Engineering Extracellular Vesicles with the Tools of Enzyme Prodrug Therapy

Extracellular vesicles (EVs) have recently gained significant attention as important mediators of intercellular communication, potential drug carriers, and disease biomarkers. These natural cell-derived nanoparticles are postulated to be biocompatible, stable under physiological conditions, and to show reduced immunogenicity as compared to other synthetic nanoparticles. Although initial clinical trials are ongoing, the use of EVs for therapeutic applications may be limited due to undesired off-target activity and potential “dilution effects” upon systemic administration which may affect their ability to reach their target tissues. To fully exploit their therapeutic potential, EVs are embedded into implantable biomaterials designed to achieve local delivery of therapeutics taking advantage of enzyme prodrug therapy (EPT). In this first application of EVs for an EPT approach, EVs are used as smart carriers for stabilizing enzymes in a hydrogel for local controlled conversion of benign prodrugs to active antiinflammatory compounds. It is shown that the natural EVs’ antiinflammatory potential is comparable or superior to synthetic carriers, in particular upon repeated long-term incubations and in different macrophage models of inflammation. Moreover, density-dependent color scanning electron microscopy imaging of EVs in a hydrogel is presented herein, an impactful tool for further understanding EVs in biological settings.


Supplementary Methods
. Characterization and loading of EVs and liposomes. Figure S2. Stability of hydrogels upon storage. Figure S3. Supplementary density-dependent scanning electron microscopy imaging. Figure S4. Complementary scanning electron micrographs of control gels. Figure S5. Activity of EV and liposome loaded hydrogels. Figure S6. Activity of enzyme control hydrogels. Figure S7. Impact of EV and liposome loaded hydrogels on inflammatory markers.
TheraPEAK™ is a chemically defined serum-free medium that contains human albumin, recombinant human insulin, pasteurized human transferrin, HEPES, and L-glutamine. Cells were conditioned for 2-3 days in the medium; conditioned medium was continuously collected, centrifuged at 300 x g for 15 min and the supernatant stored at -80 °C until further use. For EV isolation and loading, our previously established protocol [1] was adapted. Briefly, 70-140 mL supernatants (corresponding to 4-16 x 10 6 hMSC cells) were centrifuged at 4000 x g for 15 min and pelleted at 120,000 x g for 2 h. Supernatants were removed and the EV pellet re-suspended in typically 400-600 µL PBS ( Figure S1a). Subsequently, EVs were mixed with β-glucuronidase (Sigma G7646, final 1.5 mg/mL w/v) and saponin (Sigma 47036, final 1 mg/mL) and incubated at room temperature for 10 min. Then loaded EVs were purified by size exclusion chromatography (SEC) using sepharose CL-2B (17 mL, eluent PBS), where EVs typically eluted between 6 and 8 mL fraction volume. Pooled fractions were centrifuged using Amicon Ultra-15 centrifugal units (molecular weight cut-off 100 kDa) for 5 min at 3000 x g to render a final volume of 500-1000 µL. EV size and concentration was measured by nanoparticle tracking analysis (LM-10, Malvern Instruments, Figure S1a). Enzymatic activity was measured by mixing 125 µL of each fraction with 25 µL of fluorescein di-β-Dglucuronide (50 mM, Thermo) and incubation at 37 °C. Fluorescein release was measured at 0 and 16 h on an EnSpire microplate reader (PerkinElmer) at 495/520 nm (excitation/emission) and the average fluorescence signal produced by 10 8 EVs calculated ( Figure S1c). Samples were stored at -80 °C until further use.
Hydrogels were mechanically tested in unconfined compression using an ElectroForce (Bose) equipped with a 22.5 N load cell. For all tests, sample dimensions were measured in wet state using digital calipers. Samples were pre-loaded to 0.05 N and compressed to 10% strain at a crosshead speed of 0.5% strain/min. The compressive modulus was calculated from the linear region of the stress-strain curve. Each experiment was repeated with n = 5.
For stability assessments, gels were incubated in PBS at 24 °C or 37 °C and their weight was monitored over 15 days. Each sample was normalized to the initial weight ( Figure S2a). Fluorescence microscopy: EVs were labeled with a green fluorescent membrane dye PKH67 using the supplier's kit (Sigma). Typically, 200 µL of EVs were diluted in 400 µL diluent C and added to a mixture of 4 µL PKH67 stock in 196 µL diluent C. After incubation for 10 min at room temperature, EVs were purified by SEC, characterized, and loaded into hydrogels as described above. Hydrogels containing PKH67-labeled EVs were imaged using an inverted confocal laser scanning microscope (Leica TCS SP5) with incubation chamber to maintain gel hydration during imaging. In general, an APO 63x/1.4NA oil immersion objective was used for bright field imaging of gels and fluorescence detection in the green channel with a 488 nm (green) excitation laser.
