Coro1a promotes the multivesicular body and plasma membrane fusion by facilitating PKM2-mediated SNAP-23 phosphorylation Authors: Zhenzhai Cai, Zhijie Li, Nannan Li, Xinyu He, Yang Jin, Daopo Lin, Beilei Fu, Xingjian Ma, Dongzhi Hong, Kezhi Lin, Hao Wu Published: 2026 ABSTRACT The release of exosomes involves the fusion of multivesicular bodies (MVBs) with the plasma membrane (PM), a process driven by the soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complex. However, the specific SNARE complex mediating MVB–PM fusion and its regulatory mechanism remain unclear. Here, we show that Coronin-1a (Coro1a) increases extracellular vesicle (EV) biogenesis by promoting the assembly of the STX-12-SNAP-23-VAMP-7 SNARE complex independent of neddylation. Mechanistically, Coro1a activates pyruvate kinase M2 (PKM2) to enhance SNAP-23 phosphorylation, thereby driving the recruitment of STX-12 and VAMP-7. Consequently, Coro1a-induced EV biogenesis is abolished by PKM2 inhibition or SNAP-23 silencing. Furthermore, Coro1a levels are elevated in lung tumor tissues and correlate with poor patient survival, suggesting that Coro1a-mediated enhancement of EV production contributes to tumor progression. Our data demonstrate that Coro1a facilitates EV biogenesis by promoting SNARE complex assembly during MVB and PM fusion. Introduction Extracellular vesicles (EVs) encapsulate a diverse array of proteins, nucleic acids, lipids, and metabolites from their parent cells. This cargo enables EVs to act as crucial mediators of intercellular communication1. Their functional roles are complex and context-dependent. Under physiological conditions, EVs perform essential functions; for instance, placental EVs mediate fetal immune escape, while intestinal EVs help maintain intestinal immune homeostasis2,3. Conversely, EVs are implicated in the pathogenesis of various diseases, including neurodegenerative disorders4, cardiovascular diseases5, and metabolic diseases6. In the context of cancer, EVs have been demonstrated to exert pivotal effects on tumor development, primarily by promoting an immunosuppressive tumor microenvironment. However, it is critical to note that certain EV subpopulations can, under specific conditions, enhance antigen presentation or activate cytotoxic lymphocytes, highlighting the functional heterogeneity of EVs7,8. Tumor cell-derived EVs (TEVs) can profoundly influence immune effector cell activation9. A key mechanism is through TEV-associated PD-L1, which has been reported to induce systemic immunosuppression10. Furthermore, our previous work suggests that TEV PD-L1 contributes to resistance against anti-PD-L1 therapy by acting as a molecular decoy for the therapeutic antibody11. Therefore, targeting the production of specific immunosuppressive TEVs represents a promising strategy to enhance antitumor immunity. Achieving this, however, requires a deeper understanding of the molecular mechanisms that govern EV biogenesis. EVs are broadly classified into two main subtypes based on their biogenesis pathways: ectosomes and exosomes. Ectosomes are formed through the direct outward budding and fission of the plasma membrane (PM). This process involves the recruitment of Tsg101 to the PM by arrestin domain-containing protein 1, facilitating VPS4-dependent ectosome generation12. In contrast, exosomes are generated through an endosome-dependent mechanism. The initial step involves PM invagination to generate early endosomes. These structures subsequently mature into multivesicular bodies (MVBs) through inward budding of the endosomal membrane, resulting in the accumulation of intraluminal vesicles (ILVs) within the MVB lumen¹. Mature MVBs face two principal destinations. One pathway leads to lysosomal degradation, a process predominantly mediated by the small GTPase Rab7. Previous findings have shown that neddylated Coronin-1a (Coro1a) enhances Rab7 recruitment to MVBs, thereby promoting lysosomal sorting and suppressing EV production13. The alternative fate of MVBs involves their trafficking to and fusion with the PM, resulting in the extracellular release of ILVs as exosomes. This trafficking step is primarily regulated by Rab2714. The fusion event, the final step in exosome biogenesis mainly executed by soluble N-ethylmaleimide-sensitive fusion attachment protein (SNAP) receptor (SNARE) complexes. SNARE complexes serve as the fundamental molecular machinery driving membrane fusion. Typically composed of one vesicle-associated SNARE (v-SNARE) and three target-membrane-associated SNARE (t-SNARE) proteins, they assemble through interactions between the N-terminal regions of their SNARE motifs. This process nucleates the formation of a four-helix bundle trans-complex, which provides the energy required for vesicle fusion15. Indeed, specific SNARE components-including the v-SNARE VAMP-3, the t-SNARE syntaxin-3 (STX-3), and SNAP-23-have been implicated in the regulation of EV biogenesis16–18. However, fundamental questions remain unresolved, such as which specific SNARE complex orchestrates the fusion of MVBs with the PM during EV release, and what precise regulatory mechanisms govern its assembly? Clarifying these mechanisms is critical for understanding EV-mediated intercellular communication. Coro1a, a member of the coronin family known for its association with F-actin19, exhibits dual regulatory roles in EV biogenesis. As previously established, neddylated Coro1a facilitates MVB degradation, thereby inhibiting EV release. In this study, Coro1a was also identified to modulate EV biogenesis through a neddylation-independent mechanism13. However, the precise molecular pathways through which Coro1a regulates EV production independently of neddylation remain unclear. To ensure methodological precision, we note that while this study primarily investigates endosomal pathway-dependent EV biogenesis (commonly referred to as exosomes), our EV isolations likely contained a mixture of vesicles, including small ectosomes. Consequently, we consistently employ the term “EVs” throughout this work to accurately reflect this heterogeneity and avoid potential misinterpretation. Here, we confirmed that Coro1a regulates EV biogenesis through a neddylation-independent pathway. This regulatory role was demonstrated by the observation that Coro1a overexpression enhanced EV release, while its silencing suppressed it, independent of neddylation status. Mechanistically, Coro1a activates pyruvate kinase M2 (PKM2), inducing its phosphorylation. This activation enhances the association between PKM2 and SNAP-23 and consequently increases SNAP-23 phosphorylation. The phosphorylation of SNAP-23, in turn, promotes its recruitment of STX-12 and VAMP-7. This cascade facilitates the assembly of the STX-12-SNAP-23-VAMP-7 SNARE complex, which in turn drives the fusion of MVBs with the PM, ultimately augmenting EV production. Collectively, our findings elucidate a previously unrecognized mechanism wherein Coro1a orchestrates the formation of the SNARE complex to promote MVB-PM fusion and EV biogenesis. Results Coro1a enhances EV release by promoting PM and MVB fusion Coro1a is previously established to regulate EV biogenesis through both neddylation-dependent and -independent mechanisms, but the latter remained mechanistically undefined13. To address this, we overexpressed Coro1a in HeLa cells treated with MLN4924, a potent inhibitor of E1 NEDD8-activating enzyme that blocks neddylation. Consistent with prior findings, Coro1a overexpression significantly increased the release of CD9+CD63+ and CD81+ CD63+ EVs from MLN4924-treated 293 T and HeLa cells (Fig. 1A and Supplementary Fig. 1A). This was further validated by elevated EV protein content and particle counts (Supplementary Fig. 1B, C), as well as enhanced signals for EV markers (CD63, Alix, Tsg101, CD81) in purified and non-purified EVs (Fig. 1B and Supplementary Fig. 1D). Notably, Coro1a did not affect cell viability or EV size and morphology (Supplementary Fig. 1E–G). To ensure that the observed phenotype was not an artifact of MLN4924 treatment, we compared EV production between wild-type HeLa cells and HeLa CORO1A-/- cells under normal, untreated conditions. We found that HeLa CORO1A-/- cells secreted significantly fewer EVs than their wild-type counterparts, even in the absence of MLN4924 (Fig. 1C and Supplementary Fig. 1H), which was further support by the reduced marker proteins in EVs from these cells (Fig. 1D). This result provides further evidence that Coro1a functions as a positive regulator of EV biogenesis, independent of its neddylation status.Fig. 1Coro1a enhances EV release by promoting