Bioinspired engineering of fusogen and targeting moiety equipped nanovesicles

Extracellular vesicles (EVs) are promising therapeutic carriers due to their natural membrane composition and biocompatibility, yet clinical translation is hindered by inadequate targeting, inefficient cytosolic delivery, low yield, and batch inconsistency. Cell-derived nanovesicles (CNVs), produced by mechanical extrusion, offer higher yield and lower cost than natural EVs while retaining many membrane proteins and consistency, making them attractive EV substitutes. While various strategies have been developed to introduce targeting moieties and fusogenic capability to EVs, endocytosis remains the predominant uptake route and often leads to lysosomal sequestration and degradation of vulnerable cargos (nucleic acids, peptides, proteins). The study aims to engineer CNVs that combine active targeting and efficient membrane fusion to achieve rapid endo-lysosomal escape and direct cytosolic delivery, thereby improving therapeutic efficacy for cytosolic targets. Specifically, the authors develop eFT-CNVs co-displaying an engineered, binding-defective but fusion-competent Sindbis viral fusogen and a membrane-bound anti-GPC3 single-chain variable fragment (scFv) for selective binding to GPC3-overexpressing cancer cells and subsequent fusogen-mediated membrane fusion.

Prior work highlights EV advantages over synthetic carriers but points to limitations in targeting and cytosolic delivery. CNVs generated by extrusion yield 50–100× more vesicles than EVs at <10% cost and share >70% membrane proteins with limited batch variation, supporting their use as EV mimetics. Targeting strategies include physical, chemical, and genetic display of moieties on EV membranes to improve homing. Fusogen-based or fusogenic modifications (viral fusogens like VSV-G, coiled-coil peptides, fusogenic peptides, EV–transfection lipid hybrids) have been explored to enhance membrane fusion and cytosolic delivery by promoting lipid mixing and pore formation. However, these approaches often lack active targeting, show modest fusion efficiency, or require impractical pre-modification of target cells. EV hybrids with cationic lipids can aid fusion but with limited efficiency and specificity. The authors position their approach to address these gaps by combining active targeting (anti-GPC3 scFv) with a bioengineered Sindbis fusogen on CNVs for efficient, selective cytosolic delivery at scalable yield.

