Efficient cellular transformation via protein delivery through the protrusion-derived extracellular vesicles

Abstract

Extracellular vesicles (EVs) mediate the transfer of intracellular proteins from producer to recipient cells. EVs originate either from plasma membrane protrusions or endosomes, with endosome-derived EVs being extensively studied and engineered. However, the efficiency and functionality of protein transfer via both types of EVs remain poorly understood. Here, we demonstrate that natural EVs derived from cell protrusions dependent on the I-BAR protein MIM, rather than from endosomes, deliver the functional small GTPase Rac1 protein at levels similar to microinjection. Rac1-containing EVs are internalized via endocytosis, trafficked through endosomal compartments, and subsequently released into the cytosol, where they enhance cell motility. To evaluate broader applicability, the genome-editing protein Cas12f is packaged into protrusion-derived EVs by MIM and endosome-derived EVs by endosomal tetraspanin CD63. Notably, protrusion-derived EVs deliver Cas12f with significantly higher efficiency than endosome-derived EVs, highlighting their superior capability for functional protein transfer. Our findings establish the protrusion-derived EVs as a powerful platform for the efficient and bioactive delivery of both native and engineered proteins, expanding the EV-based therapeutic strategies.

Fig. 1. The cell migration induced by exogenous Rac1 in EVs.

A Schematic representation of the role of EVs in the FBS on cell migration. FBS was heat-inactivated as a conventional treatment for cell culture, and EVs in it were depleted by ultracentrifugation. Cell migration was measured by the wound healing assay. B Cell migration during the wound healing of WT or CIP4 KO PANC-1 cells in the presence of FBS (Top) or EV-depleted FBS (Bottom). Cell edges at 0 h (dotted lines) and 12 h (solid lines) are shown. Scale bars, 100 μm. C Cell migration in the presence of FBS, EV-depleted FBS, or EV-depleted FBS supplemented with l-EV fraction from FBS (4 × 10⁶ cells were treated with 6 × 10⁸ EVs/ml in 1 ml). Dynamin inhibitor dynasore was applied at 40 μM. Cell migration areas are the wounded areas occupied by migrated cells after 12 h. Data show the means ± SD from 4 independent experiments. D Illustration of Rac1-containing EVs from HEK293 cell culture medium. E Cell migration in the presence of l-EV fraction from MIM I-BAR-expressing HEK293 cells (4 × 10⁶ cells were treated with 1.4 × 10⁹ EVs/ml in 1 ml). Dynasore was applied at 40 μM. Data show the means ± SD from 4 independent experiments. F Western blotting of Rac1 in the l-EV fractions of GFP or MIM I-BAR-expressing HEK293 cells and EV fractions from FBS. The number of EVs per lane was indicated. Representative blots from three technical replicates using the same lot of FBS are shown. G Western blotting of MIM in the EV fractions from FBS. A representative blot from the three technical replicates using the same lot of FBS is shown. H Cell migration under the l-EV fraction from FBS treated with the Rac1 inhibitor EHT1864 at 20 μM. 4 × 10⁶ cells were treated with 6 × 10⁸ EVs/ml in 1 ml. Data show the means ± SD from 3 independent experiments. Statistical significance was performed by one-way ANOVA with Tukey’s honestly significant difference (HSD) test. Source data are provided as a Source data file.

Fig. 2. The internalization of EVs and transport of Rac1 in endosomes.

A Schematic illustration of EV endocytosis, endosomal transport, and cargo release. B 3D-PALM/STORM live cell imaging of the CIP4-mEos4b (magenta)-expressing CIP4 KO PANC-1 cells with l-EV fraction having SF650T-Halo-MIM I-BAR (green). Images were recorded every 10 s. Scale bar, 500 nm. C Colocalization of CIP4-mEos4b with SF650T-Halo-MIM I-BAR in (B). The percentages of voxels co-localized in each frame were calculated (“RAW”). Co-localization of the CIP4 images with the 180°-rotated SF650T-Halo-MIM images is shown as “Randomized”. The 13 super-resolution images from independently seeded 5 cells, which resulted in 4, 1, 4, 2, and 2 images, respectively, were analyzed. D Colocalization of EVs from the l-EV fractions having both Halo-Rac1 and mCherry-MIM I-BAR with EEA1 (Top) or Lamp1 (Bottom) in WT cells after 3 h of incubation, in which 10⁶ cells were treated with 1.6 × 10⁹ EV/ml in 1 ml. Arrowheads in the enlarged images show their colocalization. Scale bars, 10 μm. E Quantification of the colocalization of Halo-Rac1 with EEA1 (Top) or Lamp1 (Bottom) in WT and CIP4 KO cells using Mander’s coefficients after EV addition. The EV fraction and the number of cells and EVs were the same as (D). The red lines indicate the means of 54 cells from 3 independent experiments. F PALM imaging of the l-EVs having SF650B-Halo-Rac1 in the late endosome, having Lamp1-mEos4b (Left). Enlargements of the square regions in the left images are on the Right. Arrowheads indicate possible release of SF650B-Halo-Rac1 from the endosome. Scale bars, 10 μm (Left) and 100 nm (Right). G Quantification of the Halo-Rac1 release from the endosome in (F). The percentage of EVs with crossover of Halo-Rac1 and Lamp1 in a cell is shown (17 cells from 3 independent experiments). All EVs with Lamp1-positive endosomes were counted. The full range of images with the marking of the counting of the EVs is shown in Supplementary Fig. 6B. The percentage of Halo-Rac1-releasing EVs was calculated as the percentage of “Halo-Rac1-releasing EVs” among the Halo-Rac1 EVs in the Lamp1-positive endosomes (endosome-colocalizing EVs). The means ± SD are shown. Statistical significance was determined by a two-tailed unpaired Mann–Whitney U test (C, E) or a two-tailed t-test with Welch’s correction (G). Source data are provided as a Source data file.