PM and MVB fusion.A 293 T and HeLa cells ± Coro1a overexpression were treated with 100 nM MLN4924 for 12 h. EVs in the supernatants were captured with anti‑CD63‑coated beads, and then CD9+CD63+ or CD81+CD63+ EVs were analyzed by flow cytometry (n = 3). B EVs were isolated from equal numbers of the corresponding 293 T cells in (A). Western blotting analysis to detect the indicated EV markers. C Nano-flow cytometry analysis of EV concentration in the supernatants of HeLa and HeLa CORO1A-/- cells in the absence of MLN4924 (n = 3). D EVs were isolated from equal numbers of HeLa and HeLa CORO1A-/- cells. Western blotting analysis to detect the indicated EV markers. E 3D quantification of CD63+ MVBs in the corresponding HeLa cells in A by z-stack confocal microscope; CD63 spot numbers and particle size were statistically analyzed (n = 5). TIRF microscopy analysis of the CD63+ MVB distribution in cells in A (F) or in HeLa and HeLa CORO1A-/- cells transfected with mcherry-CD63 plasmids in the presence of MLN4924 (G); CD63 spot numbers in the subplasmalemmal region were statistically analyzed (n = 10–13 in E, n = 9 in F). Scale bar, 10 μm. H Mean diffusion coefficient (Dxy) values for the individual trajectories of MVBs in E are shown (n = 42 in HeLa and n = 83 in 293 T). I, J Measurement of MVB-PM fusion activity in individual HeLa or 293 T cells (n = 25) in A or in HeLa and HeLa CORO1A-/- cells transfected with CD63-pHluorin in the presence of MLN4924 (n = 10) as analyzed with the AMvBE macro. Representative results from three independent experiments are shown. ns not significant; *P < 0.05; **P < 0.01 and ***P < 0.001 (unpaired two-tailed Student’s t-test; mean ± s.d.; except for unpaired Mann-Whitney test and median ± interquartile ranges in H). See Supplementary Data 1 for the exact P-values. Coro1a enhances EV release by promoting PM and MVB fusion. A 293 T and HeLa cells ± Coro1a overexpression were treated with 100 nM MLN4924 for 12 h. EVs in the supernatants were captured with anti‑CD63‑coated beads, and then CD9+CD63+ or CD81+CD63+ EVs were analyzed by flow cytometry (n = 3). B EVs were isolated from equal numbers of the corresponding 293 T cells in (A). Western blotting analysis to detect the indicated EV markers. C Nano-flow cytometry analysis of EV concentration in the supernatants of HeLa and HeLa CORO1A-/- cells in the absence of MLN4924 (n = 3). D EVs were isolated from equal numbers of HeLa and HeLa CORO1A-/- cells. Western blotting analysis to detect the indicated EV markers. E 3D quantification of CD63+ MVBs in the corresponding HeLa cells in A by z-stack confocal microscope; CD63 spot numbers and particle size were statistically analyzed (n = 5). TIRF microscopy analysis of the CD63+ MVB distribution in cells in A (F) or in HeLa and HeLa CORO1A-/- cells transfected with mcherry-CD63 plasmids in the presence of MLN4924 (G); CD63 spot numbers in the subplasmalemmal region were statistically analyzed (n = 10–13 in E, n = 9 in F). Scale bar, 10 μm. H Mean diffusion coefficient (Dxy) values for the individual trajectories of MVBs in E are shown (n = 42 in HeLa and n = 83 in 293 T). I, J Measurement of MVB-PM fusion activity in individual HeLa or 293 T cells (n = 25) in A or in HeLa and HeLa CORO1A-/- cells transfected with CD63-pHluorin in the presence of MLN4924 (n = 10) as analyzed with the AMvBE macro. Representative results from three independent experiments are shown. ns not significant; *P < 0.05; **P < 0.01 and ***P < 0.001 (unpaired two-tailed Student’s t-test; mean ± s.d.; except for unpaired Mann-Whitney test and median ± interquartile ranges in H). See Supplementary Data 1 for the exact P-values. To dissect the neddylation-independent mechanism, we focused on CD63+ MVBs. 3D quantification in z-stack confocal images confirmed that Coro1a reduced CD63+ MVBs in MLN4924-treated HeLa cells (Fig. 1E). Hepatocyte growth factor-regulated tyrosine kinase substrate (HRS) has been previously used as a marker of MVBs, being present in intermediates between early and late MVBs positive for Tsg101 and LBPA20,21. A phenotype also confirmed the reduction of HRS+ MVBs by Coro1a (Supplementary Fig. 1I). However, Coro1a did not influence early endosome (EE) numbers (Supplementary Fig. 1J) or ILV budding (Supplementary Fig. 1K), excluding effects on MVB formation. Similarly, lysosomal degradation of MVBs (assessed via LAMP1 colocalization) remained unaffected (Supplementary Fig. 1L). Since Coro1a showed no effects on EV biogenesis and degradation, we assumed whether Coro1a increases EV secretion by promoting MVB and PM fusion. If so, Coro1a should induce MVB docking onto the PM. Next, we used total internal reflection fluorescent (TIRF) microscopy to evaluate PM docking of MVBs. To monitor the CD63+ MVBs, we transfected HeLa cells with plasmids expressing the CD63 fluorescent fusion protein and then transfected control or Coro1a into these cells. Within the analyzed cell populations, and in contrast to the reduced CD63+ MVBs in total cells, cells in the Coro1a-overexpression group exhibited a significant increase in CD63+ MVBs in the subplasmalemmal region of MLN4924-treated HeLa and 293 T cells (Fig. 1F). Then, we transfected plasmids expressing CD63 fluorescent fusion protein into HeLa and HeLa CORO1A-/- cells, and confirmed the comparable transfection efficacy of CD63 (Supplementary Fig. 1M). In contrast to Coro1a overexpression, genetic ablation of Coro1a led to a pronounced population-wide decrease in subplasmalemmal CD63⁺ MVBs in MLN4924-treated HeLa cells (Fig. 1G). Vesicle docking represents a critical initial step in membrane fusion22 and is accompanied by restricted vesicle mobility upon PM association23. To confirm whether non-neddylated Coro1a enhances EV production by increasing the docking of MVBs and PM, the coefficient Dxy of CD63+ MVBs, an index of mobility, was calculated along vesicle trajectories using a rolling analysis window. Consistent with enhanced docking, Dxy values were significantly reduced in both MLN4924-treated HeLa and 293 T cells upon Coro1a overexpression (Fig. 1H). To directly assess the role of Coro1a in promoting MVB fusion and release at the plasma membrane, we next performed TIRF microscopy experiments using a pH-sensitive CD63 reporter (CD63-pHluorin) in MLN4924-treated HeLa and 293 T cells. CD63-pHluorin fluorescence is quenched within the acidic environment of MVBs (low pH) but undergoes a rapid increase upon fusion with the PM and exposure to the neutral extracellular environment24. Consistent with population-level observations, cells in the Coro1a-overexpression Coro1a overexpression markedly increased the frequency of fluorescence bursts, whereas Coro1a knockout resulted in a substantial decrease in fusion-associated signals (Fig. 1I, J and Videos S1–S6). These findings indicate a marked reduction in fusion events between CD63-pHluorin-labeled MVBs and the plasma membrane, thereby supporting Coro1a’s essential contribution to the stepwise progression from MVB docking to exocytic release. Together, our data suggest that non-neddylated Coro1a drives EV release by facilitating MVB-PM fusion. Coro1a induces the formation of STX-12-SNAP-23-VAMP-7 complex SNARE complexes are essential to regulate exocytosis by promoting vesicle fusion with PM25. The SNARE complex, comprising v-SNARE VAMP-3 and t-SNAREs STX-3/SNAP-2316–18, has been demonstrated to regulate EV secretion. Our functional studies revealed that SNAP-23, but not VAMP-3 or STX-3, is specifically required for Coro1a-mediated enhancement of CD81+CD63+ EV secretion from MLN4924-treated HeLa cells (Fig. 2A and Supplementary Fig. 2A, B). SNAP-23 silencing also reduced CD9+CD63+ EV secretion from MLN4924-treated HeLa cells (Fig. 2A). This effect was SNAP-23 specific because SNAP-23 rescue recovered EV secretion from SNAP23 siRNA-transfected HeLa cells with MLN4924 treatment (Supplementary Fig. 2C). SNAP-23 silencing also reduced EV release from MLN4924-treated 293 T cells (Fig. 2B). Then, we found that decreased EV secretion from MLN4924-treated HeLa CORO1A-/- cells was also no longer detected after SNAP-23 silencing (Supplementary Fig. 2D). To confirm the specific role of Coro1a in regulating EV production in HeLa CORO1A-/- cells, we rescued its expression in these cells, restoring EV production to wild-type levels (Supplementary Fig. 2E). Mechanistically, at population level, Coro1a overexpression redistributed SNAP-23 to CD63+ MVBs in MLN4924-treated HeLa cells (Fig. 2C), demonstrating that non-neddylated Coro1a promotes EV biogenesis via SNAP-23.Fig. 2Coro1a induces the formation of the STX-12-SNAP-23-VAMP-7 