“There is nothing in the intellect that was not first in the sense”. — Aristotle

With a nod to Wise Old Aristotle, we scientists prefer to use sharpened senses and the best instruments to diligently learn what nature’s telling us, irrespective of our frothing imaginations. The questions that came in during our February webinar [1] on MSC-EV characterization were refreshingly in that same spirit. [2] Technically diverse, they still coalesced around a resonant theme. Researchers across pharma, biotech, and academia aren’t just asking how to measure EVs. They’re asking something deeper: what to measure that actually matters? [3] However, with time limited during the live Q&A, we wished to continue answering them in the form of a blog published here, below.

The field is moving from simply detecting extracellular vesicles to genuinely understanding them. [4] At the same time, our audience looks to integrate them into more scalable manufacturing processes. That palpable shift from descriptive characterization toward mechanism-anchored insight underlay many questions we received. We’ve organized those questions into five topics that reflect where the EV field stands today, and where it’s heading.

Theme 1: Does Processing Shape Potency — and How Do You Know What You’ve Lost?

These questions share a common thread: if you change your process, how do you know whether you’ve preserved what matters? Filtering, purifying, and selecting extracellular vesicle subpopulations all affect the final product. But the real concern isn’t the step itself, it’s whether the biology survives it. Researchers here are thinking about prospective potency markers such as CD73, not just purity, and that distinction is at the heart of where EV characterization is heading. [5]

Q. Does EV size affect CD73 content, and how do you account for that in your process?

Yes. Smaller EVs generally carry less CD73 per particle, which is exactly why measuring activity directly is more informative than simply counting particles. If your process shifts the size distribution, CD73 activity data can tell you whether that shift has affected functional content. This kind of metric is what may allow translational programs to define dose based on biology rather than bulk quantity.

Q. What happens to the CD73-positive particles removed at the 0.2 µm filtration step, and can you choose a different cutoff?

We have characterized both pre- and post-filtration fractions. While some larger, CD73-positive particles are removed at this step, we’ve confirmed that the sub-200 nm population retains strong CD73 enzymatic activity. In bioprocessing, filtration is usually necessary to remove larger (micron-sized) particulates from the sample. If your product goals favor retaining a broader EV population, larger pore size filters are available. So, filtration cutoffs can be evaluated during process development.

Theme 2: Designing a Downstream Process That Will Scale

The questions here were beyond basic research; they were CMC-minded. Researchers asking about TFF, SEC, affinity chromatography, and DNA removal are thinking ahead to clinical manufacturing. [6] The underlying concern is one that the whole field shares: the decisions you make early in process development can ramify into obstacles later. Building with scalability and regulatory alignment in mind from the start is what separates programs that translate from those that stall.

Q. Why is TFF preferred over ultracentrifugation, and how does the full downstream processing workflow fit together?

Tangential flow filtration (TFF) is our preferred approach for scalable manufacturing. While ultracentrifugation can be useful in early research settings, TFF supports larger processing volumes, better process consistency, and cleaner alignment with clinical manufacturing workflows. [7] The full DSP workflow builds on this foundation. First, clarification filtration removes large debris and therefore reduces pressure on the TFF system, improving overall robustness and recovery. Additives such as AgentV^TM-DSP can further minimize membrane fouling and particle loss during concentration. [8] Integrating these scalable unit operations early in development simplifies the path to clinical translation — and the decisions you make at this stage are much easier to carry forward than to undo later.

Q. Is additional purification (e.g. SEC, ion exchange, or affinity chromatography) necessary after TFF?

Adding purification steps beyond TFF is dependent on the application and the desired purity of the final product. For some applications, TFF alone delivers a strong balance of yield, purity, and functional activity. [9] Ion-exchange and mixed-mode chromatography are additional, often-preferrable options depending on your product requirements. SEC could be added as a polishing step when higher purity is required. Affinity-based approaches are promising and an active area of field-wide interest, but broader adoption depends on the availability of scalable resins suited for EV applications. At early development stages, the added cost and complexity may not be justified, though it’s worth revisiting as your program matures. As mentioned before, any changes to the process will have an impact on the quality attributes of the product. Establishing the critical quality attributes will help define the final product and determine the optimal processing. Our recommendation is to always perform quality check at every step of the process and with every change to it.

