ABA-Induced Cargo Proteins Loading in Extracellular Vesicles for Gene Editing

Highlights

What are the main findings?

• We developed an ABA-inducible proximity system that successfully directs theencapsulation of specific protein cargo (e.g., Cas9) into exosomes during biogenesis.
• Among the engineered scaffolds, the BASP1–PYL1 fusion proved most effective, enabling robust, ABA-dependent protein loading into EVs.

What are the implications of the main finding?

• This work establishes a versatile molecular switch for controllable loading of therapeutic proteins into exosomes.
• The technology provides a programmable platform for next-generation therapeutic delivery.

Abstract

Extracellular vesicles, which carry bioactive cargos such as proteins, RNAs, and lipids, represent promising drug delivery vehicles owing to their biocompatibility, low immunogenicity, and inherent tissue-targeting capabilities. To address the current limitations in controlled cargo loading, we developed an abscisic acid (ABA)-inducible proximity system that directs proteins into exosomes during biogenesis. We engineered exosomal scaffolds by fusing the ABA receptor PYL1 to EV-enriched proteins—including BASP1, CD9, PTGFRN, and a truncated form PTGFRNΔ687—thereby creating docking sites within the exosomal lumen, while the target cargo (e.g., EGFP, firefly luciferase, or Cas9) was tagged with the ABI1 phosphatase domain. We demonstrate that ABA administration in producer cells induces PYL1–ABI1 complex formation, which recruits ABI1-fused cargo for selective encapsulation into EVs. Among the scaffolds tested, BASP1–PYL1 proved the most effective, enabling robust, ABA-dependent enrichment of cargo proteins. Purified EVs maintained canonical morphology, size, and marker expression (CD63, syntenin-1, CD9), confirming preserved biogenesis. Critically, these loaded exosomes efficiently delivered functional cargo to recipient cells, enabling Cas9/sgRNA-mediated genome editing. Together, our findings establish an ABA-triggered molecular switch for controllable EV protein loading, providing a versatile platform for next-generation therapeutic delivery.

1. Introduction

Nanoscale extracellular vesicles (EVs) (30–150 nm), secreted by diverse cell types, have emerged as promising platforms for therapeutic delivery. This potential is rooted in their innate biocompatibility, low immunogenicity, and intrinsic ability to traverse biological barriers, including the blood–brain barrier [1]. These vesicles naturally transport functional biomolecules (proteins, RNAs, lipids) between cells, facilitating intercellular communication and modulating recipient cell behavior [2]. Critically, EVs can be engineered to deliver therapeutic cargoes such as small-molecule drugs, nucleic acids (like siRNA or miRNA), and proteins to specific tissues [3,4,5,6,7,8,9,10,11,12,13,14,15,16,17], offering significant advantages over synthetic nanoparticles in targeting efficiency and safety profiles [18,19,20].

Despite this promise, achieving precise cargo loading remains a major limitation in exosome-based therapeutics. Conventional loading strategies—including electroporation, sonication, and passive incubation—suffer from low efficiency, cargo degradation, or disruption of EV integrity [21]. Genetic modification of EV producer cells (e.g., fusion of cargo to exosome-enriched proteins like Lamp2b or CD63) enables direct encapsulation during biogenesis but fundamentally lacks tunability: once expressed, cargo loading is constitutive and irreversible. This limitation impedes applications requiring controlled release (e.g., conditional activation of gene editors) or sequential delivery of multiple payloads [21].

Chemically induced proximity (CIP) systems, which use small molecules to trigger reversible protein–protein interactions, offer a powerful strategy to achieve temporal and spatial control over biological processes [22]. Systems like rapamycin-induced dimerization (FKBP-FRB) have revolutionized synthetic biology [23,24] but face challenges in mammalian therapeutics due to potential immunogenicity, cytotoxicity, or poor pharmacokinetics. Plant hormone systems—notably the abscisic acid (ABA)-responsive PYL1-ABI1 module—provide an orthogonal alternative with low cross-reactivity in mammalian cells and favorable safety profiles [25]. ABA binding induces a rapid, reversible, and high-affinity interaction between the receptor PYL1 and the phosphatase ABI1, which is ideal for engineering inducible molecular switches [25,26].

