Cell cultures
An immortalized human bone marrow-derived mesenchymal stem cell (MSC) line (3A6) was a gift from Dr. Shih-Chieh Hung of Veterans General Hospital-Taipei (Taipei, Taiwan) [32] and was characterized and used in our previous studies [33, 34]. Daoy, U87-MG, PC3, and NIH/3T3 cell lines were originally obtained from American Type Culture Collection (ATCC, Manassas, VA, USA). The ONS76 cell line was originally obtained from Japanese Collection of Research Biosources Cell Bank (NIBIOHN, Osaka, Japan). Daoy-MYCN, a highly metastatic derivative of Daoy that overexpresses MYCN, was previously constructed through the service of the National RNAi Core Facility (Academia Sinica, Taipei, Taiwan) [7]. HEK293FT and HEK293A cell lines were originally obtained from Invitrogen (ThermoFisher Scientific, Waltham, MA, USA). MSCs were cultured in low-glucose Dulbecco’s modified Eagle medium (DMEM) (Cytiva, Marlborough, MA, USA) supplemented with 10% fetal bovine serum (FBS) (Cytiva) and 1% penicillin-streptomycin (PS) (Cytiva). Daoy and U87-MG cells were cultured in Eagle’s minimum essential medium (Cytiva) supplemented with 10% FBS and 1% PS. ONS76 and PC3 cells were cultured in RPMI 1640 medium (Cytiva) supplemented with 10% FBS and 1% PS. HEK293 cells were cultured in high-glucose DMEM (Cytiva) supplemented with 10% FBS, 1% PS, and 1% non-essential amino acids (Cytiva). All cell lines were incubated at 37 °C in 5% CO2 and were routinely passaged upon < 80% confluence for the 3A6 cell line and upon ~ 90% confluence for the other cell lines.
Plasmid construction and gene manipulation
Complementary (c)DNA of human Lamp2 variant B was purchased from OriGene (cat. no. RC200456, Rockville, MD, USA). A multiple cloning site was designed near the N-terminus of the Lamp2 protein to insert a nucleotide sequence coding for the desired peptide, provided in Supplementary Table S1. Then, the entire Lamp2 with the inserted sequence was cloned into the pLAS3w.Pbsd plasmid to produce lentiviruses using a protocol of the National RNAi Core Facility (Academia Sinica, Taipei, Taiwan), HEK293FT cells, and the Maestrofectin transfection reagent (Omics Bio, Taipei, Taiwan), all according to the manufacturers’ protocol. Lentiviral transduction was performed on MSCs, followed by selection with 2.5 µg/mL blasticidin to establish stable lines.
Extracellular vesicle (EV) isolation
MSCs were cultured under normal conditions to reach around 90% confluency. Then, cells were washed with 1× phosphate-buffered saline (PBS) to remove traces of the old medium and were then replenished with serum-free medium. After 48 h, conditioned medium was collected in a 50-mL centrifugation tube and centrifuged at 3000 ×g for 20 min to pelletize dead cells. The supernatant was filtered using a 0.22-µm Millex syringe filter (cat. no. SLGVR33RS, Merck Millipore, Burlington, MA, USA) to exclude larger particles. Next, the filtered supernatant was concentrated 2–3-fold using a Vivaspin Turbo 15 5000 MWCO ultrafiltration column (cat. no. VS15T12, Sartorius, Göttingen, Germany), and then was subjected to exosome isolation using an exoEasy Maxi Kit (cat. no. 76064, Qiagen, Hilden, Germany) following the manufacturer’s instructions. In the final step, EVs were eluted in 1 mL of elution buffer per column. Finally, the eluate was buffer-exchanged with 1× PBS and concentrated to the desired volume using a Vivaspin Turbo 4 3000 MWCO ultrafiltration column (cat. no. VS04T91, Sartorius). Approximately 1011 EVs were isolated and recovered from ~ 2 × 107 MSCs cultured in four p100 dishes.
Nanoparticle tracking analysis (NTA) and zeta potential analysis
The particle size and concentration analyses of EV samples was performed with a NanoSight NS300 (Malvern Panalytical, Malvern, UK) equipped with a green laser and a CMOS camera following the manufacturer’s instructions. Before measurement, samples were diluted with 1× PBS to achieve a measured particle concentration within the 108–109 particles/mL range. Zeta potential measurement was conducted with a Zetasizer Nano ZSP (Malvern Panalytical). For each sample, data acquired from at least three repetitions were analyzed using the manufacturer’s software.
