Preparation of Exosomes for siRNA Delivery to Cancer Cells

Extracellular vesicles, in particular exosomes, have recently gained interest as novel drug delivery vectors due to their biological origin, abundance, and intrinsic capability in intercellular delivery of various biomolecules. This work establishes an isolation protocol to achieve high yield and high purity of exosomes for siRNA delivery. Human Embryonic Kidney cells (HEK-293 cells) are cultured in bioreactor flasks and the culture supernatant (hereon referred to as conditioned medium) is harvested on a weekly basis to allow for enrichment of HEK-293 exosomes. The conditioned medium (CM) is pre-cleared of dead cells and cellular debris by differential centrifugation and is subjected to ultracentrifugation onto a sucrose cushion followed by a washing step, to collect the exosomes. Isolated HEK-293 exosomes are characterized for yield, morphology and exosomal marker expression by nanoparticle tracking analysis, protein quantification, electron microscopy and flow cytometry, respectively. Small interfering RNA (siRNA), fluorescently labeled with Atto655, is loaded into exosomes by electroporation and excess siRNA is removed by gel filtration. Cell uptake in PANC-1 cancer cells, after 24 h incubation at 37 °C, is confirmed by flow cytometry. HEK-293 exosomes are 107.0 ± 8.2 nm in diameter. The exosome yield and particle-to-protein ratio (P:P) ratio are 6.99 ± 0.22 × 1012 particle/mL and 8.3 ± 1.7 × 1010 particle/µg, respectively. The encapsulation efficiency of siRNA in exosomes is ~ 10-20%. Forty percent of the cells show positive signals for Atto655 at 24 h post-incubation. In conclusion, exosome isolation by ultracentrifugation onto sucrose cushion offers a combination of good yield and purity. siRNA could be successfully loaded into exosomes by electroporation and subsequently delivered into cancer cells in vitro. This protocol offers a standard procedure for developing siRNA-loaded exosomes for efficient delivery to cancer cells.


Introduction
Exosomes are a subtype of extracellular vesicles (EV) ranging from 50-200 nm in diameter, secreted by various cell types such as immune cells 1,2 , cancer cells 3,4,5,6 and stem cells 7 . Exosomes have also been shown to be present in various physiological fluids 8,9,10,11 . The combination of the inherent ability of exosomes to carry various biomolecules (e.g., RNA and proteins) 12,13,14 and the effective delivery of these biomolecules into recipient cells 15,16,17 attracted interest for their potential as nano-scale drug delivery vectors. Various small molecules that serve as anti-cancer and anti-inflammatory drugs have been demonstrated to be successfully loaded into exosomes and delivered to target cells 18,19,20,21,22,23,24,25,26,27 . Interestingly, nucleic acids such as siRNA 28,29 and microRNA 30 have also been successfully loaded into exosomes via electroporation and delivered to target cells.
Recently, RNA interference (RNAi) via small interfering RNA (siRNA) has gained more interest as the preferred mechanism in gene silencing due to its high specificity, potent effect, minimal side effects and ease of siRNA synthesis 28,29 . siRNAs are double-stranded RNA molecules ranging from 19 to 25 nucleotides in length that triggers sequence-specific catalytic mRNA knockdown. Due to its large molecular weight and polyanionic nature, passive uptake of naked siRNA into cells is hindered 28,29 . It is also not possible for naked siRNA to be injected into the systemic circulation due to rapid degradation by plasma nucleases 31 . Thus, encapsulation of siRNA in a nanocarrier would aid the effective delivery and uptake of siRNA into the target cells.
Exosomes are an ideal system for siRNA encapsulation as its structure is comprised of a hollow, aqueous core enveloped by a phospholipid bilayer. Exosomes not only have good stability in the blood but also have natural targeting properties to deliver functional RNA into cells 32 . The study conducted by Alvarez-Erviti et al. successfully demonstrated effective delivery of siRNA to the brains of mice using engineered exosomes with virtually no complications 31 . It is hypothesized that exosome-based therapy is relatively safer than other therapies as exosomes do not replicate endogenously as cells would and therefore do not exhibit metastatic properties 15 .