• Donor cell engineering: HEK293 cells were double-edited via CRISPR/Cas9 to knock out GPC3 and B2M (to reduce self-aggregation and MHC-I–mediated immunogenicity). Clones were validated by Sanger sequencing and western blot. Donor cells were then transduced (lentiviral) to express a membrane-bound anti-GPC3 scFv and an engineered HA-tagged Sindbis viral fusogen (binding-defective, fusion-competent). Expression was confirmed by qPCR and flow cytometry (HA tag detection). Mycoplasma tests were negative.
• Functional validation in cells: Co-culture of engineered HEK293 (eT-HEK293: scFv-only; eFT-HEK293: scFv+fusogen) with GPC3+ HepG2 at acidic pH (5.5) showed fusogen-dependent formation of multinucleated polykaryons (cell–cell fusion). Blocking GPC3 with antibody reduced microcluster formation; pH did not affect scFv-mediated agglomeration, supporting scFv functionality.
• CNV preparation and characterization: CNVs, eT-CNVs, and eFT-CNVs were produced by mechanical extrusion of wild-type HEK293, eT-HEK293, and eFT-HEK293, followed by differential centrifugation, filtration, and ultracentrifugation. Sizes (NTA/TEM): ~114.9 nm (CNV), 124.3 nm (eT-CNV), 120.9 nm (eFT-CNV). TEM revealed saucer-shaped vesicles; western blots confirmed EV markers (CD81, TSG101). Anti-HA Fab–gold nanoparticle labeling allowed counting fusogen molecules per eFT-CNV (mean ~7 per vesicle, n=138).
• Fusion and uptake assays: Western blot detection of HA-tagged E2 domain in HepG2 membrane protein extracts after 30 min co-incubation with graded eFT-CNV numbers indicated dose-dependent fusion; with 1×10^6 eFT-CNVs, fusion efficiency approached ~100% under conditions described. Time-course (10–30 min) confirmed fusion detectable by 20 min. Fluorescence microscopy with PKH67-labeled vesicles and LysoView staining showed eT-CNVs primarily endocytosed and colocalized with lysosomes, whereas eFT-CNVs displayed diffuse intracellular fluorescence consistent with fusion-mediated cytosolic delivery and reduced endocytosis; minimal uptake in GPC3 knockout HepG2 or GPC3-low MCF7. Endocytic pathway inhibitors implicated clathrin-, caveolae-mediated endocytosis, and macropinocytosis in eT-CNV uptake. eFT-CNVs also fused with HepG2-derived EVs, evidenced by macroscopic agglomeration, TEM, and FRET assays (fusion efficiency ~64.7% at pH 5.5 within 30 min).
• Drug loading: siR-Sox2 loaded by electroporation; gelonin (30 kDa) and paclitaxel by sonication. Loading efficiencies: siR-Sox2 3.7±0.4%, gelonin 31.2±2.8%, paclitaxel 27.4±2.3%. Drug loading did not significantly differ among CNV types. Post-loading, vesicle size increased by ~8.7–35 nm; zeta potential shifted with cargo charge (siRNA/paclitaxel more negative; gelonin slightly more positive). Release studies (37 °C, pH 5.5 vs 7.4): faster release at pH 5.5, slow gelonin release (24 h: 10.3% at pH 5.5; 6.1% at pH 7.4); paclitaxel showed burst release in 1 h then sustained (24 h: ~73.2% at pH 5.5; 45.3% at pH 7.4).
• In vitro efficacy: HepG2 cells treated with equal vesicle counts (3×10^9 vesicles per condition) loaded with siR-Sox2, gelonin, or paclitaxel. For siR-Sox2 (200 nM equivalent): qPCR and ELISA showed Sox2 mRNA reduced to 17.6% and protein to 32.8% with eFT-CNV vs 75.7%/98% (CNV) and 62.6%/76.4% (eT-CNV). For paclitaxel (200 nM equivalent): cell cycle analysis showed G2/M arrest fractions of 86.9% (eFT-CNV), 85.1% (eT-CNV), 74.5% (CNV), vs 45.4% (free drug). IC50 assays across modalities compared free drug, CNV, eT-CNV, eFT-CNV formulations. Wound-healing, EdU proliferation, and transwell invasion assays quantified functional effects.
• In vivo efficacy and biodistribution: BALB/c mouse HepG2 xenografts (n=5/group) received intravenous dosing every 2–3 days for 3 weeks: siR-Sox2 (1 mg/kg), gelonin (2.5 mg/kg), paclitaxel (7.5 mg/kg) equivalents. Tumor volumes tracked; tissues analyzed histologically. Paclitaxel biodistribution assessed by HPLC at multiple timepoints; tumor accumulation kinetics compared to free paclitaxel. Body weight monitored; major organ histology evaluated for toxicity.
• Statistics: Data as mean±SD; two-tailed t-test and one-way ANOVA with p<0.05 significant.

• Engineered donor cells co-expressing anti-GPC3 scFv and a binding-defective, fusion-competent Sindbis fusogen produced eFT-CNVs via extrusion with high consistency. Fusogen expression verified; ~7 fusogen molecules per eFT-CNV on average.
• Targeting and fusion: eFT-CNVs rapidly bound GPC3+ HepG2 and MCF7 cells via scFv and fused with plasma membranes within ~20–30 min, enabling cytosolic delivery and minimizing lysosomal trafficking. eT-CNVs (no fusogen) were endocytosed and trafficked to lysosomes; CNVs lacked efficient binding. Fusion with HepG2 EVs achieved ~64.7% efficiency at pH 5.5 (30 min) by FRET.
• Drug loading and release: siR-Sox2 (3.7±0.4%), gelonin (31.2±2.8%), and paclitaxel (27.4±2.3%) loaded efficiently; acidic pH accelerated release, with gelonin releasing slowly (24 h ≤10.3% at pH 5.5), favorable for targeted cytosolic delivery.
• In vitro efficacy: • siR-Sox2 (200 nM equiv.): Sox2 mRNA reduced to 17.6% and protein to 32.8% with eFT-CNVs vs 75.7%/98% (CNV) and 62.6%/76.4% (eT-CNV). Treatment potency improved 3.9-fold vs CNV and 1.4-fold vs eT-CNV; free siRNA had minimal effect. • Paclitaxel (200 nM equiv.): G2/M arrest 86.9% (eFT-CNV), 85.1% (eT-CNV), 74.5% (CNV), 45.4% (free). IC50 improvements with eFT-CNVs: 25.6-fold vs free, 4.1-fold vs CNV, 1.7-fold vs eT-CNV. • Gelonin: eFT-CNVs enhanced potency 40-fold vs free gelonin, 8.1-fold vs CNV, 2.5-fold vs eT-CNV. • Functional assays: eFT-CNVs significantly reduced migration (e.g., siR-Sox2-loaded eFT-CNVs inhibited migration by 23.8-fold vs free siRNA and markedly vs CNV/eT-CNV), proliferation (EdU), and invasion compared to other groups.
• In vivo efficacy: • siR-Sox2: final tumor volume ~101.6±26.4 mm^3 (eFT-CNV) vs substantially larger in PBS/free/CNV/eT-CNV; eFT-CNV improved efficacy 15.7-fold vs free, 10.4-fold vs CNV, 7.7-fold vs eT-CNV. • Gelonin: final tumor volume ~40.6±16.4 mm^3; efficacy improved 34.1-fold vs free, 13.8-fold vs CNV, 9.3-fold vs eT-CNV. • Paclitaxel: final tumor volume ~4.6±2.7 mm^3; efficacy improved 84.1-fold vs free, 44.6-fold vs CNV, 9.5-fold vs eT-CNV. • Safety/pharmacokinetics: No significant body weight changes; no extensive organ damage histologically. Paclitaxel-loaded eFT-CNVs accumulated in tumors, peaking at ~24 h with significantly different kinetics vs free drug.