Transmission electron microscopy (TEM)
TEM was performed through the service of the Imaging Core of Taipei Medical University Core Facility Center using an HT-7700 TEM (Hitachi High-Technologies, Tokyo, Japan). A standard negative staining protocol was applied to stain the EV samples. In brief, 10 µL EVs in 1× PBS were loaded onto a TEM grid for 2 min before being mostly removed with a micropipette. Next, 10 µL of a 2% uranyl acetate solution was applied onto the grid for 2 min and gently removed. Finally, 10 µL of double-distilled (dd)H2O was applied and quickly removed. The grid was air-dried for 10 min with no treatment with glow discharge before TEM image capture. Apart from EV structures, an uneven staining distribution and artificial contrast variations of the untreated surface may be observed in the background of samples prepared by this method.
EV labeling
For in vitro assays, approximately 5 × 1010 EVs were labeled in one reaction of the ExoGlow Membrane EV Labeling Kit (cat. no. EXOGM600A-1, System Biosciences, Palo Alto, CA, USA) following the manufacturer’s protocol. As recommended by the manufacturer, the ExoQuick-TC Exosome Precipitation Solution (cat. no. EXOTC10A-1, System Biosciences) was used to ensure maximal removal of free dye, followed by low-speed centrifugation of 1000 ×g at 4 °C for 60 min to precipitate the labeled EVs, which were gently washed and resuspended in 1× PBS. All labeled EVs were subjected to an NTA to re-determine EV numbers before use in downstream assays.
EV transfection with siRNAs
The LOXL1-AS1-targeted siRNA #1196 (si1196) and the non-targeted siRNA (siNC) were purchased from GenePharma (Shanghai, China). Non-targeted fluorescein isothiocyanate (FITC)-conjugated siRNA (siNC-FITC) was purchased from Santa Cruz Biotechnology (cat. no. sc-36869, Dallas, TX, USA). Sequences of these synthesized siRNAs are provided in Supplementary Table S2. In total, 5 × 109 EVs and 2.5 µL of 10 µM siRNA were transfected per reaction using an ExoFect siRNA/miRNA Transfection Kit (cat. no. EXFT200A-1, System Biosciences) following the manufacturer’s protocol. After a 1-h incubation step, 2 µL of 1 µg/mL RNase A (cat. no. 11119915001, Roche, Basel, Switzerland) was added to the reaction for 30 min at 37 °C to degrade the untransfected siRNAs, followed by clean-up steps provided by the transfection kit to obtain a total of 200 µL transfected EVs.
Fluorescence-based quantification
Absorbance measurement of EV samples and EV-treated cells. Based on NTA results of labeled EVs, the volume of each sample was adjusted with 1× PBS to achieve an equal EV concentration between samples. Then, 100 µL of sample was loaded into a black 96-well plate followed by absorbance measurement at 465/645 nm (ExoGlow) or 488/580 nm (FITC). Next, cells were seeded in black 96-well plates at 2500 cells/well overnight. Old medium was replaced with 2.5 × 107 labeled EVs (~ 104 EVs/cell) mixed in 100 µL 1× Hank’s balanced salt solution (HBSS) and incubated for 1 h at 37 °C. Then, cells were gently washed three times with 1× PBS, followed by absorbance measured at 465/645 nm (ExoGlow). All absorbance measurements were performed with a microplate reader (CLARIOstar Plus, BMG Labtech, Ortenberg, Germany).
Flow cytometry. The E1-3 peptide (FSRPAFL) [31] and a non-cell penetrating (NC) peptide (YDEEGGGE) [35] were synthesized and conjugated with the fluorescent label N-rhodamine (cat. no. IS025, Cloud-Clone, Katy, TX, USA) (Supplementary Table S1). Cells in normal monolayer culture were gently detached using Accutase (BD Biosciences, Franklin Lakes, NJ, USA) until they had completely separated into single cells, and then were washed three times with 1× PBS. In total, 106 cells suspended in 100 µL of 1× PBS were supplemented with 1 µg peptide and incubated for 1 h at 37 °C. After being washed three times with 1× PBS to remove unbound peptides, cells were subjected to flow cytometry (BD FACSCanto II, BD Biosciences) to measure the fluorescence intensity of PE: rhodamine. Flow cytometric data were analyzed using FlowJo software vers. 10.10 (BD Biosciences).