Various methods have been reported to successfully isolate exosomes from either cell culture or physiological fluids. The most popular method uses ultracentrifugation to pellet exosomes from the starting material 31,32,33 . This method can be quite harsh on exosomes and usually coprecipitates proteins from the sample. Combining ultracentrifugation with a density-based separation such as sucrose gradients is becoming more common, to reduce protein and non-exosomal contamination in the isolated exosomes 19,34 . Size-exclusion chromatography (SEC) allows separation of exosomes from other types of extracellular vesicles (EV) by size and can also result in minimal protein contamination but is limited Copyright © 2018 Creative Commons Attribution 3.0 License December 2018 | 142 | e58814 | Page 2 of 13 by small amount of starting material it can process 35,36 . Immunoaffinity capture uses beads coated with antibodies that bind to exosomal surface proteins such as tetraspanins or other cell-specific marker that allows specific capture of exosomes rather than EVs or other proteins, as well as isolating sub-population of exosomes from whole samples, but again is limited by the amount of starting material and is costly 36,37 . Polymerbased precipitation of exosomes used to be popular too, but since it is a rather crude precipitation, it leads to a higher non-exosomal vesicle and protein contamination 38,39 . Electroporation has been reported for its inefficiency as a method to load exosomes with siRNA due to protein aggregation 15,28,31 . Transfectionbased approaches were demonstrated to have better loading efficiency and protein stability, but is undesirable due to its toxicity and side effects of transfection agents in altering cellular gene expression 28 . Thus, electroporation has been more widely used in siRNA loading into exosomes as it is a safer method. However, an optimized encapsulation method needs to be established in order to deliver adequate amounts of siRNA to the target site for a potent gene knockdown.
Here, we propose an exosome isolation protocol using density-based ultracentrifugation onto just a single 25% (w/w) sucrose cushion prepared in deuterium oxide, rather than a sucrose density gradient. This is a cost-effective method that circumvents the laborious density gradient preparation and allows processing of large volumes of starting material, yet results in intact exosomes of high yield and purity suitable for subsequent loading with siRNA. Fluorescent Atto655-conjugated non-specific siRNA was loaded into Human Embryonic Kidney cells (HEK-293 cells) derived exosomes via electroporation and delivered to human pancreatic adenocarcinoma (PANC-1) cancer cells in vitro.
1. Culture HEK-293 cells in normal medium (see Table of Materials; 5% CO 2 , 37 °C) and expand them into 4 x T75 flasks (until 90% confluent). 2. Wet the membrane of the bioreactor flask by adding 50-100 mL of normal medium in the medium reservoir of the bioreactor flask. 3. Collect all HEK-293 cells from the 4 x T75 and resuspend them in 15 mL of exosome-depleted medium (see Table of Materials). 4. Add the HEK-293 cell suspension to the cell compartment of the bioreactor flask using a 20 mL syringe connected to a blunt fill needle (see Table of Materials), with care to remove any bubble that might have formed. 5. Fill the medium reservoir of the bioreactor flask with normal medium up to 500 mL and keep the flask in the incubator (5% CO 2 , 37 °C) for a week.
1. Make 1:1,000-1:50,000 dilutions of the exosome stock in 1 mL (minimum 750 µL) volume so as to obtain 20-80 particles in the viewing frame of the NTA instrument (see Table of Materials) display. 2. Inject the diluted exosome stock into the NTA instrument sample chamber using a 1 mL syringe, and insert the temperature probe of a thermometer into the temperature probe inlet. 3. Set the NTA software (see Table of Materials) for recording as follows: 3 standard measurements, 30 s each, manual temperature option unchecked; and enter the dilution factor under the Advanced tab. 4. Set the camera level to 13 and run the capture script on the NTA software, injecting a fresh batch of sample and entering the temperature of the sample chamber when prompted after each reading. 5. Set the threshold to 4 for the subsequent analysis part, and note the average modal size and particle concentration of the exosome stock from the measurements.

Characterization of Exosome Purity by Particle:Protein Ratio Determination
1. Measure the protein content of the exosome stock by a bicinchoninic acid (BCA) protein assay kit (see Table of  2. Calculate the particle:protein ratio by dividing the exosome yield obtained earlier with the protein concentration of the exosome stock measured above.  Table of Materials) to the exosome-bead mixture to achieve a 10 mM final concentration and incubate for 15 min at RT. 3. Add 1 mL of PBS and incubate for 75 min at RT in a microcentrifuge tube with mild agitation on a rocking shaker (~150 rpm). 4. Centrifuge the suspension at 580 x g for 5 min at RT and discard the supernatant. 5. Resuspend the pellet with 1 mL of 100 mM glycine solution (see Table of Materials) and incubate for 30 min at RT. 6. Centrifuge the suspension for 5 min at 580 x g. Discard the supernatant and resuspend the pellet with 1 mL of 3% FBS/PBS (see Table of Materials). 7. Repeat this washing step and resuspend the pellet in 350 µL of 3% FBS/PBS. 8. Divide the suspension into 7 tubes, each containing 50 µL of suspension and incubate them with fluorophore-conjugated anti-CD81, anti-CD9 and anti-CD63 antibodies and their corresponding isotype controls (1:10 dilution), respectively, for 45 min at 4 °C. Keep 1 of the tubes as an unstained control but undergoing the same processing. 9. Add 1 mL of 3% FBS/PBS to each tube, centrifuge for 5 min at 580 x g and discard the supernatant. 10. Resuspend the pellet with 200-400 µL of 3% FBS/PBS and analyze the sample on the flow cytometer (see Table of Materials) under the appropriate channels.