The study demonstrates that co-displaying a high-affinity targeting moiety (anti-GPC3 scFv) and a bioengineered viral fusogen on CNVs synergistically enables selective binding to tumor cells and rapid membrane fusion, achieving efficient endo-lysosomal escape and direct cytosolic delivery. This addresses key barriers in EV-based delivery—target specificity and cytosolic access—while leveraging CNV scalability and consistency. Compared to prior fusogenic EV approaches (e.g., VSV-G-decorated EVs with broad receptor usage, coiled-coil systems requiring pre-modified targets, EV–lipid hybrids with modest fusion), eFT-CNVs provide active, antigen-specific targeting and high fusion efficiency within minutes, translating into markedly enhanced efficacy especially for cytosolic macromolecules like gelonin and siRNA. Knocking out GPC3 prevents self-aggregation of vesicles presenting both antigen and scFv, and B2M knockout may reduce MHC-I–mediated immunogenicity, addressing potential safety concerns of CNVs and improving translational potential. The results show the greatest relative benefit for cargos whose activity depends on cytosolic access (gelonin, siRNA), while small-molecule paclitaxel benefits substantially from targeting with incremental gain from fusion. Overall, the findings support eFT-CNVs as a versatile platform for precision nanomedicine, with implications for gene editing and vaccination through efficient cytosolic nucleic acid/protein delivery.

The authors developed eFT-CNVs—cell-derived nanovesicles co-functionalized with an engineered Sindbis fusogen and anti-GPC3 scFv—that selectively bind GPC3-overexpressing cells and fuse with plasma membranes to deliver cargos directly into the cytosol. The platform is scalable via mechanical extrusion, shows consistent composition, and markedly enhances therapeutic efficacy versus free drugs and non-fusogenic/untargeted vesicles, particularly for cytosolic targets. In vitro and in vivo studies demonstrate superior gene silencing, cytotoxicity of gelonin, and chemotherapeutic outcomes with favorable safety. Future work should pursue pH-independent fusogens to broaden applicability beyond acidic microenvironments, optimize fusogen expression/copy number, and explore recombinant all-in-one fusion proteins integrating targeting and fusion domains. Potential applications include targeted CRISPR/Cas9 delivery for gene editing, mRNA vaccines or in situ CAR-T generation, and molecular diagnostics using beacon-loaded vesicles.

• pH dependence: The engineered Sindbis fusogen operates optimally at acidic pH, limiting efficacy in non-acidic microenvironments; pH-independent fusogens are desired.
• Fusion quantification: Western blot and fluorescence imaging used to infer fusion have inherent limitations and may not yield absolute efficiencies.
• Immunogenicity considerations: Although B2M/MHC-I knockout may reduce immunogenicity, CNVs generated by mechanical extrusion can carry membrane flips or debris that could present antigens; thorough purification is needed and residual immunogenicity remains uncertain.
• Cargo-specific assessments: Gelonin’s intracellular ribosome inhibition was not directly quantified in HepG2 cells due to assay constraints; siRNA potency may limit observable effect size.
• Stability/release: Increased release rates at acidic pH suggest potential vesicle instability; long-term stability and in vivo release kinetics beyond 24 h require further study.
• Model scope: Efficacy was demonstrated primarily in HepG2 xenografts and selected cell lines; broader tumor types, dosing regimens, and pharmacology are needed for generalizability.