complex.HeLa (A) or 293 T (B) cells with or without Coro1a overexpression were transfected with NC or SNAP23 siRNA for 24 h. Then, the cells were treated with 100 nM MLN4924 for 12 h, and the EVs in the supernatants were captured with anti‑CD63‑coated beads, followed by flow cytometric analysis of CD9+CD63+ or CD81+CD63+ EVs (n = 3), and nano-flow cytometry was employed to measure the EV concentration (n = 3). C Confocal microscopy and Pearson’s correlation coefficient (PCC) analysis of the colocalization of CD63 and SNAP-23 in MLN4924-treated HeLa cells with or without Coro1a overexpression. Scale bar, 10 μm. D Western blotting analysis of STX-12 and VAMP-7 in the lysates of MLN-treated 293 T cells with or without Coro1a overexpression after IP with anti-SNAP-23. E Confocal microscopy and fluorescence intensity analysis of STX-12, VAMP-7, and SNAP-23 colocalized in MLN4924-treated HeLa cells with or without Coro1a overexpression. Scale bar, 10 μm. 293 T (F) or HeLa (G) cells were transfected with NC, VAMP7, or STX12 siRNA for 48 h. Then, the EVs in the supernatants were captured with anti‑CD63‑coated beads, followed by flow cytometric analysis of CD9+CD63+ or CD81+CD63+ EVs (n = 3), and nano-flow cytometry was employed to measure the EV concentration (n = 3). H Confocal microscopy imaging of STX12 localization in DiI-labeled HeLa cells. Scale bar, 10 μm. I Time-dependent liposome fusion was measured from the development of FRET between the DiD-labeled STX-12 liposomes and the DiI-labeled VAMP-7Δ12-3 M liposomes. Representative results from three independent experiments are shown. ns not significant; **P < 0.01 and ***P < 0.001 (One-way ANOVA followed by Turkey test in A, B; unpaired two-tailed Student’s t test in E, F; mean ± s.d.). See Supplementary Data 1 for the exact P-values. Coro1a induces the formation of the STX-12-SNAP-23-VAMP-7 complex. HeLa (A) or 293 T (B) cells with or without Coro1a overexpression were transfected with NC or SNAP23 siRNA for 24 h. Then, the cells were treated with 100 nM MLN4924 for 12 h, and the EVs in the supernatants were captured with anti‑CD63‑coated beads, followed by flow cytometric analysis of CD9+CD63+ or CD81+CD63+ EVs (n = 3), and nano-flow cytometry was employed to measure the EV concentration (n = 3). C Confocal microscopy and Pearson’s correlation coefficient (PCC) analysis of the colocalization of CD63 and SNAP-23 in MLN4924-treated HeLa cells with or without Coro1a overexpression. Scale bar, 10 μm. D Western blotting analysis of STX-12 and VAMP-7 in the lysates of MLN-treated 293 T cells with or without Coro1a overexpression after IP with anti-SNAP-23. E Confocal microscopy and fluorescence intensity analysis of STX-12, VAMP-7, and SNAP-23 colocalized in MLN4924-treated HeLa cells with or without Coro1a overexpression. Scale bar, 10 μm. 293 T (F) or HeLa (G) cells were transfected with NC, VAMP7, or STX12 siRNA for 48 h. Then, the EVs in the supernatants were captured with anti‑CD63‑coated beads, followed by flow cytometric analysis of CD9+CD63+ or CD81+CD63+ EVs (n = 3), and nano-flow cytometry was employed to measure the EV concentration (n = 3). H Confocal microscopy imaging of STX12 localization in DiI-labeled HeLa cells. Scale bar, 10 μm. I Time-dependent liposome fusion was measured from the development of FRET between the DiD-labeled STX-12 liposomes and the DiI-labeled VAMP-7Δ12-3 M liposomes. Representative results from three independent experiments are shown. ns not significant; **P < 0.01 and ***P < 0.001 (One-way ANOVA followed by Turkey test in A, B; unpaired two-tailed Student’s t test in E, F; mean ± s.d.). See Supplementary Data 1 for the exact P-values. To identify the SNARE complex mediating fusion between MVBs and the PM, we performed immunoprecipitation of ectopically expressed SNAP-23 in MLN4924-treated 293 T cell lysates, followed by MS/MS analysis. This revealed that SNAP-23 specifically interacted with SNARE proteins, including STX-12 and VAMP-7 (Supplementary Fig. 2F). Notably, Coro1a overexpression significantly increased the co-immunoprecipitation of VAMP-7 and STX-12 with SNAP-23 in MLN4924-treated 293 T cells (Fig. 2D), whereas the opposite trend was observed in MLN4924-treated HeLa CORO1A-/- cells (Supplementary Fig. 2G). Immunofluorescence staining further confirmed that Coro1a promoted colocalization of SNAP-23, VAMP-7, and STX-12 in MLN4924-treated HeLa cells, supporting a role for Coro1a in enhancing SNARE complex assembly at MVB-PM fusion sites. Statistical analysis further demonstrated a enhanced tendency in colocalization among SNAP-23, VAMP-7, and STX-12 at population level, which exhibited a significant difference (Fig. 2E). Furthermore, VAMP-7 or STX-12 silencing significantly reduced CD81+CD63+ and CD9+CD63+ EV secretion by 293 T and HeLa cells (Fig. 2F, G and Supplementary Fig. 2H). Again, impaired EV production by VAMP7 or STX12 siRNA transfection was fully reversed by VAMP7 or STX-12 rescue (Supplementary Fig. 2I). A previous study has reported an endosomal distribution of STX-1226. Through immunofluorescence detection of STX-12 localization, we found a potential enrichment of STX12 in proximity to the PM region (Fig. 2H), which also suggests the possibility that STX-12 regulates EV secretion. We next conducted lipid-mixing experiments to explore whether VAMP-7, STX-12, and SNAP-23 form a functional SNARE complex to drive MVB-PM fusion. Since VAMP-7 favors a closed conformation with its N-terminal longin domain bound to its SNARE motif, we introduced Y45E/I139S/I144S mutation (3 M) to facilitate the opening of VAMP-727–29 and eliminated the C-terminal 12-amino acids to improve VAMP-7 solubility. With lipid-mixing assays, we observed significant membrane fusion between VAMP-7Δ12-3 M liposomes and STX-12 liposomes in the presence of SNAP-23 (Fig. 2I and Supplementary Fig. 2J), demonstrating that the SNARE complex composed of STX-12, SNAP-23, and VAMP-7 is competent in mediating MVB-PM fusion. These results suggest that Coro1a promotes STX-12-SNAP-23-VAMP-7 complex formation in a neddylation-independent manner. Coro1a facilitates PKM2-mediated SNAP-23 phosphorylation Next, we wondered how non-neddylated Coro1a regulates the formation of the STX-12-SNAP-23-VAMP-7 complex. PKM2 is involved in EV biogenesis by inducing Serine/Threonine phosphorylation (p-Ser/Thr) of SNAP-2318. MS/MS results revealed that PKM2 was also pulled down by SNAP-23 (Supplementary Fig. 2E). We did detect the interaction between PKM2 and SNAP-23 in 293 T cells, which was not affected by MLN4924 (Fig. 3A). So, we explored the role of PKM2 in non-neddylated Coro1a-mediated regulation of the formation of the STX-12-SNAP-23-VAMP-7 complex. First, we found that Coro1a induced SNAP-23 phosphorylation in MLN4924-treated 293 T cells (Fig. 3B). On the contrary, lowered phosphorylated SNAP-23 was detected in MLN4924-treated HeLa CORO1A-/- cells (Supplementary Fig. 3A). Consistent with its function on EV biogenesis, we found that constitutive-active SNAP-23 mutant (SNAP23-S95E) remarkably promoted CD81+CD63+ EV release by 293 T cells, but dominant-negative SNAP-23 mutant (SNAP23-S95A) did not show this effect (Fig. 3C). In contrast, SNAP23-S95A started to inhibit CD81+CD63+ EV release with higher overexpression (Supplementary Fig. 3B). Furthermore, we found that SNAP-23-S95E overexpression abolished the difference in CD81+CD63+ EV secretion between MLN4924-treated HeLa CORO1A-/- and HeLa cells (Fig. 3D). These results indicate that Coro1a regulates neddylation-independent EV secretion by affecting p-Ser/Thr of SNAP-23. Then, we found that PKM2 inhibitor Shikonin eliminated Coro1a-induced SNAP-23 phosphorylation (Fig. 3E). Correspondingly, Shikonin commensurated the total numbers of CD63+ MVBs between MLN4924-treated HeLa CORO1A-/- and HeLa cells at population-level observations (Fig. 3F). Furthermore, we found comparable STX-12-SNAP-23-VAMP-7 complexes across the entire cell population (Fig. 3G). Similar results were obtained by PKM2 silencing (Supplementary Fig. 3C–E). MLN4924 has been reported to increase glycolysis by activating PKM2 via promoting its tetramerization30. To exclude the interference of MLN4924 on this process, we inhibited neddylation by NEDD8 silencing and found that PKM2 silencing abolished Coro1a-induced p-Ser/Thr of SNAP-23 in 293 T cells with NEDD8 silencing (Supplementary Fig. 3F). Furthermore, Coro1a deficiency-induced decrease in SNAP-23 interaction with VAMP-7 and STX-12 was also negated in NEDD8-silenced HeLa cells along with PKM2 silencing (Supplementary Fig. 3G). Subsequently, we found a similar subplasmalemmal distribution of CD63+ MVBs (Fig. 3H), and a similar fusion activaity of CD63+ MVBs and PM (Fig. 3I and Videos S7–S10) between Shikonin-treated HeLa CORO1A-/- and HeLa cells in the presence of MLN4924. Therefore, Coro1a increases SNAP-23 phosphorylation by facilitating PKM2 activation.Fig. 3Coro1a facilitates PKM2-mediated SNAP-23 phosphorylation.A Western blotting analysis of PKM2 in the lysates of 293 T cells ± 100 nM MLN4924 for 12 h after IP with anti-SNAP-23. B Western blotting analysis of p-Ser/Thr of SNAP-23 in the lysates of MLN4924-treated 293 T cells ± Coro1a overexpression after IP with anti-SNAP-23. C 293 T cells were transfected with SNAP23-S95E or SNAP23-S95A for 48 h. Then, the EVs in the supernatants were captured with anti‑CD63‑coated beads, followed by flow cytometric analysis of CD81+CD63+ EVs (n = 3). D MLN4924-treated HeLa CORO1A-/- and HeLa cells were transfected with SNAP23-S95E for 48 h. Then, the EVs in the supernatants were captured with anti‑CD63‑coated beads, followed by flow cytometric analysis of CD81+CD63+ EVs (n = 3). E Western blotting analysis of p-Ser/Thr of SNAP-23 in the lysates of MLN4924-treated 293 T cells ± Coro1a overexpression, accompanied by 1 μM Shikonin treatment for 12 h after IP with anti-SNAP-23. F 3D quantification of CD63+ MVBs in Shikonin-treated HeLa and HeLa CORO1A-/- cells in the presence of MLN4924 by z-stack confocal microscope; CD63 spot numbers and particle size were statistically analyzed (n = 5). G Western blotting analysis of VAMP-7 and STX-12 in Shikonin-treated HeLa and HeLa CORO1A-/- cells in the presence of MLN4924 after IP with anti-SNAP-23. H TIRF microscopy analysis of the CD63+ MVB distribution in Shikonin-treated HeLa and HeLa CORO1A-/- cells in the presence of MLN4924; CD63 spot numbers in the subplasmalemmal region were statistically analyzed (n = 6–10). Scale bar, 10 μm. I Fusion activity analysis of the CD63+ MVB distribution in Shikonin-treated HeLa and HeLa CORO1A-/- cells in the presence of MLN4924 (n = 9). Representative results from three independent experiments are shown. ns not significant; *P < 0.05; ***P < 0.001 (unpaired two-tailed Student’s t test; mean ± s.d.). See Supplementary Data 1 for the exact P-values. Coro1a facilitates PKM2-mediated SNAP-23 phosphorylation. A Western blotting analysis of PKM2 in the lysates of 293 T cells ± 100 nM MLN4924 for 12 h after IP with anti-SNAP-23. B Western blotting analysis of p-Ser/Thr of SNAP-23 in the lysates of MLN4924-treated 293 T cells ± Coro1a overexpression after IP with anti-SNAP-23. C 293 T cells were transfected with SNAP23-S95E or SNAP23-S95A for 48 h. Then, the EVs in the supernatants were captured with anti‑CD63‑coated beads, followed by flow cytometric analysis of CD81+CD63+ EVs (n = 3). D MLN4924-treated HeLa CORO1A-/- and HeLa cells were transfected with SNAP23-S95E for 48 h. Then, the EVs in the supernatants were captured with anti‑CD63‑coated beads, followed by flow cytometric analysis of CD81+CD63+ EVs (n = 3). E Western blotting analysis of p-Ser/Thr of SNAP-23 in the lysates of MLN4924-treated 293 T cells ± Coro1a overexpression, accompanied by 1 μM Shikonin treatment for 12 h after IP with anti-SNAP-23. F 3D quantification of CD63+ MVBs in Shikonin-treated HeLa and HeLa CORO1A-/- cells in the presence of MLN4924 by z-stack confocal microscope; CD63 spot numbers and particle size were statistically analyzed (n = 5). G Western blotting analysis of VAMP-7 and STX-12 in Shikonin-treated HeLa and HeLa CORO1A-/- cells in the presence of MLN4924 after IP with anti-SNAP-23. H TIRF microscopy analysis of the CD63+ MVB distribution in Shikonin-treated HeLa and HeLa CORO1A-/- cells in the presence of MLN4924; CD63 spot numbers in the subplasmalemmal region were statistically analyzed (n = 6–10). Scale bar, 10 μm. I Fusion activity analysis of the CD63+ MVB distribution in Shikonin-treated HeLa and HeLa CORO1A-/- cells in the presence of MLN4924 (n = 9). Representative results from three independent experiments are shown. ns not significant; *P < 0.05; ***P < 0.001 (unpaired two-tailed Student’s t test; mean ± s.d.). See Supplementary Data 1 for the exact P-values. PKM2 dictates Coro1a-induced EV biogenesis Subsequently, we investigated the functional relevance of PKM2 to non-neddylated Coro1a-mediated regulation of EV biogenesis. PKM2 inhibitor or silencing eliminated Coro1a overexpression-mediated increase in CD81+CD63+ EV release from MLN4924-treated HeLa cells (Fig. 4A, B). Coro1a overexpression also could not increase CD81+CD63+ EV release from NEDD8 knockout HeLa (HeLa NEDD8-/-) cells in the presence of PKM2 inhibitor (Supplementary Fig. 4A, B). A similar result was observed in MLN4924-treated 293 T cells along with Shikonin treatment (Supplementary Fig. 4C). On the contrary, Shikonin negated Coro1a knockout-mediated decrease in CD81+CD63+ EV release from MLN4924-treated HeLa cells (Supplementary Fig. 4D). Then, we examined the role of non-neddylated Coro1a in regulating EV production in vivo, we established murine subcutaneous HeLa tumors and B16F10 melanoma with or without Coro1a knockout (Supplementary Fig. 4E) and simultaneously treated these mice with MLN4924. We found that EV amounts in TTs from both Coro1a knockout tumor-bearing mice were significantly reduced relative to mice inoculated with wild-type tumors. However, these differences were abolished in both tumor-bearing mice receiving Shikonin treatment (Fig. 4C, D). Shikonin also abrogated the decrease in TT EV amount of B16F10 Coro1a-/-; Nedd8-/- tumor-bearing mice relative to B16F10 Nedd8-/- tumor-bearing mice (Supplementary Fig. 4F, G). Others and our publications have demonstrated that robust antitumor immunity can be induced by reducing tumor EV production, thereby inhibiting tumor growth10,13. Consistent with the above results, the B16F10 tumors exhibited obvious growth priority relative to B16F10 Coro1a-/- tumors in immune complete mice, while both tumors showed similar development in Shikonin-treated mice (Fig. 4E). Similar results were obtained in B16F10 Nedd8-/- and B16F10 Coro1a-/-; Nedd8-/- tumor-bearing mice with Shikonin treatment (Supplementary Fig. 4H). In addition, when EVs from B16F10 cells (B16F10-EVs) were supplied to B16F10 Coro1a-/- tumor-bearing mice, comparable tumor growth was observed between MLN4924-treated B16F10 and B16F10 Coro1a-/- tumor-bearing mice (Fig. 4F), suggesting that Coro1a deficiency may affect tumor growth by influencing the production of TEV subpopulations. Thus, these results suggest that Coro1a regulates EV biogenesis by affecting PKM2 activation.Fig. 4PKM2 dictates Coro1a-induced EV biogenesis.HeLa cells ± Coro1a overexpression were treated with Shikonin (A) or transfected with PKM2 siRNA (B). Then, the cells were treated with 100 nM MLN4924 for 12 h, and the EVs in the supernatants were captured with anti‑CD63‑coated beads, followed by flow cytometric analysis of CD81+CD63+ EVs (n = 3), and nano-flow cytometry (A) was employed to measure the EV concentration (n = 3). Nude mice were inoculated with HeLa and HeLa CORO1A-/- cells for 3 weeks (C) or B16F10 and B16F10 Coro1a-/- cells for 2 weeks (D). TTs were collected, and EV amounts were quantified (n = 3). E After being inoculated with B16F10 and B16F10 Coro1a-/- cells for 7 days, the tumor-bearing mice were intratumorally injected with DMSO or Shikonin (2 mg kg−1 per injectin per 2 days). The tumor mice also received an intratumoral injection with MLN4924 (10 mg kg−1 per injectin per 2 days). The tumor size was monitored every other day (n = 5). F After being inoculated with B16F10 and B16F10 Coro1a-/- cells for 7 days, the tumor-bearing mice received an intratumoral injection with MLN4924 (10 mg kg−1 per injectin per 2 days). Some B16F10 Coro1a-/- tumor-bearing mice were intravenously supplied with 20 μg of B16F10-EVs 3 times per week for 2 weeks. The tumor size was monitored every other day (n = 5). Representative results from two independent experiments are shown. ns, not significant; **P < 0.01 and ***P < 0.001 (Unpaired two-tailed Student’s t test in A, B; One-way ANOVA followed by Turkey test in C–F; mean ± s.d.). See Supplementary Data 1 for the exact P-values. PKM2 dictates Coro1a-induced EV biogenesis. HeLa cells ± Coro1a overexpression were treated with Shikonin (A) or transfected with PKM2 siRNA (B). Then, the cells were treated with 100 nM MLN4924 for 12 h, and the EVs in the supernatants were captured with anti‑CD63‑coated beads, followed by flow cytometric analysis of CD81+CD63+ EVs (n = 3), and nano-flow cytometry (A) was employed to measure the EV concentration (n = 3). Nude mice were inoculated with HeLa and HeLa