Poster M18 - Modification of EV Production Workflow: Impacts of Growth Media and Filtration Methods on EV Metrics

Introduction: Extracellular vesicles (EV) have been shown to play a vital role in disease progression and treatment response, as they are released by nearly all types of cells. To establish a standardized and efficient workflow to generate EVs for translational studies, we evaluated methods to reduce confounding effects of co-isolates at cell culture and EV concentration steps. First, we compared the performance of serum-free media to EV-depleted media in our EV-production and analysis pipeline. EV Depleted Media (EVDM) uses ultracentrifugation to remove bovine serum-derived EVs from the media, but even after extended ultracentrifugation carryover EVs remain. So, we hypothesized that use of the serum free media XVIVO 15 might reduce the detection of non-specific EV particles or proteins. Second, we evaluated whether the process of concentrating liter-scale volumes of conditioned cell culture media could be streamlined by use of tangential flow filtration instead of ultrafiltration. We hypothesized that TFF should outperform UF by reducing processing time and increasing EV recovery.

Materials and

Methods: Complete RPMI1640 was EV-depleted by ultracentrifugation at 100,000xg for 18 hours at 4 degrees C. Mouse dendritic cells (DC2.4) were grown in 10 T225 flasks containing 50 mL EVDM or XVIVO 15 each, with the CCM volume being 500 mL for each. Cell-free CCM was recovered after reaching 70-80% confluency. Cells and debris were removed by centrifugation at 500 and 2500xg for 10 min. Ultrafiltration (Pall Jumbosep 100K) was used to concentrate from 500mL to 5-10mL prior to size exclusion chromatography (Izon qEV10 70nm) to remove free-protein and other co-isolates. Particle and protein concentrations in EV fractions were counted using Nanoparticle Tracking Analysis and NanoDrop 8000. The peak fractions were also analyzed using flow cytometry, microBCA, Western Blot, TEM, and protein mass spectrometry.

To compare UF and TFF, DC2.4 cells were grown in XVIVO 15 and the EVs were generated using the same parameters as before. After the cell and debris spins, the CCM was filtered by either Jumbosep or by using the Minimate TFF System equipped with a 100K Omega capsule. The TFF system was also used to filter 1 L and 4 L of CCM to test the scalability of the system. The filtered CCM was then put through SEC and the EV fractions were analyzed as above.

Results, Conclusions, and Discussions: We generated production-scale preparations of DC2.4 EVs using different growth media and concentration methods to evaluate the impact of media type and concentration method on EV particle concentration, EV protein profiles, and EV size distributions. EVs produced in EVDM had peak particle and protein concentration of 4.99E11 p/mL and 73.06 ug/mL while those produced in XVIVO 15 had concentrations of 1.43E11 p/mL and 582.23 ug/mL. FCM analyses of these preparations demonstrated a distinctive 130-140nm population in addition to a broad distribution of ~60-150nm EVs, and this peak was accentuated in XVIVO media, compared to EVDM. Proteomics analyses of EVs produced in these preparations demonstrated that more protein peptides were identified from EVs produced in EVDM compared to XVIVO.

Comparing the UF vs TFF preps, we observed similar particle (3.3E10 vs 2.4E10 p/mL) and protein (23.0 vs 47.7 ug/mL) concentrations. When the CCM volume was scaled up to 1 L and 4 L for TFF, we saw higher total particle (1.30E11 vs 1.90E11 p/mL) and protein concentration (701.7 vs 2032.7 ug/mL) recovery levels. When factoring the total volume harvested, the 1 L prep had higher particle (6.50E8 p/mL and 3.51 ug/mL) and protein concentrations (2.38E8 p/mL and 2.54 ug/mL) than the 4 L prep. This could be an indicator of an input saturation limit of the TFF system. FCM analyses showed similar peaks as before except for the 4 L prep.

Ultimately, choice of media and CCM concentration prior to further processing and analyses need to be informed by considerations in the downstream assays. In our experiments presented here, we find that cell culture in XVIVO media, CCM concentration by ultrafiltration, and subsequent SEC is a robust approach for producing highly reproducible EV preparations. However, EVDM and / or TFF are acceptable alternatives in this EV production and isolation workflow.

Acknowledgements (Optional): This work was supported by NIH NCI CCR Laboratory of Pathology, the NIH NCI CCR Radiation Oncology Branch, and NIH 1ZIABC011502.