Confocal fluorescence microscopy. Approximately 2.5 × 105 cells/well were seeded on a coverslip in a six-well plate overnight. Then, cells were treated with 2.5 × 109 labeled EVs (~ 104 EVs per seeded cell) well mixed in either 2 mL 1× HBSS for 1–3 h of incubation or 2 mL culture medium for 16 h of incubation. Then, coverslips were gently washed three times with 1× PBS, fixed with 4% paraformaldehyde in 30 min, rewashed three times with 1× PBS, and mounted on a glass slide using the ProLong Gold Antifade mountant with the DAPI DNA Stain (cat. no. P36935, Invitrogen). Images were captured with a Stellaris 8 Confocal Microscope (Leica Microsystems, Wetzlar, Germany), acquiring multiple fluorescence channels, including DAPI, green fluorescent protein (GFP), FITC, and Cy3 (ExoGlow). Fiji/ImageJ software was used to quantify the fluorescent intensity and analyze signal co-localization in three or four different captures per sample.
Cell-based functional assays
Cytotoxicity assay. EV toxicity toward cells was assessed using the WST-1 Cell Proliferation Reagent (cat. no. 11644807001, Roche) according to the manufacturer’s protocol. Briefly, cells were seeded into a 96-well plate at 1000 cells/well in 100 µL of culture medium and supplemented with different EV doses as indicated. After 48 h, 10 µL of WST-1 reagent was added to each well, followed by incubation at 37 °C for 4 h and absorbance measurement at 440/630 nm wavelengths.
EV treatment on cells. Daoy-MYCN cells were seeded at 2.5 × 105 cells per p60 dish. The next day, cells were treated with 2.5 × 109 EVs (~ 104 EVs per seeded cell) in 3 mL of fresh culture medium for 16 h. For comparison, cells with the same seeding density were directly transfected with an equal amount of siRNA using the HiPerFect transfection reagent (cat. no. 301705, Qiagen, Hilden, Germany) according to the manufacturer’s protocol. Cells were subsequently replenished with fresh medium and cultured for another 32 h (i.e., 48 h after treatment began). Then, cells were collected for a reverse-transcription qualitative polymerase chain reaction (RT-qPCR) analysis to determine the knockdown efficiency or subjected to downstream functional assays, including wound-healing, transwell, and tumor sphere-formation assays, performed as previously reported [7] and briefly described as follows.
Wound-healing assay. In total, 2.5 × 104 EV-treated cells in 100 µL of culture medium were seeded into each well of two-well culture inserts (Ibidi, Gräfelfing, Germany) to create a non-cell gap for microscopic image captures at indicated time points.
Transwell migration and invasion assays. Hanging culture inserts with 8-µm pores (Merck Millipore, Burlington, MA, USA), either non-coated for the migration assay or precoated with 50 µL of 1 mg/mL Matrigel (Life Sciences, Corning, NY, USA) for the invasion assay, were placed in a 24-well plate containing 1% FBS culture medium. Each insert was seeded with 5 × 104 EV-treated cells in serum-free medium followed by incubation for 3 h (migration) or 12 h (invasion). Migrated cells were fixed and stained with 0.5% crystal violet for quantification by cell counting using microscopic images or by absorbance measurement of crystal violet dissolved from stained cells.
Sphere-formation assay. Approximately 4 × 104 EV-treated cells were seeded at a very low seeding density of 2000 cells/mL in defined sphere-forming culture medium using an ultra-low-attachment T175 flask (day 0). At days 4, 7, and 10, 5 mL of fresh medium and 5 × 109 EVs were added to each flask. Then, the entire culture of each sample was subjected to sphere size measurement and counting, RNA extraction, and protein extraction.
Cellular RNA extraction, cDNA synthesis, and RT-qPCR analysis
Total cellular RNA was extracted with the NC RNA extraction reagent (EBL Biotechnology, Taipei, Taiwan) and then transcribed into cDNA using the PrimeScript RT reagent kit (Takara Bio, Shiga, Japan). Then, a real-time RT-qPCR was performed using gene-specific primer pairs listed in Supplementary Table S3, a QuantiNova SYBR Green PCR kit (Qiagen), and LightCycler 96 (Roche), all according to the manufacturers’ instructions. The 2− ΔΔCt method was applied to calculate gene expressions, with HSPCB as the housekeeping gene for normalization.
