful in a monolayer becomes a systemic liability in vivo. These enhancers therefore remain valuable in the laboratory yet unpromising as systemic therapeutics, which is precisely the argument for confining the enhancing activity to the target cell or the intracellular trigger, as the carrier-based strategies above do.

A unifying design motif

Across these strategies a single principle emerges: the escape-enhancing activity must be confined – by location (cell-restricted: EEV cyclic CPPs, ligand-tethered enhancers) or condition (stimulus-gated: the CIL pH-switch, masked fusogens, enzyme-activated pro-drugs) – so that endosomal membranes are destabilized only in the target cell and only once the cargo has arrived. That constraint, rather than raw membrane-lytic potency, is what separates a therapeutically viable enhancer from a toxic one. A final qualification tempers the enterprise: efficient escape is not always the goal. Where the cargo is sufficiently stabilized, the endosome itself can serve as a slow-release depot – GalNAc-siRNA conjugates such as vutrisiran achieve durable silencing not through efficient cytosolic delivery but through gradual leakage of retained, nuclease-resistant oligonucleotide – so that for some modalities, the design target is a sustained low flux rather than a maximized burst.¹⁰

---

References

1. Gilleron J, Querbes W, Zeigerer A, et al. Image-based analysis of lipid nanoparticle-mediated siRNA delivery, intracellular trafficking and endosomal escape. Nat Biotechnol. 2013;31(7):638-646. doi:10.1038/nbt.2612
2. Sahay G, Querbes W, Alabi C, et al. Efficiency of siRNA delivery by lipid nanoparticles is limited by endocytic recycling. Nat Biotechnol. 2013;31(7):653-658. doi:10.1038/nbt.2614
3. Wittrup A, Ai A, Liu X, et al. Visualizing lipid-formulated siRNA release from endosomes and target gene knockdown. Nat Biotechnol. 2015;33(8):870-876. doi:10.1038/nbt.3298
4. Hafez IM, Maurer N, Cullis PR. On the mechanism whereby cationic lipids promote intracellular delivery of polynucleic acids. Gene Ther. 2001;8(15):1188-1196. doi:10.1038/sj.gt.3301506
5. Hajj KA, Ball RL, Deluty SB, et al. Branched-tail lipid nanoparticles potently deliver mRNA in vivo due to enhanced ionization at endosomal pH. Small. 2019;15(6):e1805097. doi:10.1002/smll.201805097
6. Akinc A, Querbes W, De S, et al. Targeted delivery of RNAi therapeutics with endogenous and exogenous ligand-based mechanisms. Mol Ther. 2010;18(7):1357-1364. doi:10.1038/mt.2010.85
7. Cheng Q, Wei T, Farbiak L, Johnson LT, Dilliard SA, Siegwart DJ. Selective organ targeting (SORT) nanoparticles for tissue-specific mRNA delivery and CRISPR-Cas gene editing. Nat Nanotechnol. 2020;15(4):313-320. doi:10.1038/s41565-020-0669-6
8. Dilliard SA, Cheng Q, Siegwart DJ. On the mechanism of tissue-specific mRNA delivery by selective organ targeting nanoparticles. Proc Natl Acad Sci U S A. 2021;118(52):e2109256118. doi:10.1073/pnas.2109256118
9. Wang X, Liu S, Sun Y, et al. Preparation of selective organ-targeting (SORT) lipid nanoparticles (LNPs) using multiple technical methods for tissue-specific mRNA delivery. Nat Protoc. 2023;18(1):265-291. doi:10.1038/s41596-022-00755-x
10. Hagedorn L, Jürgens DC, Merkel OM, Winkeljann B. Endosomal escape mechanisms of extracellular vesicle-based drug carriers: lessons for lipid nanoparticle design. Extracell Vesicles Circ Nucl Acids. 2024;5(3):344-357. doi:10.20517/evcna.2024.19
11. Dilliard SA, Siegwart DJ. Passive, active and endogenous organ-targeted lipid and polymer nanoparticles for delivery of genetic drugs. Nat Rev Mater. 2023;8(4):282-300. doi:10.1038/s41578-022-00529-7
12. Crooke ST, Wang S, Vickers TA, Shen W, Liang XH. Cellular uptake and trafficking of antisense oligonucleotides. Nat Biotechnol. 2017;35(3):230-237. doi:10.1038/nbt.3779
13. Perrin P, Janssen L, Janssen H, et al. Retrofusion of intralumenal MVB membranes parallels viral infection and coexists with exosome release. Curr Biol. 2021;31(17):3884-3893.e4. doi:10.1016/j.cub.2021.06.022
14. Chahal GS, Helbig KJ, Parton RG, Monson EA. The biology of endosomal escape: strategies for enhanced delivery of therapeutics. ACS Nano. 2026;20:1789-1813. doi:10.1021/acsnano.5c18112
15. Li X, Kheirabadi M, Dougherty PG, et al. The endosomal escape vehicle platform enhances delivery of oligonucleotides in preclinical models of neuromuscular disorders. Mol Ther Nucleic Acids. 2023;33:273-285. doi:10.1016/j.omtn.2023.06.022
16. Herman L, Guagliardo R, Zamborlin A, et al. Surfactant protein B-derived peptides as endosomal escape enhancers for pulmonary delivery of siRNA. J Control Release. 2025;381:113571. doi:10.1016/j.jconrel.2025.02.067
17. Gorecki J, de la Fuente IF, Corrigan PM, et al. Enhancing the membrane destabilization capacity of DNA-surfactant conjugates for improved cytosolic delivery of nucleic acids. ACS Appl Mater Interfaces. 2026;18(34):45954-45970. doi:10.1021/acsami.6c01718. Originally posted as a preprint: ChemRxiv. 2025. doi:10.26434/chemrxiv-2025-5wn12
18. Hashiba K, Fukasaka M, Okuma C, et al. Balancing multivalent avidity and receptor availability governs mRNA delivery by antibody-functionalized lipid nanoparticles. Nano Lett. 2026;26(14):4630-4641. doi:10.1021/acs.nanolett.5c06525
19. Ming X, Carver K, Fisher M, et al. The small molecule Retro-1 enhances the pharmacological actions of antisense and splice switching oligonucleotides. Nucleic Acids Res. 2013;41(6):3673-3687. doi:10.1093/nar/gkt066
20. Yang B, Ming X, Cao C, et al. High-throughput screening identifies small molecules that enhance the pharmacological effects of oligonucleotides. Nucleic Acids Res. 2015;43(4):1987-1996. doi:10.1093/nar/gkv060
21. Gilleron J, Paramasivam P, Zeigerer A, et al. Identification of siRNA delivery enhancers by a chemical library screen. Nucleic Acids Res. 2015;43(16):7984-8001. doi:10.1093/nar/gkv762
22. Du Rietz H, Hedlund H, Wilhelmson S, Nordenfelt P, Wittrup A. Imaging small molecule-induced endosomal escape of siRNA. Nat Commun. 2020;11(1):1809. doi:10.1038/s41467-020-15300-1
23. Boya P, Kroemer G. Lysosomal membrane permeabilization in cell death. Oncogene. 2008;27(50):6434-6451. doi:10.1038/onc.2008.310