Here, we leverage the ABA-PYL1-ABI1 module to develop a chemically regulated system for on-demand protein loading into exosomes. We chose HEK293 cells as the EV-producing platform due to their advantages in EV production [27]. We first explored several EV-enriched scaffold proteins (BASP1, CD9, and PTGFRN) [28,29] by fusing them to the PYL1 module. Concurrently, the cargo protein was tagged with the ABI1 domain. Upon ABA administration, PYL1–ABI1 complex formation drives the translocation of the ABI1-fused cargo to the BASP1-PYL1 docking sites on nascent exosomes, enabling controlled encapsulation during vesicle biogenesis. Using EGFP as the initial cargo, we found that BASP1 was the only scaffold protein that showed robust ABA-controlled encapsulation. We then engineered this induced loading system to successfully encapsulate other proteins, including firefly luciferase and Cas9 protein, and demonstrated their efficient delivery and functional activity, specifically achieving Cas9/sgRNA-mediated genome editing in recipient cells. This work establishes a versatile and controllable molecular switch for EV cargo loading, significantly advancing the field of exosome-based therapeutic delivery.

2. Materials and Methods

2.1. Plasmid Construction

The BASP1 (NM_006317.5) gene was cloned from HEK293T cell cDNA using primers 5′- GCAGGATCCGCCACCATGGGAGGCAAGCTCAGCAAG-3′ and 5′- TGCTACCGCCTCCGCCCTCTTTCACGGTTACGGTTTG-3′. Chemically synthesized, codon-optimized PYL1 sequences were amplified with primers 5′- AAAGAGGGCGGAGGCGGTAGCACACAGGACGAATTCACCCAG-3′ and 5′- GGTAGGATCCGCCTCCGTTCATAGCCTCGGT-3′. Overlap PCR was performed to fuse these two fragments. The resulting product was digested with BamHI (Takara, Cat#1010, Dalian, China) and ligated into the BamHI-treated pCS2-LbCpf1-3xHA vector [30] using T4 DNA ligase(Thermo Fisher, Cat#EL0011, Waltham, MA, USA)to generate the pCS2-BASP1-PYL1-HA plasmid. The PTGFRN (NM_020440.4) and CD9 (NM_001769.4) genes were commercially provided (Hainan Shengzhe Biotech, Haikou, China). The full coding region of PTGFRN and the truncated sequences for PTGFRNΔ687 were amplified with the forward primers 5′-TTGCAGGATCCGCCACCATGGGGCGCCTGGCCT-3′ and 5′-TGCAGGATCCGCCACCATGGGTCCTATATTTAATGCTTC-3′, respectively, and the shared reverse primer 5′- TGCTACCGCCTCCGCCGGATCCGTCCATCTCCATCGAC-3′. The CD9 coding region was amplified with forward primer 5′-TGCAGGATCCGCCACCATGCCGGTCAAAGGAG-3′ and reverse primer 5′-TGCTACCGCCTCCGCCGGATCCGACCATCTCGCGGTTC-3′. These amplified fragments were subsequently fused with the amplified pCS2 vector using primers 5′-GGTGGCGGATCCTGCAAAAAGAAC-3′ and 5′-GGCGGAGGCGGTAGCACACAG-3′ derived from the pCS2-BASP1-PYL1-HA plasmid. ABI1 coding sequences were cloned from Arabidopsis cDNA with primers 5′- GAATTCGGCAGTGGAGTGCCTTTGTATGGTTTTACT-3′ and 5′- TATAGTTCTAGAGGTCACTTCAAATCAACCACCACCAC-3′ and inserted into the EcoRI-digested pX495 vector using an in-fusion method to generate the Cas9-ABI1 expression construct. EGFP coding sequences were amplified with primers 5′- ATGACAAGCTTGTGAGCAAGGGCGAGGAGCT-3′ and 5′- CACACTTCCGCCTCCCTTGTACAGCTCGTCCATGC-3′. This EGFP PCR product was fused with the ABI1 fragment (amplified with primers 5′-GGAGGCGGAAGTGTGCCTTTGT-3′ and 5′-CCCGGGATCCTCACTTCAAATCAACCACCAC-3′) via overlap PCR. The fused EGFP-ABI1 fragment was digested with HindIII (Takara, Cat#1615, Dalian, China) and BamHI (Takara, Cat#1605, Dalian, China) and ligated into the identically restricted pCMV-3XFLAG vector. Similarly, the firefly Luciferase coding sequences (primers: 5′-TGACAAGCTTATGGAAGATGCCAAAAAC-3′ and 5′- tccCACGGCGATCTTGCCGCCCTTCTT-3′) and the ABI1 region (primers: 5′-CAAGATCGCCGTGGGAGGCGGAAGTGTGCCTTTGT-3′ and 5′-CCCGGGATCCTCACTTCAAATCAACCACCAC-3′) were fused using overlap PCR, digested with HindIII and BamHI, and ligated into the pCMV-3XFLAG vector with T4 DNA ligase (ThermoFisher, Cat#EL0011, Waltham, MA, USA). All PCR reactions were performed using Q5 High-Fidelity DNA Polymerase (NEB, Cat#M0491, Ipswich, MA, USA), and all resulting constructs were sequence-verified.

