Split Inteins Offer a Viral-Protein-Free Route to Load Functional Enzymes into Extracellular Vesicles

Extracellular vesicles have long been touted as nature’s own delivery vehicles: nanoscale membrane-bound particles released by virtually every cell type, capable of ferrying proteins, RNA molecules, and other cargo between cells. Yet turning that biological promise into a reliable therapeutic platform has proven stubbornly difficult, largely because getting functional cargo into vesicles — and then getting that cargo to do useful work once it reaches a recipient cell — remains an inefficient and technically awkward process. A new study published in Scientific Reports by Razieh Ghazi Birjandi and Hesam Dehghani of Ferdowsi University of Mashhad in Iran now describes a one-step, viral-protein-free strategy that uses pH-sensitive split inteins to pack proteins into extracellular vesicles and demonstrate that those proteins can arrive at target cells in a biologically active form.

The central innovation rests on a pair of molecular tools called split inteins. Inteins are self-splicing protein elements that can excise themselves from a larger protein while stitching the flanking fragments back together with a peptide bond. When an intein is split into two non-functional halves — an N-terminal fragment and a C-terminal fragment — the two halves can reassociate spontaneously and reconstitute splicing activity without any external catalyst. The researchers exploited the well-characterized Npu split intein pair, fusing the N-terminal component, NpuN, to CD9, a tetraspanin membrane protein that is abundantly sorted into extracellular vesicles, and fusing the C-terminal component, NpuC, to the protein cargo of interest. The logic is elegant: because CD9 is naturally enriched on vesicle membranes, any cargo tethered to it through the reconstituted NpuN/NpuC interaction should be dragged along into the vesicle lumen or membrane during vesicle biogenesis.

What distinguishes this design from earlier intein-based loading schemes is the placement of a pH-sensitive linker between NpuC and the cargo. The researchers reasoned that this sequence would remain intact at neutral pH but would cleave under acidic conditions — precisely the environment found inside the intraluminal vesicles of producing cells and within the endosomes of recipient cells after vesicle uptake. In principle, this means the cargo is carried into vesicles while attached, then released from its membrane anchor when the vesicle interior or the recipient endosome acidifies, freeing the protein to diffuse into the cytosol and perform its function rather than remaining stuck to the vesicle membrane.

To test whether the system actually improves loading, the team fused two fluorescent reporter proteins, DsRed2 and tdTomato, to NpuC and compared engineered vesicles against vesicles loaded by passive means. The split-intein approach improved the loading of both reporters by approximately 1.3-fold relative to passive loading. That figure is modest rather than dramatic, but it represents a consistent, reproducible gain achieved with a single genetic construct and no chemical transfection reagents, electroporation, or viral envelope proteins — all of which complicate conventional loading workflows and raise safety or manufacturing concerns for clinical translation.

The most consequential demonstration, however, came when the researchers swapped the fluorescent reporters for a genuinely functional enzyme: the piggyBac transposase. This enzyme recognizes specific inverted terminal repeat sequences in DNA and excises any genetic material flanked by them, a property widely used in genome engineering. The team engineered donor cells to produce vesicles carrying NpuC-fused piggyBac transposase, isolated those vesicles, and added them to reporter cells whose genomes contained a 5′ITR–EGFP–polyadenylation signal–3′ITR cassette positioned so that its excision would switch on expression of mCherry. When the vesicles were taken up, the transposase excised the flanked cassette and the reporter cells turned red — direct evidence that a catalytically active enzyme had been loaded into vesicles, delivered across the recipient cell membrane, released into the cytosol, and had performed a genome-modifying reaction.

This functional delivery assay matters because most cargo-loading studies stop at showing that a fluorescent protein is present inside vesicles. Fluorescence can be detected without the protein being folded correctly, enzymatically competent, or accessible to the recipient cell’s cytoplasm. The piggyBac experiment closes that gap: excision of a genomic cassette is an all-or-nothing molecular event that cannot occur unless the transposase arrives intact and active. It also showcases a potential application beyond delivery for its own sake — vesicle-borne transposases could, in future work, be used to remove or rearrange specific genetic elements in living cells without exposing them to viral vectors.

A notable design choice is the deliberate avoidance of VSV-G, the vesicular stomatitis virus glycoprotein that is commonly used to pseudotype extracellular vesicles and enhance their fusion with target cell membranes. VSV-G pseudotyping improves delivery efficiency but imports a viral envelope protein into the system, which can alter vesicle tropism, trigger immune responses, and blur the line between a genuinely extracellular-vesicle-based therapy and a disguised viral vector. By building a loading and delivery system that works entirely with endogenous vesicle components — CD9 as the membrane anchor and the split intein as the molecular glue — the authors offer a platform that keeps the vesicles closer to their natural composition.

The characterization of the engineered vesicles relied on transmission electron microscopy to confirm vesicle morphology, dynamic light scattering to measure size distributions, and dot blot detection to verify the presence of the engineered mEmerald-CD9 membrane protein and the DsRed2 cargo in the vesicle fraction. These are standard and appropriate first-line techniques for confirming that the particles being studied are indeed membrane-bound vesicles of the expected size range and that the engineered constructs are physically associated with them.

The authors are candid about an important limitation: no endogenous extracellular vesicle markers or negative contamination markers were assessed in this study, which means vesicle identity and sample purity were not established by protein marker profiling. In the extracellular vesicle field, where preparations can be contaminated with protein aggregates, lipoproteins, and other non-vesicular particles, marker-based validation according to community guidelines is considered essential for attributing observed effects unambiguously to vesicles themselves. The functional piggyBac result strongly suggests that active enzyme was delivered in a membrane-associated form, but rigorous marker profiling will be needed before the platform’s mechanism of uptake and release can be stated with full confidence.

Even with those caveats, the study adds a genuinely useful tool to the extracellular vesicle engineering toolbox. The split-intein system is modular — any protein of interest can, in principle, be fused to NpuC — and it operates in a single step during natural vesicle production, avoiding post-isolation manipulation that can damage vesicles. The pH-sensitive release element adds a built-in trigger keyed to the acidic compartments that vesicles naturally encounter after uptake. If follow-up work confirms marker-defined purity, optimizes the 1.3-fold loading gain, and extends the functional delivery results to additional enzyme classes and primary cells, the approach could help extracellular vesicles move closer to their long-promised role as programmable, virus-free carriers of therapeutic proteins.

Subject of Research: Split-intein-based engineering of extracellular vesicles for protein cargo loading and functional delivery

Article Title: Split-intein-mediated engineering of VSV-G-free extracellular vesicles for enhanced cargo loading and delivery of functional proteins

Article References: Ghazi Birjandi, R., & Dehghani, H. (2026). Split-intein-mediated engineering of VSV-G-free extracellular vesicles for enhanced cargo loading and delivery of functional proteins. Scientific Reports. https://doi.org/10.1038/s41598-026-75437-9

Image Credits: AI Generated

DOI: 10.1038/s41598-026-75437-9

Keywords: extracellular vesicles, split inteins, cargo loading, CD9, tetraspanins, protein delivery, piggyBac transposase, VSV-G-free, pH-sensitive linker, nanocarriers, Scientific Reports, vesicle engineering