Fig. 3. Localization of the EV-derived exogenous Rac1 and lamellipodia formation in the recipient cells.

A Lamellipodia formation by the l-EV fraction from the cells expressing MIM. The WT or CIP4 KO PANC-1 cells were incubated with l-EV fraction from Halo-MIM I-BAR- or Halo-single expressing cells, in which 10⁶ PANC-1 cells were treated with 1.4 × 10⁷ EVs/ml in 1 ml for 3 h. After fixation, cells were stained for WAVE2 to visualize lamellipodia (arrowheads). Scale bars, 40 μm. Representative images from three independent experiments are shown. B PANC-1 cells incubated with l-EV fraction from mCherry-MIM I-BAR and Halo-Rac1 expressing cells, in which 2.5 × 10⁶ PANC-1 cells were treated with 0.1% DMSO or 100 nM Bafilomycin for 30 min and then with 2 × 10⁹ EVs in 0.5 ml of EV-depleted medium for 12 h. After fixation, cells were stained for HaloTag, mCherry, and F-actin. Arrowheads show Halo-Rac1 localization at the leading edge. Scale bars, 10 μm. C Quantification of cells with Halo-Rac1 localization at leading edges after incubating with the EVs for 12 h. Data show the means ± SD from 3 independent experiments, in which 7–10 microscopic fields were analyzed per experiment. Statistical significance was determined by a two-tailed unpaired t-test with Welch’s correction. Source data are provided as a Source data file.

Fig. 4. The stoichiometry of the EV-derived Rac1 in the recipient cells.

A Western blotting of the WT or CIP4 KO PANC-1 cells (10⁶ cells) treated with l-EV fraction from MIM I-BAR- and Halo-Rac1-co-expressing cells (8 × 10⁸ EVs/ml in 1 ml) for the indicated times to quantify the internalized Halo-Rac1 to endogenous Rac1. Each lane contains l-EV fraction from MIM I-BAR- and Halo-Rac1-co-expressing cells (5 × 10⁶ EVs for the blotting with anti-Halo or anti-β-actin antibodies, and 5 × 10⁷ EVs for the blotting with anti-Rac1 antibody), and the cell lysates from EV-treated cells (2 × 10⁵ cells) per lane. B Western blotting of purified Rac1 (25 ng) and l-EV fraction from Halo-Rac1- and MIM-co-expressing cells (5 × 10⁷ EVs) on the same membrane as (A) to determine the amount of Halo-Rac1 and endogenous Rac1 in EVs. C Halo-Rac1 molecules internalized per cell calculated from (A, B). Data show the means ± SD from 3 independent experiments. D The amount of the recipient cell-derived endogenous Rac1 molecules per cell from the western blotting of (A). Data show the means ± SD from 3 independent experiments. E Ratio of the internalized Halo-Rac1 to cellular endogenous Rac1 calculated from (C, D). Data show the means ± SD from 3 independent experiments. F Illustration of sandwich ELISA for Halo-Rac1. G Halo-Rac1 molecules internalized per cell were quantified by ELISA. PANC-1 cells incubated with l-EV fraction expressing MIM I-BAR and Halo-Rac1 for 1 and 3 h were measured, in which 6 × 10⁶ cells were treated with 1.5 × 10⁹ EVs/ml in 6 ml EV-depleted FBS-containing medium. The means ± SD are shown from 4 independent experiments. H The comparison of microinjected cells with the EV-treated cells. PANC-1 cells were microinjected with Halo-tagged Rac1 (300 μg/ml) or treated with Halo-tagged Rac1-containing EVs from MIM and Rac1 expressing cells at 4 × 10⁹ EVs/ml, 0.5 ml medium, and at 1 × 10⁵ or 1 × 10⁶ cells, respectively. After incubation for 30 min or 12 h, respectively, cells were fixed and stained with an anti-Halo antibody. Rhodamine-labeled dextran was co-injected to identify the injected cells. The injection amount, as monitored by Rhodamine fluorescence, was correlated with Halo protein amount. Actin filaments were visualized by phalloidin staining. Scale bars, 10 μm. Representative images from three independent experiments are shown. I Summary of the Rac1 molecule transfer via EVs. Statistical significance was determined by one-way ANOVA with Tukey’s HSD. Source data are provided as a Source data file.