Assessment of TNF-α gene and protein expression: Bone marrow cells were harvested from 6-8 week old CD-1 mice (Charles River) and matured to macrophages over 7 days in culture medium containing macrophage colony stimulating factor according to a well-established Submitted to 5 protocol. [2] Bone marrow derived macrophages (BMDMs) were incubated with EV-hydrogels and liposome-hydrogels loaded with 3.5 × 10 8 β-glucuronidase-encapsulated vesicles per gel.
Bacterial lipopolysaccharide (100 ng/mL, from Escherichia coli 026:B6, Sigma L2654) and curcumin-β-D-glucuronide (0.8 µM) were added and were incubated for 24 h at 37 °C and 5% CO2. RNA was isolated from macrophages using TRIzol (Thermofisher) and converted to cDNA using a High-Capacity cDNA Reverse Transcription kit (Qiagen). TNF gene expression was measured via qPCR using TaqMan     Empty poly(vinyl alcohol) hydrogels or hydrogels containing EVs were imaged by DDC-SEM. Imaging of hydrogels revealed optically more dense uranyl-labeled EVs (EV-uranyl, 3.5 × 10 8 EVs/gel, indicated by arrows) which were not observed in control gels (non-loaded-hydrogel). Images were obtained by in-lens electron detector (a and b) and in backscattered electron mode (c and d).
They were recorded at a tilt angle of 70° and post-corrected to a tilt angle of 0° for display. The density-dependent color SEM analysis was executed by assigning the in-lens or secondary electron image to the green channel and the backscattering signal to the red channel (e and f). Figure S4. Complementary scanning electron micrographs of a) gels with unlabeled EVs or b) non-loaded gels (control gel). c) and d) PVA hydrogels containing optically more dense uranyl-labelled EVs (EV-uranyl) or heme-labelled EVs (EV-heme) as indicated by arrows. Representative images were obtained by in-lens electron detector (left column) and conventional secondary electron detector (middle column), and in backscattered electron mode (right column). Figure S5. Activity of EV and liposome loaded hydrogels. Enzymatic activity of vesicleloaded hydrogels. EV-hydrogels and liposome-hydrogels were loaded with 3.5 × 10 8 βglucuronidase-encapsulated vesicles per gel. Incubation of a) EV-encapsulated βglucuronidase in hydrogels or b) liposome-encapsulated β-glucuronidase in hydrogels with fluorescein di-β-D-glucuronide for up to 7 days (first incubation). Enzymatic cleavage was assessed by measuring increasing cumulative fluorescence produced by fluorescein. After 7 days, gels were washed thoroughly with PBS and incubated with fresh fluorescein di-β-Dglucuronide substrate to assess the enzyme activity upon long-term application (recycling). Values are represented as mean ± SD, n = 3-5, *p<0.05 vs hydrogels at 0 h (ANOVA on Ranks with Dunn's post-hoc test was performed on raw data). Normalization was executed against PBS control sample (set to 0%, not included) and the highest observed fluorescein release (100%). In a and b, 0% and 100% are equal to facilitate comparison between both cycles. Figure S6. Activity of non-encapsulated β-glucuronidase in hydrogels over time.
Incubation of hydrogels containing free β-glucuronidase (0.1 mg/mL) with fluorescein di-β-D-glucuronide for up to 7 days. Enzymatic cleavage was assessed by measuring increasing fluorescence produced by fluorescein. After 7 days, gels were washed thoroughly with PBS and incubated with fresh fluorescein di-β-D-glucuronide substrate to assess the enzyme activity upon long-term application. Values are represented as mean ± SD, n = 3-5, normalization was executed against PBS control sample (set to 0%, not included) and the highest observed fluorescein release (100%). Arrows indicate percentage loss of activity after 5 and 7 days of repeated incubation. Figure S7. Impact of EV and liposome loaded hydrogels on inflammatory markers. EVhydrogels and liposome-hydrogels were loaded with 3.5 × 10 8 β-glucuronidase-encapsulated vesicles per gel. Gels were incubated with primary bone-marrow derived murine macrophages and with LPS plus 0.8 µM curcumin-β-D-glucuronide, or PBS alone for 48 h. Gene expression and protein concentration of TNF alpha was assessed by PCR and ELISA. Values are represented as mean ± SD, n = 3, *p<0.05 vs β-glucuronidase-hydrogel + PBS (in a) (oneway ANOVA with Fisher post-hoc test). Normalization was executed against a control sample containing only LPS in cell culture medium.