2.2. Cell Culture, Transient Transfection and ABA-Induced EV Production

Human embryonic kidney (HEK293T) (ATCC, CRL-3216), cervical carcinoma (HeLa) (ATCC, CRM-CCL-2), and hepatocellular carcinoma (HepG2) (ATCC, HB-8065) cell lines were cultured in Dulbecco’s Modified Eagle Medium (DMEM, high glucose) (ProCell, Cat#PM150210, Wuhan, China) supplemented with 10% fetal bovine serum (FBS) (Sigma, 12003C, Australia) and 1% penicillin–streptomycin (Gibco, Cat#15140122, Grand Island, NY, USA). All cell lines were maintained at 37 °C in a humidified atmosphere containing 5% CO₂. Cells were passaged at approximately 80% confluence using 0.25% trypsin-EDTA (Sigma, Cat#T4049, St. Louis, MO, USA) and plated into 12-well, 24-well, or 96-well plates at appropriate densities. Twenty-four hours post-plating, transfections were performed using Lipo8000 Transfection Reagent (Beyotime, Cat#C0533, Shanghai, China) according to the manufacturer’s protocol. Transient co-transfection was performed using a 1:1 ratio (by mass) of the Scaffold protein-PYL1 and ABI1-cargo expression plasmids. The DNA was complexed with Lipo8000 at a ratio of 1 μg DNA: 0.8 μL Lipo8000 in serum-free DMEM. Abscisic acid (ABA) induction was performed by adding 100 μM ABA (Beyotime, Cat#BS951, Shanghai, China) or vehicle (ethanol) to the culture medium of cells co-transfected with BASP1-PYL1 and the ABI1-tagged protein constructs.

2.3. Extracellular Vesicles Isolation and Purification

Extracellular vesicles were purified using two methods: PEG Precipitation for small-scale cultures [31] and Ultrafiltration for large-scale cultures according to previous reports [32]. All isolation steps were performed at 4 °C.

2.3.1. PEG Precipitation

Culture medium was collected 48 h after ABA induction. The medium was sequentially centrifuged at 4 °C: 1000× g for 10 min; 3000× g for 10 min; and 12,000× g for 10 min. Following clarification, 200 μL of freshly made 50% PEG8000 (Sigma, Cat#92897, Buchs, Switzerland) solution was added to 800 μL supernatant, mixed, and incubated at 4 °C for 2 h. Exosomes were pelleted by centrifugation at 3000× g for 10 min, and the supernatant was discarded. The pellet was resuspended in 500 μL PBS, and the washing step was repeated once. The final pellet was resuspended and lysed in RIPA lysis buffer (Beyotime, Cat#P0013B, Shanghai, China) for further analysis.

