Maximising Extracellular Vesicle Output: A Small Bioreactor Goes a Long Way

Cell culture is traditionally conducted in flasks, with cells grown in a ‘2D’ manner on a flat plastic surface. Unfortunately, this is a limited and poor approach to extracellular vesicle (EV) production, with an unwieldy number of flasks needed to isolate sufficient EVs. It is not uncommon to see the incubator of an EV researcher packed with T175 flasks, growing up cells for an EV harvest. Whilst this EV farming from 2D culture is inconvenient for EV research projects, it is prohibitively incompatible with larger scale operations such as the production of EVs for therapeutics or cosmeceuticals. Not only do these cultures take up large amounts of space, but they consume large quantities of media, demand frequent care and have low reproducibility. Which raises the question: could there be a better way?

For some time, bioreactors have been in commercial use across a range of applications; for example, for antibody production. More recently, bioreactors have been used to successfully produce large quantities of EVs.¹ However, little is known about the stability of these bioreactors and many are prohibitively large for research laboratories where space is often at a minimum. To address this knowledge gap, Hisey et al. (2022)² investigated the yield and stability of a 1L flask-type bioreactor. In the bioreactor, cells were grown on a woven 3D surface within a 15 mL cell compartment, which was separated from a larger media compartment by a 10 kDa membrane. This separation between compartments was implemented so that any EVs produced in the smaller cell compartment would remain separate from the larger media compartment, keeping them concentrated in a much smaller volume. These bioreactors are a highly effective use of space – several can fit within standard laboratory incubators – but just how productive are they?

‍

Study methodology for culturing and isolating EVs  

Five different breast cancer cell lines were seeded into the cell chamber of their own bioreactor before being gradually acclimated to an optimised, serum-free media (Figure 1). The 10 kDa membrane allowed for fetal bovine serum to be maintained in the media chamber without risking bovine EVs contaminating those produced by the cells in the cell chamber. The cell chamber was filled with 15 mL of media, whilst the media chamber contained 500 mL media.

EV-containing media was then collected according to the schedule in Figure 1, alongside measurement of media compartment volume (to check for membrane damage) and quantification of shed cells – and their viability – for the assessment of bioreactor viability.

‍

The EV-containing media (15 mL) was removed and centrifuged at 2,000 x g for 10 minutes to remove cells and large debris. The supernatant was then centrifuged at 10,000 x g for 30 minutes to pellet large EVs. For the concentration of small EVs, a 100,000 x g centrifugation for 70 minutes was used. The pellet was then resuspended in 700 µL PBS prior to qEV isolation using a qEVoriginal 35 nm column with the aid of the Automatic Fraction Collector (AFC). EV-rich fractions were pooled.

‍

Key findings and open questions from Hisey et al. (2022)

1. Bioreactor health can be monitored by counting shed cells. Inaccessibility makes the cells in bioreactors difficult to assess for viability and growth, making bioreactor stability difficult to assess. Hisey et al. counted shed cells in the media removed for EV isolation at twice weekly intervals. They found the number of shed cells gradually decreased over time as bioreactors presumably stabilised (Figure 2a) – with one exception. The MCF7 bioreactor had a sudden increase in shed cells, indicating bioreactor instability and leading the authors to terminate that particular bioreactor. Media volume was also monitored to check for membrane breakage.

‍

‍

2. Different cells provide different yields of EVs over the bioreactor lifespan. Interestingly, whilst most cells increased their EV yields over time, the yield of EVs from cells positive for HER2 decreased sharply after an initial few weeks of good yield. This suggests that some cells may not be suited to long-term bioreactor cultures for EV production. Investigating why this occurs may be key to optimising EV bioreactors.
3. The yield of EVs from these bioreactors is astronomically high. Saying that the yield is astronomically high may sound like an over-hyping of the results, but when the EV harvest from a 1L bioreactor is comparable to one hundred T175 flasks, we believe the expression is justified (Figure 2B). Given these harvests occur twice weekly, this is equivalent to over 200 of the large T175 flasks each week. As the flask-style bioreactor in question is roughly the size of two stacked T175 flasks, this represents a huge leap forward in EV production capacity.
4. EV protein content remains stable over bioreactor lifespan. For each of three cell lines, EV proteomics was conducted for 6 different EV collections. Each cell line clustered separately in a principal components analysis (Figure 3). The authors found the EV protein cargo to be relatively stable across the 6 sampling periods. This suggests that at least where protein is considered, these flask-style bioreactors can produce comparable EV harvests for months.

‍

How did using qEV columns contribute to the success of this study?

This study was conducted over a long period of time. For four months of that time, EVs were harvested twice weekly for a total of nearly 200 EV harvests across the bioreactors. It is not clear whether EVs were isolated from all harvests, but we do know that at least 65 EV isolations were conducted using qEV columns. Given that one of the most important questions asked in this study was regarding the stability of the EVs produced by the bioreactors over time, using a reproducible technique to isolate EVs was critical. SEC is widely recognised as a reproducible technique which isolates highly functional EVs in a gentle manner. And, alongside use of the Automatic Fraction Collector (which was also used in this study) user-introduced variation in sample preparation can be further reduced, thereby maximising the reproducibility of EV isolation.

‍

The future of flask-based bioreactors for EV production

This study has shown that flask-based bioreactors are stable for long periods and produce large quantities of EVs. Importantly, media volume and shed cells can be monitored to identify when bioreactors have become unstable and thus unsuitable for EV production. The yield of EVs was extremely high as compared to traditional 2D culture, despite the bioreactor taking up the space of roughly two T175 flasks.

‍

Final thoughts on further maximising yield

EV yield may have been even higher if tangential flow filtration had been used to concentrate media prior to qEV isolation instead of ultracentrifugation, as has been shown previously.³ The team at Izon Science offers expertise in tangential flow filtration, which can be used as a concentration step prior to EV isolation using qEV columns. Our extensive experience in EV isolation puts us in the perfect position to work with you to implement customised isolation solutions, which can include both methodological and physical integration into bioreactor workflows with a high degree of automation.