Fig. 5. The internalization of the engineered EV cargo proteins through the MIM-dependent EVs.

A Schematic illustration of Cas12f loading into EVs and genome editing. Cas12f-Halo-FRB binds to FKBP-MIM in the presence of rapamycin and is loaded to the MIM-dependent EVs. The reporter cassette contains mCherry and two kinds of frame-shifted GFP, expressing GFP after the cutting of the target sequence by Cas12f and the addition/deletion of nucleotides during repair. B Western blotting of HEK293 cell lysates and their l-EV fractions from mCherry-FKBP-MIM I-BAR- and Cas12f-Halo-FRB-co-expressing cells, with or without rapamycin treatment. Lysates of the same number of cells (6 × 10³ cells) and EVs (7 × 10⁹ EVs) were analyzed. Representative blots from three independent experiments are shown. C–F Genome-editing of reporter cells by Cas12f-carrying EVs. Cas12f-reporter cells (2.5 × 10⁴ cells), including HEK293 (C, D) and PANC-1 (E, F), were treated with l-EV fraction (1.5 × 10¹⁰ EVs/ml in 0.2 ml). After 3 days, GFP-positive cells were counted under a fluorescence microscope (C and E, Left: representative images; Right: quantification from 3 independent experiments) or by flow cytometry (FCM, D and F, D from 6 independent experiments and F from 4 independent experiments). The bars in the graphs show the addition of EVs treated with (gray) or without (light gray) rapamycin or no EV addition (white). The l-EV fraction from cells expressing Cas12f and MIM 5KA was also examined (E). G–I Effect of heating or freezing on the genome editing efficiency of l-EVs. l-EV fractions (1.5 × 10¹⁰ EVs/ml in 0.2 ml) were heated or frozen as indicated and incubated with the reporter HEK293 cells (G, H) or PANC-1 cells (I) for genome editing as in (C–F). GFP-positive cells were counted by fluorescence images (G from 6 independent experiments, I from 3 independent experiments) or by FCM (H from 5 independent experiments). The bars in the graphs show the addition of EVs treated with (gray) or without (light gray) rapamycin or no EV addition (white). Data show the means ± SD. Statistical significance was performed by one-way ANOVA with Tukey’s HSD test (C, E–I) or Kruskal–Wallis test with Dunn’s multiple comparisons test (D). Source data are provided as a Source data file.

Fig. 6. The superiority of the protrusion-derived EVs in the Cas12f delivery over the endosome-derived EVs.

A Western blotting of the l-EV fraction and s-EV fraction from HEK293 cells co-transfected with the vector for expression of mCherry-FKBP-MIM I-BAR- or CD63-GFP-FKBP and Cas12f-Halo-FRB-co-expressing cells, with MIM-derived EVs or CD63-derived EVs with or without rapamycin treatment. Those from the cells transfected with a vector expressing Cas12f, followed by the T2A sequence and the I-BAR domain, were also analyzed. Lysates of the same number of cells (6 × 10³ cells) and EVs (7 × 10⁹ EVs) were analyzed. B Quantification of (A). Bar graphs show the addition of EVs from cells transfected with mCherry-FKBP-MIM I-BAR (green) or CD63-GFP-FKBP (yellow) followed by rapamycin treatment, those without rapamycin treatment (light gray), or no EV addition (white) throughout the Fig. 6B–F. C Genome-editing of reporter cells by Cas12f-carrying EV fractions. Cas12f-reporter HEK293 cells (2.5 × 10⁴ cells) were treated with l-EV fractions or s-EV fractions prepared as in (A) (1.5 × 10¹⁰ EVs/ml in 0.2 ml). After 3 days, their fluorescence was observed under microscope and the number of GFP-expressing cells was counted. D Genome editing of reporter cells by Cas12f-carrying EV fractions detected by FCM. The l-EV fractions and s-EV fractions prepared after rapamycin treatment were added to reporter cells as described in (C). GFP-expressing cells were counted by FCM. E, F Genome-editing efficiency of the Cas12f-containing EVs. The efficiency per Cas12f molecule (E) was determined by the GFP-positive cell number divided by the amount of Cas12f per EV (B), and the efficiency per added EV volume (F) was determined by GFP-positive cell number divided by the relative EV volume described in Supplemental fig. 14D. The efficiency was indicated as arbitrary units (arb. units). Data show the means ± SD (3 independent experiments for (B, C, E, F); 5 independent experiments for (D)). Statistical significance was performed by one-way ANOVA with Tukey’s HSD test (B, C) to compare among all data sets, or with Dunnett’s multiple comparisons test (D–F) to compare each data set with that of no EV (D) or that of l-EVs treated with DMSO (E, F). Source data are provided as a Source data file.