2.3.2. Ultrafiltration

For larger-scale cultures, the cell culture medium was sequentially centrifuged at 4 °C: 1000× g for 10 min; 3000× g for 10 min; and 5000× g for 20 min. The resulting supernatant was passed through a 0.22 μm filter to remove residual particulates. The filtrate was then concentrated using a 15 mL 100 kDa molecular weight cut-off (MWCO) ultrafiltration unit (Millipore, Cat#UFC910096, Darmstadt, Germany) via centrifugation at 6500× g for 1 h at 4 °C. The retentate was washed with PBS. Purified exosomes were stored at −80 °C.

2.4. Nanoparticle Tracking Analysis (NTA)

Extracellular vesicles obtained by ultrafiltration were diluted 1000-fold in sterile PBS. Particle size distribution and concentration were measured using a ZetaView PMX 110 instrument (Particle Metrix, Meerbusch, Germany).

2.5. Protein Extraction and Western Blot Analysis

Cell culture supernatants were discarded, and pre-cooled RIPA lysis buffer (Beyotime, Cat#P0013B, Shanghai, China) containing a protease inhibitor cocktail (Roche, Basel, Switzerland, 11697498001) was added. Cells were lysed on ice for 20 min. Lysates were centrifuged at 12,000× g for 15 min at 4 °C. The resulting supernatants were collected, and protein concentrations were measured using the BCA assay (Biosharp, Cat#BL521A, Hefei, China). Protein samples were then normalized to equal concentrations using RIPA buffer, mixed with 5× loading buffer, and denatured at 100 °C for 10 min.

Proteins were separated by SDS-PAGE (100 V for the stacking gel, 120 V for the separating gel) and transferred to a PVDF membrane using a standard sandwich setup at 350 mA for 90 min. Membranes were blocked with 5% non-fat milk powder in TBST for 1 h at room temperature. Primary antibodies (dilutions specified in the Supplementary Table S1) were incubated overnight at 4 °C. FLAG and HA tag antibodies were used for detecting fusion proteins respectively. After washing, HRP-conjugated secondary antibodies (1:10,000 dilution) were incubated for 1 h at room temperature. Chemiluminescent detection was performed using an ECL kit (Biosharp, Cat#BL520A, Hefei, China), mixing equal volumes of Reagent A and B, and visualized using a gel imaging system (Tanon 5200, Shanghai, China). Densitometric analysis was performed using ImageJ software (NIH, ImageJ 1.54g). The integrated densities of the FLAG-ABI1-EGFP and HA-BASP1-PYL1 bands were measured. The FLAG-ABI1-EGFP signal was then normalized to its corresponding HA-BASP1-PYL1 signal within the same sample to calculate an EGFP/BASP1 ratio in EVs. Finally, this ratio was normalized to the vehicle control within each independent experiment to determine the ABA-induced fold change. To evaluate total cargo protein packaging efficiency, the band intensity of the cargo protein was normalized to that of the EV marker syntenin-1 within the same sample. The resulting value from ABA-treated samples was subsequently normalized to the average value from vehicle-treated controls in each independent experiment.

2.6. Dual-Luciferase Reporter Assay

Cells cultured in 24-well plates were co-transfected with Renilla plasmids and the ABI1-tagged firefly luciferase expression constructs. After 48 h of treatment with vehicle or ABA, culture medium was collected, and EVs were isolated via the PEG precipitation protocol described above. The isolated EV pellets were resuspended in 100 μL PBS. Luciferase activity in the lysed extracellular vesicles was measured using the Dual-Luciferase Reporter Assay System (Kaiji, Cat#KGE3308, Nanjing, China) according to the manufacturer’s instructions.