CORO1A-/- cells for 3 weeks (C) or B16F10 and B16F10 Coro1a-/- cells for 2 weeks (D). TTs were collected, and EV amounts were quantified (n = 3). E After being inoculated with B16F10 and B16F10 Coro1a-/- cells for 7 days, the tumor-bearing mice were intratumorally injected with DMSO or Shikonin (2 mg kg−1 per injectin per 2 days). The tumor mice also received an intratumoral injection with MLN4924 (10 mg kg−1 per injectin per 2 days). The tumor size was monitored every other day (n = 5). F After being inoculated with B16F10 and B16F10 Coro1a-/- cells for 7 days, the tumor-bearing mice received an intratumoral injection with MLN4924 (10 mg kg−1 per injectin per 2 days). Some B16F10 Coro1a-/- tumor-bearing mice were intravenously supplied with 20 μg of B16F10-EVs 3 times per week for 2 weeks. The tumor size was monitored every other day (n = 5). Representative results from two independent experiments are shown. ns, not significant; **P < 0.01 and ***P < 0.001 (Unpaired two-tailed Student’s t test in A, B; One-way ANOVA followed by Turkey test in C–F; mean ± s.d.). See Supplementary Data 1 for the exact P-values. Coro1a strengthens the interaction between PKM2 and SNAP-23 Next, we wondered how Coro1a regulates PKM2-mediated SNAP-23 phosphorylation. First, we found that Coro1a interacted with PKM2 rather than SNAP23 in 293 T cells independently of neddylation (Fig. 5A). PKM2 is particularly likely to form a dimer when tyrosine 105 (Y105) is phosphorylated31. Then, dimerized PKM2 is relocated to the vesicular structures close to the PMs, associating with SNAP-2318. We found that Coro1a induced PKM2 phosphorylation at the Y105 site in the absence of Coro1a neddylation (Fig. 5B). Consistent with this result, Coro1a increased exogenous and endogenous interaction between PKM2 and SNAP-23 (Fig. 5C, D). In contrast, PKM2 and SNAP-23 interaction was inhibited in HeLa CORO1A-/- cells (Fig. 5E). Therefore, Coro1a increases SNAP-23 and PKM2 interaction by promoting PKM2 phosphorylation.Fig. 5Coro1a strengthens the interaction between PKM2 and SNAP-23.A Western blotting analysis of PKM2 and SNAP-23 in the lysates of MLN4924-treated 293 T cells ± Coro1a-Flag overexpression after IP with anti-Flag. B Western blotting analysis of Y105 phosphorylation of PKM2 in MLN4924-treated 293 T cells ± Coro1a overexpression. C Western blotting analysis of Myc in the lysates of MLN4924-treated 293 T cells ± Coro1a-Flag overexpression, accompanied by PKM2-His and SNAP-23-Myc overexpression after IP with anti-His. D Western blotting analysis of PKM2 in the MLN4924-treated 293 T cell lysates ± Coro1a-Flag overexpression after IP with anti-SNAP-23. E Western blotting analysis of PKM2 in the lysates of MLN4924-treated HeLa and HeLa CORO1A-/- cells after IP with anti-SNAP-23. Representative results from three independent experiments are shown. See Supplementary Data 1 for the exact P-values. Coro1a strengthens the interaction between PKM2 and SNAP-23. A Western blotting analysis of PKM2 and SNAP-23 in the lysates of MLN4924-treated 293 T cells ± Coro1a-Flag overexpression after IP with anti-Flag. B Western blotting analysis of Y105 phosphorylation of PKM2 in MLN4924-treated 293 T cells ± Coro1a overexpression. C Western blotting analysis of Myc in the lysates of MLN4924-treated 293 T cells ± Coro1a-Flag overexpression, accompanied by PKM2-His and SNAP-23-Myc overexpression after IP with anti-His. D Western blotting analysis of PKM2 in the MLN4924-treated 293 T cell lysates ± Coro1a-Flag overexpression after IP with anti-SNAP-23. E Western blotting analysis of PKM2 in the lysates of MLN4924-treated HeLa and HeLa CORO1A-/- cells after IP with anti-SNAP-23. Representative results from three independent experiments are shown. See Supplementary Data 1 for the exact P-values. Advanced tumors increase EV release by upregulating Coro1a Tumors secrete more EVs as progression32. We found sequentially increased Coro1a in TTs of 2-, 3- and 4-week tumor mice bearing B16F10 tumors and increased EVs from these TTs (Fig. 6A and Supplementary Fig. 5A). The increase in Coro1a of these TTs was further confirmed by immunohistochemistry (IHC) (Fig. 6B). In addition, increased extent of EVs in TTs were greatly blunted with Coro1a deficiency (Fig. 6C), suggesting tumors secret more EVs as progression by upregulating Coro1a.Fig. 6Advanced tumors increase EV release by upregulating Coro1a.Coro1a proteins in TTs from mice bearing B16F10 tumors for 2, 3 and 4 weeks were detected by western blotting (A) and IHC (B, n = 15). C Nude mice were inoculated with HeLa and HeLa CORO1A-/- cells for 2 or 4 weeks. TTs were collected, and EV amounts were quantified (n = 3). D, E Coro1a proteins in TTs from lung cancer patients were detected by IHC. Coro1a protein levels in patients at different stages were statistically analyzed (D, n = 4 - 30), and the PFS of Coro1ahi and Coro1alo patients (E). Scale bar, 50 μm. ns not significant; *P < 0.05 (unpaired two-tailed Student’s t test in C; Log-rank test in D; mean ± s.d.). See Supplementary Data 1 for the exact P-values. Advanced tumors increase EV release by upregulating Coro1a. Coro1a proteins in TTs from mice bearing B16F10 tumors for 2, 3 and 4 weeks were detected by western blotting (A) and IHC (B, n = 15). C Nude mice were inoculated with HeLa and HeLa CORO1A-/- cells for 2 or 4 weeks. TTs were collected, and EV amounts were quantified (n = 3). D, E Coro1a proteins in TTs from lung cancer patients were detected by IHC. Coro1a protein levels in patients at different stages were statistically analyzed (D, n = 4 - 30), and the PFS of Coro1ahi and Coro1alo patients (E). Scale bar, 50 μm. ns not significant; *P < 0.05 (unpaired two-tailed Student’s t test in C; Log-rank test in D; mean ± s.d.). See Supplementary Data 1 for the exact P-values. As an extension of our findings in human cancers, we observed significantly higher Coro1a in cancerous (Ca) tissues of non-small cell lung cancer (NSCLC) patients than in precancerous (Para-Ca) tissues (Supplementary Fig. 5B and Supplementary Table 1). Compared with Stage I patients, Coro1a protein levels in TTs of Stage II and III patients were significantly higher. However, this increase was not obtained in Stage IV patients, probably due to the small sample size (Fig. 6D). We previously demonstrated that Coro1a deficiency inhibited mouse MC38 colon tumor growth and stimulated antitumor CD8+ T cell responses, which were abolished by supplementation of TEVs, indicating Coro1a accelerates tumor growth by suppressing antitumor immunity through inducing TEV production13. Consistent with the oncogenesis role of TEVs and the regulatory effect of Coro1a on EVs, we found that patients with low Coro1a protein levels (Coro1alo) had a favorable progress-free survival (PFS) relative to patients with high Coro1a protein levels (Coro1ahi) (Fig. 6E). These results suggest that increased TEVs are produced in a Coro1a-dependent manner, which may facilitates tumor progression. Discussion While mounting evidence underscores the critical involvement of EVs in disease progression, their functional outcomes are highly context-dependent, varying with cellular origin, cargo composition, and the surrounding microenvironment. Consequently, although certain pathogenic EV subpopulations present attractive therapeutic targets, any intervention strategy must acknowledge the potential duality of EV functions. Although several approaches have been developed to reduce EV production, they remain unsuitable for clinical translation33. A deeper understanding of EV biogenesis mechanisms is therefore imperative for the identification of specific and druggable targets suitable for clinical translation. Previous work demonstrated that neddylated Coro1a localizes to MVB membranes, where it directly binds and recruits Rab7, thereby directing MVBs toward lysosomal degradation and suppressing EV release. Nevertheless, we observed that Coro1a knockout still suppressed EV secretion even in the presence of MLN4924 neddylation inhibitor, suggesting an alternative, neddylation-independent regulatory role for Coro1a in EV biogenesis13. In this study, we uncover that non-neddylated Coro1a promotes EV secretion by facilitating MVB-PM fusion. Mechanistically, Coro1a activates PKM2, promotes PKM2-mediated phosphorylation of SNAP-23, and drives the assembly of the STX12-SNAP23-VAMP7 SNARE complex, which is essential for the tethering and fusion of MVBs with the PM to enable EV release. To directly validate the necessity of STX-12 in mediating MVB-PM fusion, we performed functional knockdown and rescue experiments for STX-12. The results show that