Representative Results
The physicochemical characterization of exosomes isolated from HEK-293 cells (HEK-293 Exo) are summarized in Table 1. The size measured using nanoparticle tracking analysis (NTA) instrument was 107.0 ± 8.2 nm. Exosome yield from the HEK-293 cells, also analyzed using the NTA instrument, was 6.99 ± 0.22 x 10 12 p/mL from ~24 mL of CM (obtained from 2 rounds of harvest). Purity of the HEK-293 Exo assessed by calculating the particle-to-protein ratio (P:P) was 8.3 ± 1.7 × 10 10 p/µg. Figure 3A. Morphological analysis using transmission electron microscopy (TEM) showed the HEK-293 Exo were spherical structures slightly above 100 nm in size ( Figure 3B). This result agrees with that from NTA measurement ( Figure 3A). The isolated HEK-293 Exo were positive for CD81, CD9 and CD63, which are canonical markers used to identify vesicles as exosomes ( Figure 3C).

The size distribution of isolated HEK-293 Exo is shown in
For purification of exosomes using size exclusion chromatography (Figure 4), the percentage recovery of exosomes was calculated by dividing the total exosome particle number recovered in the 10 fractions collected (F0-F9) with the initial exosome particle number used, while the percentage recovery of siRNA was calculated by dividing the total fluorescence intensity obtained from F3, F4 and F5 with the total fluorescence intensity obtained from all 10 fractions collected. The recovery of exosome and siRNA post-purification was calculated as 75.0% and 80.4%, respectively. The encapsulation efficiency of siRNA in exosomes was ~10-20%, calculated using the siRNA standard curve established ( Figure  4C).
Qualitative analysis of in vitro uptake of exosomes loaded with the fluorescent Atto655-siRNA by flow cytometry showed that PANC-1 cells treated with siRNA-encapsulated exosomes recorded the largest shift in fluorescence signal ( Figure 5A). PANC-1 cells treated with siRNAencapsulated exosomes recorded a higher percentage of population positive for the Atto655 signal (39.4%) compared to that treated with unloaded exosomes and siRNA mixture (0.56%), which corroborated the observation above ( Figure 5B). The degree of cellular uptake of siRNA (expressed as the fold difference in mean fluorescence intensity (MFI) values from that of untreated cells) was also observed to be significantly higher in PANC-1 cells treated with siRNA-encapsulated exosomes (MFI fold difference = 5.1) compared to that treated with the exosome-siRNA  . The large medium reservoir continuously supplies nutrients to and removes wastes from the cell compartment through a 10 kDa semi-permeable membrane, allowing prolonged culture without requiring a large volume of medium to be in contact with the cells, or regular flasks changing, which can ultimately save the overall cost and labor of high-scale exosome production 40 . It was also demonstrated that the morphology, phenotype as well as the immunomodulatory functions of exosomes isolated cells long-term bioreactor flasks cultures are similar to that sourced from cells cultured in regular 75 cm 2 flasks 40 . Culture of other immortalized cell lines as exosome sources in the bioreactor flask would therefore help increase their exosome yield while maintaining their integrity and function. This form of culture is however not applicable to primary cells with limited division cycles, and those that cannot be cultured in high density.