2.7. EV Delivery and Functional Assay

HeLa and HepG2 cells were used as recipients. For EGFP delivery, 2 × 10⁵ cells/dish were seeded in confocal dishes. For Cas9 delivery, 2.5 × 10⁴ cells/well were seeded in 96-well plates. Purified exosomes were diluted in DMEM containing 1% PS and added at a standardized particle-to-cell ratio of 10,000:1. Twenty-four hours after the addition of EVs, the cell membrane was stained with PKH26 (Beyotime, Cat#C2071M, Shanghai, China) according to the manufacturer’s protocol. Briefly, cells were washed twice with serum-free medium. An equal volume of the supplied assay buffer and a 2× PKH26 staining solution were combined, added to the cells, and incubated at 37 °C for 5 min. Cells were then washed with complete medium. EGFP and PKH26 fluorescence signals were subsequently visualized and analyzed using confocal laser scanning microscopy.

EV delivery of Cas9 was confirmed by assessing the target gene modification via PCR according to a previous report [33]; briefly, a pair of sgRNAs, which were designed to remove one exon of the TERT gene, were transfected to Hela cells one day before EV delivery of Cas9 protein. Two days after Cas9 protein delivery, cells were lysed with direct PCR lysis buffer. PCR amplification was carried out to analyze the TERT gene exon deletion.

2.8. Statistical Analysis

All data were presented as the mean ± standard deviation (SD) of three independent experiments. A one-sample t-test was used to compare the ratio values of the ABA group with the baseline value of 1 (Vehicle group), with p < 0.05 considered statistically significant. A value of p < 0.05 was considered statistically significant, and p < 0.01 was considered highly statistically significant.

3. Results

3.1. Development and Screening of the ABA-Inducible EV Loading System

Based on the known orthogonal activity and safety profile of the Abscisic Acid (ABA) proximity system in mammalian cells [25,34], we designed an inducible system for controlled protein cargo loading into extracellular vesicles (Figure 1). We explored four EV-enriched proteins—BASP1, PTGFRN, a truncated PTGFRN (PTGFRNΔ687), and CD9—as candidate scaffolds. Given that BASP1 is anchored inner leaflet of cellular membranes via N-terminal myristoylation [28], CD9 is a tetra-transmembrane protein with both terminal inside vesicle lumen [29], and PTGFRN is single-transmembrane proteins with their C-termini extending into the EV lumen [28], we fused the ABA receptor PYL1 to the C-terminus of each scaffold protein to create target protein docking sites inside nascent EVs (Figure 2a). The protein cargo, EGFP, was tagged with the phosphatase domain ABI1 (ABI1-EGFP). We found the fluorescence intensity of ABI1-tagged EGFP was in comparable scale with EGFP (Supplementary Figure S1a).

Upon transient co-transfection of the EGFP-ABI1 cargo with the scaffold-PYL1 constructs, initial analysis showed that ABA treatment altered the cellular localization of EGFP fluorescence in the producer cells. This effect was particularly pronounced for the BASP1 scaffold protein, where EGFP accumulated in the cytoplasm following ABA induction (Supplementary Figure S2a,b). This redistribution may indicate an ABA-induced PYL1–ABI1 interaction and subsequent recruitment to the scaffold docking site.

To evaluate cargo packaging efficiency, we purified extracellular vesicles and performed Western blot analysis. In the absence of a PYL1-fused scaffold protein, the cargo ABI1-EGFP was rarely detected in EVs, with or without ABA induction. While all tested scaffold-PYL1 fusions facilitated ABI1-EGFP packaging, only the BASP1-PYL1 scaffold supported robust, ABA-dependent cargo enrichment in EVs (Figure 2b). Given that BASP1 is myristoylated at its N-terminus and that a short N-terminal peptide (N12) is sufficient for its EV localization [28], we investigated whether this domain alone could function as a scaffold. A BN12-PYL1 fusion protein packaged ABI1-EGFP but was less efficient than the full-length BASP1-PYL1 (Supplementary Figure S3), indicating that regions beyond the myristoylation signal contribute to optimal function. We next asked if this scaffold mechanism could be generalized. Proteins like MARCKS and MARCKSL1, which share no sequence homology with BASP1 but are similarly myristoylated and associate with the plasma membrane inner leaflet, are also enriched in EVs [28]. We hypothesized that they could function as analogous scaffolds. Indeed, C-terminal fusions of PYL1 to either MARCKS or MARCKSL1 also enabled ABA-induced cargo packaging (Supplementary Figure S4).