specific siRNA-mediated knockdown of STX-12 significantly suppressed the Coro1a overexpression-induced increase in EV secretion. This inhibitory effect was fully rescued by STX-12 overexpression. This “knockdown-rescue” paradigm establishes the ‌causal role‌ of STX-12 downstream of the Coro1a-PKM2 signaling axis in driving SNARE complex assembly and MVB-PM fusion. In addition, PKM2 inhibitor Shikonin or PKM2 silencing treatment blocked Coro1a-induced SNAP-23 phosphorylation and STX-12-SNAP-23-VAMP-7 complex formation. This phenotype is entirely consistent with that observed in the direct STX-12 knockdown experiments. Together with the in vitro liposome fusion data, these results ‌collectively construct a comprehensive evidence chain that Coro1a, by activating PKM2, promotes SNAP-23 phosphorylation, enhances its interaction with STX-12 and VAMP-7, thereby specifically assembling the fusion-competent STX-12-SNAP-23-VAMP-7 SNARE complex that ultimately executes the MVB-PM fusion event. Altogether, Coro1a exerts a dual regulatory influence on EV biogenesis, operating through neddylation-dependent and independent pathways. Notably, previous data indicated that even in the absence of MLN4924, Coro1a overexpression still enhances EV production13, indicating that the non-neddylated form predominates under physiological conditions, whereas the inhibitory neddylated form is amplified under external stress. Collectively, these findings position Coro1a as a central molecular regulator enabling context-specific EV release through intricate dual-pathway control. Based on the established role of STX12 in SNARE complexes facilitating MVB-PM fusion, it is hypothesized that STX12 would localize to the PM. Although our immunofluorescence analysis suggests a potential enrichment of STX12 in proximity to the PM region, we acknowledge limitations in resolution that preclude definitive assignment. Alternative interpretations, including cytoplasmic retention or association with sub-plasmalemmal vesicles, remain plausible and warrant further investigation using complementary approaches such as immunoelectron microscopy. Coro1a enhances the phosphorylation of SNAP-23 through its regulation of PKM2, which exists in two forms dimer and a tetramer. The dimeric PKM2 exhibits low pyruvate kinase activity but possesses high protein Ser/Thr kinase activity34. Phosphorylation of PKM2 at the Y105 site inhibits its tetramerization and increases its protein kinase activity35. In this study, we observed that Coro1a specifically promotes the phosphorylation of PKM2 at Y105, which aligns with the observed increase in Ser/Thr phosphorylation of SNAP-23 under Coro1a overexpression. These results suggest that Coro1a reduces PKM2’s metabolic activity while augmenting its kinase activity. Notably, Coro1a lacks a kinase domain, and as a Ser/Thr kinase, PKM2 cannot auto-phosphorylate on tyrosine residues. Therefore, we hypothesize that Coro1a facilitates the interaction between PKM2 and an unidentified tyrosine kinase, leading to enhanced PKM2 phosphorylation. Subsequently, the hyperactive PKM2 promotes SNAP23 phosphorylation, thereby facilitating EV secretion. The identity of the tyrosine kinase involved in this pathway remains to be elucidated. It should be noted that the majority of mechanistic investigations in this study were conducted under MLN4924 treatment to specifically isolate neddylation-independent pathways. While this approach enabled precise dissection of Coro1a’s non-canonical function, the reliance on pharmacological perturbation presents inherent limitations in observing baseline physiological dynamics. Future studies employing real-time imaging in unperturbed model systems will be essential to fully characterize the spatiotemporal regulation of Coro1a-mediated MVB-PM fusion without potential confounders. Previous work demonstrated that Coro1a deficiency markedly delays the growth of MC38 colon tumors in mice by attenuating EV-mediated immunosuppression in the tumor microenvironment13. Here, we observe a progressive increase in Coro1a expression during tumor development, accompanied by elevated EV secretion in advanced-stage tumors. In addition, clinical analysis revealed that patients with high tumor Coro1a protein levels exhibited significantly shorter progression-free survival (PFS). Given that TEVs are potently immunosuppressive and contribute significantly to tumor malignancy10,36, the observed correlation between elevated Coro1a and poor PFS may be attributed to enhanced production of immunosuppressive TEVs. However, it is critical to note that the correlation between EV abundance and clinical outcomes is complex and likely influenced by multiple factors beyond simple quantitative increases. While this study establishes a clear association between Coro1a expression and tumor progression, the upstream factor(s) driving Coro1a upregulation during tumorigenesis remain unclear. Further elucidation of this regulatory mechanism would be valuable, as it could uncover promising therapeutic targets for intervention in cancer progression. In summary, this study elucidates a neddylation-independent pathway through which Coro1a regulates EV biogenesis, revealing how non-neddylated Coro1a promotes MVB-PM fusion. These findings not only extend our understanding of the dual regulatory roles of Coro1a in EV production but may also inform the future development of targeted strategies aimed at mitigating the impact of pathogenic EVs. Materials and methods Human samples Tumor tissues from lung cancer patients were obtained from the Second Affiliated Hospital, Zhejiang University School of Medicine, with approval from the institutional Ethics Committee. All participants were informed about the use of their samples, and written informed consent was obtained. All ethical regulations relevant to human research participants were followed. Mice Female mice of the C57BL/6 J and nude mice, aged between 6 and 8 weeks and weighing 18–20 g, were acquired from Joint Ventures Sipper BK Experimental Animal Co. in Shanghai, China. All the mice were housed in a controlled facility that maintained specific pathogen-free conditions. The experimental procedures on the mice received the necessary approval from the Animal Care and Use Committee of the Zhejiang University School of Medicine. Cell lines and cell culture Human 293 T embryonic kidney cells, human HeLa cervical cancer cells, and mouse B16F10 melanoma cells were purchased from the Chinese Academy of Sciences Institute (Shanghai, China). The cells were cultured in DMEM with 10% exosome-depleted fetal bovine serum (FBS) (Thermo Fisher Scientific, Waltham, CA, USA) and 1% penicillin/streptomycin (Keyi, Hangzhou, Zhejiang, China). All cells were cultured at 37 °C in a humidified atmosphere with 5% CO2. Plasmid and DNA, siRNA transfection Flag-human-CORO1A, Flag-human-PKM2, Flag-human-VAMP7, Myc-human-SNAP23, His-human-PKM2, and SNAP-23 mutants, all driven by the CMV promoter in pLV3 vectors, were constructed and acquired from the Miao Ling Plasmid Sharing Platform (Wuhan, Hubei, China). Rab5Q79L-GFP was provided by Prof. Yuehai Ke (Zhejiang University, Hangzhou, Zhejiang, China). According to the manufacturer’s protocol, 293 T, HeLa, and B16F10 cells were transfected with 2 μg (10 μg in rescue experiments) plasmids using JetPEI Transfection Reagent (Polyplus, Beijing, China) or PEI (Polysciences, Warrington, PA, USA) or were transfected with 10 nM scramble negative control (NC) or target siRNA using INTERFERin (Polyplus). The sequences of the siRNAs used in this study are listed in Supplementary Table 2. Semiquantitative EV assay To detect EVs in the cell culture supernatants, supernatants containing FBS with depleted EVs were cleared by centrifugation at 300 × g for 10 min and 2000 × g for 20 min. 4-μm aldehyde sulfate beads (Thermo Fisher Scientific) were first coated with purified anti-CD63 antibodies and then blocked with FBS with depleted EVs at room temperature (RT) for 1 h. The beads were washed twice in PBS and centrifuged at 3000 × g for 5 min. The cleared supernatants (1 ml) were incubated with anti-CD63-coupled beads overnight at 4 °C with shaking. The beads were washed twice in PBS and incubated with anti-CD9 or anti-CD81 antibodies for 30 min at 4 °C. After washing twice in PBS, the beads were acquired on an ACEA NovoCyte, and the data were analyzed with NovoExpress software (ACEA Biosciences, San Diego, CA, USA). The negative staining threshold was obtained from beads incubated with an unconditioned medium. The antibodies used are listed in Supplementary Table 2. Nano-flow cytometry EV assay EVs