(499a) Purification of Exosomes Using Tangential Flow Filtration
Methods: Cell supernatant is isolated from various cell lines of interest that are known to produce exosomes, HEK293, MDA 231, and CHO. After the supernatant is collected, it undergoes two benchtop centrifuge steps that are a part of the conventional ultracentrifugation process. The first, 300g for 5 minutes removes the cells, and the second 2000g for 30 minutes removes most cell debris. The supernatant after these two centrifugation steps is then loaded for filtration, with a small sample removed to allow for comparison to initial levels. Filtration is performed using a polysulfone hollow fiber filter with varying pore sizes, 10 kD, 50 kD, and 500 kD (Spectrum Labs). The supernatant is pumped through the filter, at 24 mL/min using a peristaltic pump. After passing through the filter, the retentate is returned to the initial sample container for recirculation through the system. The filtration continues until the retentate volume reaches around 3 to 5 mL remaining. The retentate of filter contains the concentrated exosomes, and the permeate media and other small components and is discarded. When collecting the retentate, the filter is flushed with 1 mL of PBS to remove exosomes that might still be on the filter. After the filtration, the initial sample, the retentate, and samples from the permeate are then tested using Nanoparticle Tracking Analysis (Nanosight) to determine the concentration and size distribution of the exosomes. We then calculate the total number of particles in both the initial and final samples to see the recovery.
Results and Discussion: Our initial trials were all run using the 10 kD filter, further experimentation is needed on the 50 and 500 kD filters. Additionally, HEK293 cells were not used in these initial runs. For the MDA 231 cells, we started with 33 mL of supernatant, which is measured to have a total particle count on the order of 1011 particles, with some variability between individual harvests. For most runs, the final pellet is measured to be around 6 mL, and has a similar total particle count, on the order of 1011 total particles. In these initial runs we are seeing approximately a 5 fold concentration of the exosomes in the sample, while reducing the volume by approximately an order of 5. Additionally, the particle size distribution is far more similar to the initial distribution than is seen in ultracentrifugation. When running the permeate samples, the concentration of particles is far lower than in the retentate, indicating very few exosomes are going through the filter. There is still a degree of variability seen in the filtration results, though far less severe than with ultracentrifugation. While most samples show a final particle count very close to the initial sample, an occasional trial shows a concentration significantly lower than the initial, or slightly higher than the initial.
The CHO samples exhibit far more variability than the MDA 231 samples. These samples started with more volume, around 45 mL, and were concentrated to around 5 mL. The final particle count is greater than the initial count consistently across the CHO samples, in many cases significantly higher. Due to the higher initial volume, we have been able to concentrate the CHO exosomes to a final concentration of around 1012 particles/mL. Size distribution is far more consistent to the initial sample compared to ultracentrifugation.
Conclusion: Ultracentrifugation as a method of purifying exosomes has significant drawbacks, from the cost and length of operation, to low recovery and size inconsistencies. Tangential Flow filtration has the potential to be a better method for the concentration of exosomes from cell culture supernatant. Initial runs have shown that a far greater amount of the initial exosome concentration is recovered after TFF than after ultracentrifugation, and that the size distribution after TFF is far more indicative of the initial population than an ultracentrifuged sample. There are still concerns that need to be addressed with future work. Ultracentrifugation contains an additional step that has not yet been accounted for, a 10,000g centrifugation step that is intended to remove apoptotic bodies. Further runs will be performed using larger filters, where the exosomes should pass through as the permeate, and apoptotic bodies should remain in the retentate. This will likely impact the recovery of exosomes, as some might remain in the retentate. Other concerns include shear stress on the exosomes causing damage or deforming them enough to pass through the small membrane, and the length of time required to reach the desired volume. Back pressure applied plays a role in these concerns, a higher back pressure allowed us to reach the concentrated volume faster, but it could cause more exosomes to be squeezed through the pores of the membrane leading to lower recovery. Overall, initial results indicate that there is a potential of the TFF system being a promising way to concentrate exosomes, alleviating some of the issues associated with ultracentrifugation.
1 Guo, W., Gau, Y., et al. 2017, âExosomes: New players in cancerâ, Oncol Rep. Vol 38, pp. 665-675.
2Lane, R., Korbie, D, et. Al. 2015, âAnalysis of exosome purification methods using a model liposome system and tunable-resistive pulse sensingâ, Scientific Reports, Vol 5, article number 7639. doi:10.1038/srep07639
3 Li, P. Kaslan, M. et al. 2017 âProgress in Exosome Isolation Techniquesâ, Theranostics. Vol 7, pp. 789-804.
4 Helwa, I. Cai, J. et al. 2017 âA Comparative Study of Serum Exosome Isolation Using Differential Ultracentrifugation and Three Commercial Reagentsâ. PLOS ONE doi:10.1371/journal.pone.0170628
5 Livshits, M. Khomyaova, E. et al. 2015. âIsolation of exosomes by differential centrifugation: Theoretical analysis of a commonly used protocolâ. Scientific Reports, Volume 5, article number 17319 doi:10.1038/srep17319
6 van Reis, R., Leonard, L. et al. 1991. âIndustrial scale harvest of proteins from mammalian cell culture by tangential flow filtrationâ. Biotechnology and Bioengineering, Vol 38, pp 413-422.