To ensure the ABA-induced loading system did not compromise canonical EV properties, we characterized the purified EVs loaded with cargo against passively loaded EVs. Transmission Electron Microscopy (TEM) confirmed that the EVs maintained the canonical cup-shaped morphology characteristic of exosomes (Figure 3a). Western blot analysis of key EV markers, including CD9, CD63, and Syntenin-1, showed no significant difference between the vehicle and ABA-treated groups. Critically, the negative EV marker Calnexin was not detected in EVs treated with either vehicle or ABA (Figure 3b). Furthermore, Nanoparticle Tracking Analysis (NTA) showed that EVs shared a similar size distribution with EVs, with a peak diameter typical of exosomes (Figure 3c). These data confirm that ABA efficiently loads cargo protein without significantly affecting EV diameter, morphology, or key marker expression.

3.2. Inducible Loading of Cargoes (EGFP, Luciferase and Cas9)

We selected BASP1–PYL1 as the scaffold protein for further analysis. After co-transfecting HEK293T cells with expression plasmids for BASP1–PYL1 and ABI1-EGFP, we observed that ABA treatment triggered a redistribution of fluorescence in the producer cells, resulting in accumulation beneath the plasma membrane (Figure 4a). To further analyze scaffold and cargo proteins in extracellular vesicles (EVs), we purified EVs and performed Western blotting. Interestingly, the BASP1–PYL1 fusion protein itself also accumulated in EVs following ABA induction. To quantify the active loading effect and exclude passive loading, we quantified the band intensities of EGFP and BASP1 proteins in the purified EVs using ImageJ (ImageJ 1.54g version) and calculated the ratio of cargo protein (ABI1-EGFP) to scaffold protein (BASP1-PYL1). This analysis demonstrated that the EGFP/BASP1 ratio in EVs from ABA-treated cells was approximately 2 to 3-fold higher on average than the vehicle-treated control (Figure 4b), confirming that ABA actively enhances cargo protein encapsulation via the induced proximity system. We further evaluated the overall cargo packaging efficiency of the ABA-induced system by comparing the ABI1-EGFP band intensity between ABA-treated and vehicle-treated samples, which showed an enhancement of approximately 4–6 fold (Figure 3c). Such quantitative enrichment was consistently observed across all tested cargos, including luciferase and Cas9 (Figure 5b,c and Figure 6b,c).

We next investigated the effect of ABA on the expression of the BASP1-PYL1 scaffold in the producer cells. Western blot analysis of cell lysates revealed that ABA induction surprisingly increased the level of BASP1–PYL1 protein within the producing cells (Supplementary Figure S5a). Given that the same amount of plasmid was transfected into each sample, we hypothesized that this increase might be due to altered protein turnover rather than gene expression. Quantitative real-time PCR (RT-PCR) showed no significant difference in BASP1-PYL1 mRNA levels between vehicle and ABA treatment (Supplementary Figure S5b). This suggests that the ABA-induced BASP1–PYL1 interaction with ABI1-EGFP may decrease the BASP1–PYL1 turnover rate, thereby increasing scaffold stability in the producer cells and drastically enhancing the subsequent excretion of both BASP1–PYL1 and the targeted cargo through EVs.