isolated from cell culture media, sera, or TTs were resuspended in PBS at a concentration of 2 × 109 particles ml−1. For nano-flow cytometry analysis, 50 µl aliquots of the EV suspension were directly loaded into the Flow NanoAnalyzer (N30E, NanoFCM Inc., Xiamen, Fujian, China) without any antibody staining. Particle concentration and size distribution were quantified based solely on light scattering signals using manufacturer-calibrated protocols. EV isolation procedures Cells were seeded onto 15-cm plates at a density of 3 million cells per plate. These cells were cultured for 48 h, and then the media from 10 plates were collected, followed by differential centrifugation. Firstly, the supernatants were centrifuged at 300 × g for 10 min, followed by a centrifugation step at 2000 × g for 20 min, both carried out at 4 °C. Subsequently, another centrifugation was performed at 10,000 × g for 30 min at 4 °C. To isolate EVs from tumor tissues (TTs), TTs were detached and ground in 5 ml PBS buffer with a syringe plunger. Then type IV collagenase (Sigma–Aldrich, St. Louis, MO, USA) was added into this buffer at a final concentration of 1 mg ml-1 and enzymatically digested for 1 h at 37 °C. The TT fragments containing supernatants were centrifuged at 300 × g for 10 min. Then, all the resulting supernatants were filtered through 0.22 μm syringe filters (Millipore, Billerica, MA, USA) and collected in 35 ml ultracentrifuge tubes (Beckman Coulter, Brea, CA, USA). EVs were concentrated by ultracentrifugation using an SW32Ti rotor (L-90K with SW32Ti rotor, Beckman Coulter) at 100,000 × g for 70 min at 4 °C. Subsequently, the EV pellets were resuspended in sterile PBS. The protein content of the isolated EVs was quantified using the BCA protein assay without adding any detergent (Thermo Fisher Scientific). Transmission electron microscopy (TEM) and nanoparticle tracking analysis (NTA) EV suspension was placed on 200-mesh carbon-coated copper grids at RT for 2 min. Any excess suspension was carefully removed from the grids using filter paper. Subsequently, the EVs were subjected to negative staining by uranyl acetate, performed at RT for 5 min. After the staining step, the EVs were washed twice with PBS and left to dry. The dried EVs were then observed using an FEI Tecnai T10 electron microscope operating at 100 kV (Thermo FEI, Hillsboro, OR, USA). To measure particle sizes and concentrations, NTA analyzed EVs using a NanoSight NS300 system (Malvern PANalytical, Malvern, UK) configured with a 488 nm laser and high-sensitivity sCMOS camera. Immunoprecipitation and western blotting EVs and cells were lysed in the SDS buffer. The lysates were then boiled at 100 °C for 10 min. Following the lysis and boiling step, the samples were resolved using SDS-polyacrylamide gel and transferred onto PVDF membranes (Millipore). These membranes were blocked with 5% milk for 2 h. Next, the membranes were incubated overnight with the appropriate primary antibodies at 4 °C, followed by incubation with the corresponding secondary antibodies at RT for 2 h. An Enhanced Chemiluminescence Kit (MultiSciences, Hangzhou, Zhejiang, China) was employed to detect the protein bands. For immunoprecipitation (IP), the cells were lysed in co-immunoprecipitation lysis buffer consisting of 50 mM Tris-HCl, 5 mM EDTA, 150 mM NaCl, 0.5% (v/v) NP-40, and 10% (v/v) glycerol at pH 7.4. The lysis buffer was supplemented with 1 mM PMSF, 1 mM Na3VO4, and 10 mM NaF. The lysate was then incubated with M2-Flag beads (Sigma–Aldrich) overnight at 4 °C for IP. The immunoprecipitates were washed at least three times with lysis buffer and subsequently analyzed using the indicated antibodies by western blotting. For the detection of SNAP-23 phosphorylation, SNAP-23 IP was performed under denaturing conditions. Briefly, 293 T or HeLa cells were lysed and boiled for 5 min in IP buffer supplemented with 1% SDS. The lysates were then diluted with IP buffer containing 0.1% SDS. After centrifugation at 12,000 × g for 10 min at 4 °C, the supernatants were collected and incubated with protein A/G magnetic beads pre-bound to the anti-SNAP-23. Following overnight incubation, the beads were washed three times with IP buffer, and the immunoprecipitates were eluted by boiling in 1% (w/v) SDS sample buffer. The Abs used are listed in Supplementary Table 2. Scanned images of uncropped blots scans are presented in Supplementary Data 1. Iodixanol density gradient fractionation Iodixanol density gradient medium (StemCell, Vancouver, BC, Canada) was prepared in ice-cold PBS immediately before use to generate a discontinuous step gradient (12–28%). The 28% iodixanol solution was added to the bottom of a centrifugation tube, and iodixanol at decreasing concentrations in PBS was carefully layered on top of the iodixanol solution, yielding the complete gradient. Identical gradients without samples were generated and ultracentrifuged in the same way for later determination of fraction densities. The density of fractions was measured by refractometry. Crude EV pellets resuspended in ice-cold PBS were added to the top of the centrifugation tube. The gradient was subjected to ultracentrifugation at 120,000 × g for 16 h at 4 °C using Beckman XPN-100 with an SW 41 Ti swinging bucket rotor. Nine individual 1-ml fractions were collected from the top of the gradient. For western blotting analysis, each 1 ml fraction was transferred to a new ultracentrifugation tube, diluted in PBS, and subjected to ultracentrifugation at 120,000 × g for 4 h at 4 °C using an SW 32 Ti swinging bucket rotor. The resulting pellets were lysed in 5 × SDS buffer on ice and boiled for 10 min at 100 °C. Immunofluorescence The cells were fixed with precooled methyl alcohol for 10 min at −20 °C and then permeabilized with 0.1% Triton X-100 for 10 min at RT. Non-specific binding sites were blocked by incubating the cells with a blocking buffer containing 5% bovine serum albumin and 3% goat serum in PBS. Then, the cells were incubated with the corresponding primary antibodies overnight at 4 °C in the blocking buffer. After 3 washes in PBS the following day, the cells were incubated with fluorophore-conjugated secondary antibodies for 1 h at RT and washed in PBS. For plasma membrane labeling, cells were incubated with 25 μM DiI (Thermo Fisher Scientific) at room temperature for 30 min, followed by three washes with PBS to remove excess dye. Nuclei were stained with DAPI (Thermo Fisher Scientific). The stained sections were imaged using an Olympus IX83-FV3000 confocal microscope (Olympus Corp, Tokyo, Japan). Images were analyzed with ImageJ software (NIH, Bethesda, MD, USA). For protein co-localization analysis, PCC was used to quantify the overlap between two proteins, whereas fluorescence intensity profile analysis was performed to assess the spatial overlap among three proteins. The antibodies used are listed in Supplementary Table 2. CRISPR-Cas9-mediated construction of gene knockout cells The gRNAs, with sequences for their target genes listed in Supplementary Table 2, were annealed and cloned into the Lenti-CRISPR-v2 vector. To delete target genes, HeLa and B16F10 were transiently transfected with Lenti-CRISPR-v2 plasmid carrying target gRNAs and selected with 2.5 µg ml−1 puromycin for 3 days. Cells were then transferred into fresh medium without puromycin and seeded at super-low density to allow colony formation from single cells. Colonies were then picked and expanded for knockout validation by immunoblot. TIRF microscopy HeLa cells were transfected with the GFP-CD63-His or mCherry-CD63-HA plasmid for 36 h, and cells were incubated in PBS before analysis by TIRF microscopy. For TIRF microscopy with an Olympus IX83 microscope (Olympus), the penetration depth (δ) of the evanescent field used to excite the fluorophore was set to 150 nm. Frames were acquired at 10 Hz in stacks of 400 images with an exposure time of 100 ms. Fluorescence image acquisitions were collected with CellSens software (Olympus). GFP+ or mCherry+ CD63 vesicles were quantified using ImageJ software (NIH). To analyze vesicle motion, time-series images were captured with a Nikon N-STORM & A1 Cell TIRF system using a DU897 EMCCD 100× oil TIRF objective, and the fluorescence image acquisitions were collected with Nis-Elements software (Nikon, Tokyo, Japan). Imaris 9.5 software (Bitplane AG, Zurich, Switzerland) calculated the mean square displacement. The diffusion coefficient, the slope of the linear fit of the first 15 points of the mean square displacement curve, was then calculated as Dxy = s/4. Real-time imaging of MVBs-plasma membrane fusion Real-time imaging