Since harvest of the CM is done weekly, and the cells in culture were never passaged, it can be assumed that the cells in the bioreactor flask are not growing in a monolayer like the regular cell culture. They are most likely to form clusters with necrotic centers, or simply detach from the surface and die when the cells are too confluent for a monolayer. Visual inspection of the cell compartment of the bioreactor flask is not possible to confirm this assumption, but is reflected by the large number of dead cells obtained during the CM harvesting. Regular removal of poorly adherent and non-viable cells from the bioreactor flask can prevent the build-up of materials on the semi-permeable membrane that can adversely affect the exchange of gas, nutrients and waste between the cell compartment and the medium reservoir, thus allowing prolonged culture in the bioreactor flasks for >6 months 40 . In this context, this non-regularity of cell growth in the bioreactor flasks is ideal as we speculate that it mimics the actual condition of tumor growth in vivo more closely than the conventional monolayer cell culture, and it is hoped that the exosomes produced by the cancer cells in the bioreactor flask would be more similar to that secreted by tumors in vivo. This would be particularly beneficial in studies looking into the role of tumor-derived exosomes in the progression of the tumor pathology. Tumorderived exosomes have been reported to intrinsically and preferentially home to their tissue of origin 32 , therefore having exosomes produced in a system mimicking their in vivo production would also be desirable in studies looking at exploring the passive targeting ability of exosomes as drug nanocarriers.
The P:P ratio was reported as a parameter to assess the purity of isolated exosomes from contaminating proteins from the culture medium of physiological fluids from which exosomes were sourced from 41 . The P:P ratio of 8.3 ± 1.7 × 10 10 p/µg obtained in this study falls within the high purity range proposed in the study. This ratio highlights the danger of using protein concentration to express the yield or dose of exosomes isolated or used in downstream studies respectively, as this does not reflect the true amount of exosomes available in the sample given the problem of protein contamination during isolation. NTA via instruments such as NanoSight, which measures the concentration of exosomes in terms of particle number, is a more sensible and accurate way of quantifying exosomes.
Highly accurate weighing during the preparation of the 25% sucrose solution in deuterium oxide is crucial as this method is a density-based isolation. Exosomes have a rather narrow range of flotation density in sucrose solution so accurate preparation of the sucrose cushion will reduce contamination of non-exosomal vesicles such as apoptotic bodies or Golgi-derived vesicles during isolation 42 . It is advised not to keep leftover sucrose solution and using it even after one day so as to avoid risk of factors that can alter its density such as loss or addition of water in the solution by either evaporation or condensation of air in the tube. Use of a swing-out rotor is also essential during centrifugation onto the sucrose cushion to allow even migration of exosomes from the CM to the sucrose solution.
Withdrawing the sucrose solution post-centrifugation is also a delicate step, and it involves finding a compromise between maximizing the amount of exosomes recovered, and not too much that protein from culture medium is introduced to the exosome sample withdrawn. The interface between the sucrose solution and the condition medium is where proteins from the culture medium would collect post-centrifugation, and can usually be seen as a dark brown ring that sits on the interface. In our hands, withdrawing 2 mL of the sucrose cushion from the initial 3 mL added is the optimum volume that agrees with the compromise mentioned above. The volumes described in this protocol are for the specific rotors used; therefore, it is advised to optimize the volume of sucrose to be withdrawn when scaling up or down the volumes for the types of rotors available in different facilities. It is also important to avoid the area right at the center of the bottom of the tube when withdrawing the sucrose, as this is where particles of higher density than sucrose will sediment and can usually be seen as an off-white pellet.
The washing step with a relatively large amount of PBS helps to further reduce the degree of protein contamination during exosome isolation 41 . This step is also essential in removing excess sucrose from the exosomes so as to avoid osmotic damage to the exosomes themselves or the biomolecules within the exosomal lumen, as well as reducing the risk of bacterial and/or fungal growth in the exosome stock. Preparing the sucrose solution in deuterium oxide rather than water helps to reduce the amount of sucrose needed to achieve the exosome flotation density for isolation, hence reducing the risk of both osmotic damage and microbial contamination. After the first centrifugation onto the sucrose cushion, the exosome-containing sucrose layer withdrawn and added to the PBS can be stored at 4 °C and processed the following day if faced with time constraints.
To the best of our knowledge, the exosome/siRNA molar ratio is an important factor in determining the efficiency of electroporation. In this protocol, we used 1:60 as the exosome to siRNA molar ratio. As the encapsulation ability of different types of exosomes are different, we strongly suggest this to be optimized on a case-by-case basis. However, the encapsulation efficiency proposed herein can always be a parameter for selecting the optimal electroporation conditions.
In addition, aggregation of siRNA is believed to be one of the most common problem in electroporation. It is proven that electroporation can induce strong aggregation of siRNA, making it even harder to enter exosomes. siRNA aggregations are often mistakenly interpreted as encapsulation of siRNA into exosome therefore proper controls were used in this study as the formation of siRNA aggregates is unavoidable during electroporation