To confirm the versatility of the ABA system, we applied it to load other payloads: firefly luciferase and gene-editing protein Cas9. We co-transfected BASP1-PYL1 and ABI1-tagged firefly luciferase expression plasmids into producer cells; after ABA induction, we isolated EVs from the producer cell culture medium. Western blot analysis confirmed that ABA induction not only enhanced the loading of the ABI1-tagged firefly luciferase cargo but also increased the accumulation of the BASP1–PYL1 scaffold in the EVs (Figure 5a–c). We also directly measured firefly luciferase activity in isolated EVs and found that ABA induction significantly enhanced the activity within EVs (Figure 5d), consistent with the Western blot results.

EVs are emerging carriers for gene editing tools [35,36,37,38,39,40,41,42]. Since Cas9 is a large (∼160 kDa) nuclear-localized protein, we tested our system’s capacity for large cargo. We tagged Cas9 with the ABI1 domain and co-transfected it with BASP1-PYL1. Western blot analysis of purified EVs confirmed that ABA efficiently induced the loading of the large Cas9-ABI1 protein into EVs (Figure 6a–c), demonstrating that the ABA-induced system is highly effective for encapsulating target proteins even with a larger size.

3.3. Delivery of Cargo-Packed EVs to Recipient Cells

Since the in-EV luciferase cargo maintained its enzymatic activity (Figure 5d), we next assessed the delivery of the encapsulated proteins to recipient cells. We enriched cargo-loaded EVs via ultrafiltration and incubated them with recipient HepG2 cells for EGFP and HeLa cells for Cas9. Delivery of EGFP-loaded EVs resulted in green fluorescent signals in recipient cells (Figure 7). Quantification revealed that EVs from ABA-treated producer cells were delivered more efficiently than those from vehicle-treated controls, as shown by a higher percentage of EGFP-positive cells and stronger fluorescence intensity (Supplementary Figure S6a,b). For the gene editing application, we added Cas9-loaded EVs to recipient HeLa cells that were co-transfected with an sgRNA pair targeting the TERT gene. Our data demonstrated that the delivered Cas9 protein was functionally active, as evidenced by a successful PCR-amplified deletion band (Figure 8). This confirms that the ABA-induced system successfully loads functional cargo proteins that retain their activity upon delivery to target cells.

4. Discussion

Extracellular vesicles (EVs) are secreted by nearly all cell types and function as essential signal messengers through the encapsulated transfer of proteins, nucleic acids, and lipids. Their inherent characteristics—low immunogenicity, ability to cross biological barriers, and protective environment for cargo—position them as ideal natural carriers for in vivo delivery of biologics. However, the therapeutic application of EVs is hindered by their significant heterogeneity in size and molecular composition, and, critically, a poor understanding of how specific scaffold proteins facilitate cargo enrichment. Leveraging these scaffold proteins is essential for developing methods that enable precise and controllable loading of therapeutic molecules.

To address the limitations of constitutive loading, the rapamycin-induced FKBP-FRB dimerization system has been used to package target proteins into EVs, achieving an encapsulation efficiency approximately 2-fold higher than that of passive loading [6,24,37]. Building on this approach, we developed and implemented a chemically induced proximity system based on the plant hormone abscisic acid (ABA), which enables on-demand protein encapsulation into EVs. This ABA-inducible system increased loading efficiency by approximately 3- to 6-fold compared to passive loading (Figure 4d, Figure 5c and Figure 6c). ABA is a non-toxic plant hormone routinely consumed in small amounts and has recently been investigated for its potential anti-aging effects [34], making it an excellent, orthogonal small-molecule inducer for potential clinical translation. Here, ABA functions as a molecular bridge, facilitating the high-affinity interaction between the ABA receptor PYL1 and the phosphatase domain ABI1. We engineered the EV-enriched inner membrane protein BASP1 as the scaffold, fusing PYL1 to its C-terminus to create an internal docking site, while the target cargo proteins (EGFP, Luciferase, and Cas9) were tagged with the ABI1 domain. This strategy is conceptually aligned with that of Stranford et al., who employed a single transmembrane domain from PDGFR as a scaffold [43]. A key distinction lies in the functional outcome. While Stranford et al. observed that ABI1-tagged cargo enrichment was constitutive and independent of ABA in their PDGFR-based system, our more extensive investigation reveals this to be a general characteristic of transmembrane scaffolds. Specifically, we demonstrate that both single-pass (PTGFRN) and multi-pass (CD9) transmembrane domain scaffolds similarly lack ABA-inducible cargo loading. In contrast, and representing a significant advance, we identified that scaffolds based on myristoylated proteins—including BASP1, MARKS1, and MARKSL1—robustly retain ABA-dependent inducibility. This critical difference suggests that the membrane anchoring modality (lipid-based versus transmembrane) is a primary determinant for preserving controllable loading, a mechanism that warrants further investigation.