of MVBs and plasma membrane fusion was performed as previously described37. Briefly, HeLa cells were transfected with the CD63-pHluorin reporter for 36 h. CD63-pHluorin, a pH-responsive optical reporter, exhibits minimal fluorescence due to quenching by the low pH characteristic of the MVB lumen. When fusion occurs, the luminal pH is swiftly neutralized, causing an immediate and dramatic rise in fluorescent emission24. Cells were then imaged using a total internal reflection fluorescence (TIRF) microscope with 561-nm laser lines for imaging. The TIRF angle was optimized to visualize MVB-PM fusion events at the basal plasma membrane. For dual-color imaging, TetraSpeck microspheres were used for channel alignment. The TIRF angle was adjusted to visualize MVB-PM fusion events, focusing on the basal plasma membrane. Cells were imaged at ≥2 Hz for 3 min per field, selecting at least five fields with similar expression levels of CD63-pHluorin. Imaging was repeated for quantification of MVB-PM fusion events. Fusion events were quantified using the AMvBE macro in FIJI. Channel alignment was performed when using dual-color imaging. The AMvBE macro was used to analyze MVB–PM fusion events by tracking CD63-pHluorin fluorescence dynamics. Lipid-mixing assay The lipid-mixing assay was performed according to a previous publication with minor modifications38. Lipid powders (POPC and DOPS) were stored at −20 °C and brought to RT before use. POPC and DOPS were dissolved in chloroform at concentrations of 10 mM and 5 mM, respectively. DiD and DiI fluorescent dyes were dissolved in absolute ethanol at a 1 mM concentration. All lipid and dye solutions were prepared under an inert nitrogen atmosphere to prevent oxidation. Make T-liposomes (donor): prepared at 2.5 μM total lipid concentration with molar ratio POPC:DOPS:DiD = 78%:20%:2% — for example, 148.2 μL of POPC stock solution, 40.5 μL of DOPS stock solution, and 48.1 μL of DiD stock solution were mixed in a glass tube. Make V-liposomes (acceptor): prepared at 2.5 μM total lipid concentration with molar ratio POPC:DOPS:DiI = 78%:20%:2% — for example, 148.2 μL of POPC stock solution, 40.5 μL of DOPS stock solution, and 46.7 μL of DiI stock solution were mixed in a glass tube—using an identical film formation protocol. The lipid mixture is dried under nitrogen flow through a continuously rotating tube for at least 3 h, thereby forming a uniform film on the glass tube wall. Complete dehydration was achieved in light-protected vacuum desiccators. The dry lipid membrane was hydrated with 400 μl buffer A (50 mM Tris-Cl, pH 7.5,150 mM sodium chloride, 0.5 mM TCEP, 50 mM β-OG), and then mixed by vortex for 5 minutes until the solution became clear. Purified full-length STX-12 proteins (5 nM) were mixed with preformed lipid-detergent mixed micelles in Buffer A at a protein-to-lipid ratio of 1:500, and incubated for 20 min at RT in the dark to reconstitute the proteins into T-liposomes. The final volume of the protein-lipid-detergent mixture should be less than the maximum capacity of the desalting column. Load the protein-lipid-detergent mixtures into the Sephadex G-25 desalting column (GE Healthcare, America). The fluorescently colored fractions were collected, resulting in DiD-labeled T-liposomes harboring functional STX-12. Acceptor Proteoliposomes (V-liposomes): purified VAMP-7Δ12-3M proteins (5 nM) were incorporated using the identical methodology with DiI-containing micelles, with the protein-to-lipid ratio also set at 1:500 and incubated at RT in the dark for 20 min, yielding VAMP-7-reconstituted V-liposomes. The total lipid concentration of liposomes obtained after desalting column elution was generally 1 mM. Since proteoliposomes typically have a characteristic diameter ranging from 40 to 100 nm and should be monodispersed, dynamic light scattering (DLS) analysis was performed to verify liposome preparation quality. 100 μM T-liposome harboring STX-12 and 100 μM V-liposome harboring VAMP7 are mixed with the addition of 5 μM SNAP-23 to a final volume of 70 μl with Buffer B (50 mM Tris-Cl, pH 7.5, 150 mM NaCl, 0.5 mM TCEP) in the well of a standard 96-well plate. A negative control group without SNAP-23 protein was set up simultaneously. Liposome fusion analysis was performed using the FluoDia T70 fluorescent flat screen Reader (Protein Technologies, America) equipped with a 530/10 BP excitation filter, 580/10 BP, and 667/10 BP emission filters at 37 °C. Donor (DiI) and acceptor (DiD) fluorescence were monitored every 25 s. Monitoring the fluorescence of the donor liposome (DiI) and recipient liposome (DiD) can calculate the initial original FRET efficiency, proximity ratio, in DiD and DiI, indicating the liposome fusion signal: Proximity Ratio (EPR) = IDiD/(IDiD + IDiI). Protein purification The recombinant proteins in this study were expressed in E. coli BL21 DE3. In general, cells were cultured to an optical density 600 nm (OD 600 nm) = 0.4–0.8 and induced by 0.1 mM isopropyl-β-d-thiogalactoside at 100 rpm for 18–24 h at 18 °C. 1.2 l of cultured cells were collected by centrifugation at 7000 g for 5 min. Cell pellets were resuspended in lysis buffer (1 M Tris-Cl, pH 7.5, 2.5 M NaCl) supplied with 1 M dithiothreitol, 1 mM PMSF, and 1% Triton X-100. Resuspended cells were broken by sonication at 40 W on ice for 10 min and centrifuged at 9500 rpm for 30 min at 4 °C to separate supernatants and cell debris. 1 ml of Ni-NTA prepacked column(Sangon Biotech, China) was washed with PBS and imidazole for 10 column volumes. Then, the collected supernatant was rotated in the column and bonded overnight at 4 °C.On the second day, the PBS was first rotated and combined in the column for 10 min at 4 °C, then washed with 10-column volume PBS. Then, imidazole with different concentration gradients was washed out of the column, and effluent samples were collected. Tandem mass spectrometry mass spectrum (MS/MS) To analyze the proteins interacting with SNAP-23, ectopically expressed SNAP-23 was pulled down from 293 T cell lysates using anti-Myc-coated beads, separated by SDS-PAGE, and stained with Coomassie brilliant blue R-250 (Solarbio, Beijing, China). Then, the gel was cut into 1 cm × 0.5 cm pieces. The protein content was analyzed by MS/MS using the Q Exactive system. Tumor model and treatment On day 0, mice were subcutaneously injected with 2 × 106 tumor cells (B16F10, B16F10 Nedd8-/-, B16F10 Coro1a-/-, B16F10 Nedd8-/- Coro1a-/- cells) or 1 × 107 tumor cells (HeLa and HeLa CORO1A-/- cells). On day 7, the tumor-bearing mice were randomized into groups and received treatments as described below. The B16F10 tumor-bearing mice were intratumorally injected with DMSO or Shikonin (2 mg kg−1 per injection per 2 days) for 4 injections. The tumor mice also received an intratumoral injection with MLN4924 (2 mg kg−1 per injection per 2 days) for 4 injections. Tumor growth was monitored every 2 days by measuring the length and width of the tumors. Tumor volume was calculated using the formula: 0.5 × length × width2. Statistics and reproducibility All statistical analyses were performed using GraphPad Prism 10.4 (GraphPad Software Inc., San Diego, CA, USA), and all the data are expressed as the mean ± s.d. The normal distribution of data was validated by the Shapiro-Wilk test. Unpaired Student’s t-test was used to compare two groups, and one-way ANOVA for comparisons among multiple groups. For confocal microscopy experiments, the colocalization of CD63 with SNAP-23, and of STX-12, VAMP-7, and SNAP-23, was quantified using Pearson’s correlation coefficient. Coefficients were calculated across multiple regions of interest per cell to evaluate the extent of spatial overlap. The resulting Pearson’s coefficient values were subsequently subjected to statistical analysis as described above to determine significance. Statistical significance was considered if a P value was < 0.05. Supplementary information Transparent Peer Review file Supplementary Information Description of Additional Supplementary files Supplementary Data 1 Video S1 Hela Video S2 Hela-Coro1a Video S3 293T Video S4 293T-Coro1a Video S5 Hela-Ctrl Video S6 Hela-Coro1a-ko Video S7 Hela-DMSO Video S8 Hela-Coro1a-KO-DMSO Video S9 Hela-Shikonim Video S10 Hela-Coro1a-KO-Shikonim reporting-summary Transparent Peer Review file Supplementary Information Description of Additional Supplementary files Supplementary Data 1 Video S1 Hela Video S2 Hela-Coro1a Video S3 293T Video S4 293T-Coro1a Video S5 Hela-Ctrl Video S6 Hela-Coro1a-ko Video S7 Hela-DMSO Video S8 Hela-Coro1a-KO-DMSO Video S9 Hela-Shikonim Video S10 Hela-Coro1a-KO-Shikonim reporting-summary