RT-qPCR-based quantification of siRNA copy numbers
Samples of EV were subjected to RNA extraction using a Total Exosome RNA and Protein Isolation Kit (cat. no. 4478545, Invitrogen) following the manufacturer’s protocol of small RNA enrichment. As the final step, 100 µL of nuclease-free H2O was used to elute RNA. To determine the absolute si1196 copy number in these samples, a titration of 1011 si1196 copies/µL was made, followed by 10-fold serial dilutions to make six titration samples ranging from 106 to 1011 copies/µL. Next, reverse-transcription into cDNA was performed with a designed si1196-specific stem-loop primer, followed by RT-qPCR detection with designed primer pairs (Supplementary Figure S2A and Table S3). A no-RNA control for the reverse-transcription reaction and a no-template control for the RT-qPCR were also included, and showed very high Cq values as the background signal (Supplementary Figure S2B). Titrations were used to establish a standard curve of Cq ~ si1196 copy number/µL values (Supplementary Figure S2C). The si1196 copy number was determined for each EV sample and was used to calculate the copy number per EV, the loading efficiency, and the stability of si1196 (Supplementary Figure S2D). All titration and control samples were run alongside EV samples. The entire experiment was repeated three times using independently prepared EV, titration, and control samples.
Protein extraction and Western blot analysis
Protein extraction, determination of protein concentrations, and sample blotting were performed as previously described [36]. To collect EV lysates, an EV sample was further concentrated using a Vivaspin Turbo 500 3000 MWCO PES ultrafiltration column (cat. no. VS04T91, Sartorius) until ~ 75–100 µL remained. Then, 1 volume of homemade 4× RIPA solution was added to 3 volumes of EVs to achieve a final lysate in 1× RIPA buffer. Equal amounts of protein for whole-cell lysate and EV lysate were loaded into each lane. Another sample of low-glucose DMEM + 0.01% FBS lysate was also included to assess contamination of FBS in EV samples. First antibodies, including anti-β-actin (cat. no. GTX109639, GeneTex, Irvine, CA, USA), anti-Bmi1 (D20B7, cat. no. #6964, Cell Signaling Technology, Danvers, MA, USA), anti-bovine serum albumin (BSA; cat. no. 2A3E6, Santa Cruz Biotechnology), anti-calnexin (cat. no. #2433, Cell Signaling Technology), anti-cluster of differentiation 63 (CD63; cat. no. 10628D, ThermoFisher Scientific), anti-CD81 (cat. no. sc-7637, Santa Cruz Biotechnology), anti-CD9 (cat. no. sc-59140, Santa Cruz Biotechnology), anti-CD90 (cat. no. 13801, Cell Signaling Technology), anti-lamin A/C (cat. no. 10298-1-AP, ProteinTech, Rosemont, IL, USA), anti-Lamp2 (cat. no. GTX103214, GeneTex), anti-Nanog (D73G4, cat. no. 4903, Cell Signaling Technology), anti-Oct4 (cat. no. GTX101507, GeneTex), anti-Sox2 (cat. no. GTX101507, GeneTex), anti-transforming growth factor (TGF)-β2 (cat. no. MAB612, R&D Systems, Minneapolis, MN, USA), anti-TSG101 (cat. no. 72312, Cell Signaling Technology), and anti-V5 (cat. no. A01724, GenScript, Piscataway, NJ, USA) were probed at a 1:1000 dilution overnight at 4°C. Second antibodies, including anti-mouse (cat. no. 111-035-146, Jackson ImmunoResearch, West Grove, PA, USA) and anti-rabbit (cat. no. 111-035-144, Jackson ImmunoResearch), were probed at a 1:104 dilution for 1 h at room temperature. Blots were visualized using a chemiluminescent detection kit (T-Pro Biotechnology, New Taipei City, Taiwan) and Amersham Imager 600 (Cytiva). The Fiji/ImageJ software was used to quantify blot lanes.