Our data explicitly demonstrate that ABA induction dramatically enhances the enrichment of ABI1-tagged cargo, and this effect is dependent on the BASP1-PYL1 docking site. More interestingly, we found that the BASP1–PYL1 scaffold protein itself was also dramatically increased in the purified EVs following ABA treatment. While both BASP1–PYL1 and the cargo protein showed a slight increase within the producer cells, their mRNA levels remained comparable with or without ABA induction (Supplementary Figure S5). This suggests a novel regulatory mechanism where the ABA-induced PYL1–ABI1 interaction alters the turnover rate or stability of the BASP1–PYL1 scaffold, thereby increasing the protein’s steady-state level and drastically enhancing its overall incorporation and co-excretion with the cargo via EVs (Figure 4b, Figure 5a and Figure 6a). Further investigation is warranted to fully elucidate the molecular mechanisms behind this ABA-induced scaffold stability and co-enrichment. Finally, we confirmed the functional utility of this platform by demonstrating the successful delivery of encapsulated therapeutic cargo to recipient cells. We successfully delivered EGFP protein to recipient HepG2 cells, observing significantly higher fluorescence signals from EVs generated under ABA induction (Figure 7 and Supplementary Figure S6a). Most importantly, we demonstrated that the system is capable of loading large, therapeutically relevant proteins, such as the gene-editing tool Cas9. The Cas9-loaded EVs were efficiently delivered to pre-transfected sgRNA HeLa cells, successfully achieving gene editing effects at the TERT gene locus (Figure 8). To better evaluate the Cas9 delivery gene editing efficiency, we delivered an sgRNA targeting PCSK9 [44] to HepG2 cells and quantified the editing efficiency through TIDE analysis [45]. Results showed successful editing at the target locus, with EVs derived from ABA-treated producer cells mediating significantly higher editing efficiency than those from vehicle-treated controls (Supplementary Figure S7a,b). In conclusion, we have successfully developed an ABA-induced cargo protein loading system for EVs, providing a robust, chemically regulated, and highly efficient platform that holds significant promise for protein biologics loading in both research and clinical therapeutic applications.

5. Conclusions

In this study, we developed a chemically inducible protein loading system for extracellular vesicles based on the abscisic acid (ABA)-responsive PYL1–ABI1 proximity module. By fusing PYL1 to the EV-enriched myristoylated protein BASP1, we established a tunable docking platform that enables selective, on-demand encapsulation of ABI1-tagged cargo proteins during EV biogenesis. This system achieves 3- to 6-fold enrichment over passive loading and supports efficient packaging of diverse functional payloads—including EGFP, luciferase, and the therapeutically relevant Cas9 nuclease—without compromising EV integrity, size, or marker expression. Importantly, EVs loaded via ABA induction successfully delivered functional proteins to recipient cells, as demonstrated by EGFP transfer and Cas9-mediated genome editing at endogenous loci. Unlike transmembrane domain-based scaffolds, the lipid-anchored BASP1–PYL1 fusion preserves strict ABA dependence, offering a key mechanistic advantage for controlled therapeutic delivery. This platform expands the synthetic biology toolkit for EV engineering and provides a robust, orthogonal strategy for biologic loading in both research and therapeutic applications.