In situ hybridization
In situ hybridization in 1-µm-sectioned formaldehyde-fixed paraffin-embedded mouse tissues was performed using an RNAScope 2.5 HD Assay-RED kit (Advanced Cell Diagnostics, Newark, CA, USA) with designed gene-specific probes for human LOXL1-AS1, human peptidyl-prolyl cis-trans isomerase B (PPIB, as positive control RNA), and bacterial dapB (as negative control RNA) following the manufacturer’s protocol. RNA expression was determined by automated detection of chromogenic dots using Fiji/ImageJ software and was normalized to the positive control RNA signal.
Immunohistochemical (IHC) staining
IHC staining were performed using an automated staining module (BenchMark XT system, Roche/Ventana Medical Systems, Tuscon, AZ, USA). Briefly, tissue sections were deparaffinized with EZ Prep (Roche/Ventana Medical Systems) at 75 °C, heat pretreated in Cell Conditioning 1 (CC1) for antigen retrieval at 76–100 °C and then incubated with the TGF-β2 antibody (1000x, Abnova, PAB12739) for 64 min at 37 °C after inactivation of the endogenous peroxidase using UV inhibitor for 4 min at 37 °C. Slides were then incubated with the OptiView DAB IHC Detection Kit (Roche/Ventana Medical Systems). Antibody binding was detected using DAB as chromogen and slides were counterstained with hematoxylin.
Animal studies
Orthotopic xenograft mouse model. All animal experiments were conducted under rules approved by the Taipei Medical University (TMU) Ethics Committee (LAC-2021-0475). All animal care, monitoring, and experiments followed guidelines of the TMU Laboratory Animal Center, where the animal study was conducted. To establish a highly metastatic MB model, 4 × 105 luciferase-tagged Daoy-MYCN cells were inoculated into the left hemisphere of NOD/SCID mice following a previously described protocol [7]. Bioluminescence imaging (BLI) was performed on day 3 and day 7 post-inoculation to confirm the formation of xenograft tumors, and then imaging was performed weekly until the endpoint. Histological studies were performed to confirm the presence and location of xenograft tumors.
Microbubble-enhanced focused ultrasound (FUS). Microbubble-enhanced FUS was performed similarly to our previous work [37]. Briefly, anesthetized mice were stabilized in the prone position using a stereostatic instrument pre-mounted with an apparatus comprised of an ultrasound transducer and a focal cone. The cone tip was positioned to point at the site of tumor inoculation on the mouse head with a layer of ultrasound gel applied to the scalp to ensure effective transmission. Next, a 150 µL/kg dose of reconstituted microbubbles (SonoVue, cat. CE101301, Bracco Suisse, Switzerland) was mixed with the EV sample and injected into the tail vein. Then, FUS was immediately performed using a 50-MHz arbitrary waveform generator (model 645, Rohde & Schwarz, Munich, Germany) and a 10 kHz-250 MHz 75 W amplifier (model 75A250A, Amplifier Research, Souderton, PA, USA) at 220 mV amplitude, 1 MHz frequency, and 3% duty cycle (30,000 cycles/s) for 60 s.
In vivo EV treatment and EV tracking. For weekly EV treatment in mice, ~ 2 × 1010 of siRNA-loaded EVs per mouse were injected into the tail vein, followed by a FUS session. For in vivo EV tracking, ~ 1.5 × 1011 EVs per mouse were labeled with the ExoGlow Vivo EV Labeling Kit (Near IR) (cat. no. EXOGV900A-1, System Biosciences) following the manufacturer’s protocol and then were injected into the tail vein, with or without application of FUS. Fluorescence imaging was subsequently performed to capture images at excitation/emission wavelengths of 768/845 nm. In all experiments, mouse organs of interest were subsequently obtained for histological studies.
Statistical analysis
Data are presented as the mean ± standard deviation (SD) with the difference between two groups determined by a two-tailed Student’s t-test unless otherwise specified in the figure legends. Correlations between two variables were assessed using a simple linear regression and Pearson’s correlation coefficients. Survival analyses were carried out using a Kaplan-Meier curve and the Gehan-Breslow-Wilcoxon test. All statistical analyses were performed with GraphPad Prism vers. 9 (GraphPad, San Diego, CA, USA) and Microsoft Office Excel vers. 16 (Microsoft, Redmond, WA, USA). p < 0.05 was considered statistically significant.