Q. How do you manage contaminants like free DNA in the downstream workflow, and what role does clarification filtration play?

The downstream processing workflow is designed to systematically remove non-EV components from conditioned media, including free DNA, residual proteins, and other contaminants. Nuclease treatment can be incorporated at an optimized point in the process to address DNA specifically. The clarification step(s) also play an important role here. Once cellular debris and large aggregates are removed by filtration, there is reduced variability in the material to enter downstream steps for a more consistent, better-defined preparation.

Theme 3: From Correlation to Confidence — How Robust Is the CD73 Framework?

Beneath the technical specifics of detection limits and data scatter, these questions reflect a deeper need: researchers want to know whether CD73 content is a meaningful measurement. [5],[10] Is it a window into therapeutic activity, or just a convenient marker? One major challenge of any emerging critical quality attribute is to build enough evidence that the field can rely on. The questions here signal that the audience is ready to move in that direction, but they wisely want the data to lead the way.

Q. You mentioned a lower limit of detection of 7.5 molecules of CD73 per particle — how can you then quantify 5 molecules per particle after DSP?

As outlined in the methods overview part of the webinar, each particle shows either a detectable number of molecules or an undetectable number of molecules. When calculating the average number of molecules per particle, we chose to account for those particles with an undetectable number of molecules by setting their value to zero. Therefore, the average number of molecules per particle is unconstrained by detection limit and can exist within the range 0-7.5.

Q. The correlations for CD73 activity seem driven by the highest data point. Would more intermediate data points strengthen the correlation?

This is a fair and important observation. We acknowledge that, as with any emerging critical quality attribute (CQA), expanding the dataset across more production lots will strengthen the correlation. That work is actively ongoing. Furthermore, we have gathered additional data to strengthen the correlation since generation of the original data presented in the webinar.

Q. Have you evaluated correlations between CD73 and other functional assays?

Yes. Correlating CD73 activity with additional functional assays, such as macrophage-based immunosuppressive potency assay is an active area of development. CD73 enzymatic activity is directly tied to the adenosine-mediated immunomodulatory mechanism of action of MSC-EVs, which makes it one of the most promising functional CQAs for this modality. Expanding the correlation dataset across orthogonal functional methods remains a key focus.

Q. Is CD73 expressed on neuronal or hiPSC-derived EVs?

Expression varies by cell source, but our analytical platform can detect and quantify CD73 across different extracellular vesicle sources. It’s not solely limited to MSC-derived EVs. Whether CD73 is a functionally meaningful marker on neuronal EVs is a separate biological question that would need to be explored in the context of your specific application.

Q. Have EV functional CQAs like CD73 activity been explored as biomarkers in Alzheimer’s disease or MCI?

This remains an open and interesting area of research. CD73 expression does vary by cell source, and the analytical platform can detect and quantify it across different EV types beyond MSC-derived EVs.

Theme 4: Getting Upstream Right with Production, Collection, and the Foundation of a Therapeutic EV

Questions about collection media, harvest timing, storage buffers, and GMP-grade drug loading all point to a shared concern. How is the quality of your extracellular vesicle product before downstream processing begins? Upstream decisions such as cell source, collection conditions, timing, formulation set the ceiling for everything that follows. Researchers here are building manufacturing processes from the ground up and want to know what the best foundation looks like. [6, 11]

Q. Can you tell us more about the RoosterBio EV collection product and what it does?

RoosterBio’s EV production platform includes a chemically defined, low particulate EV collection medium, [12] specifically formulated for collecting EVs, exosomes, and other secretome materials in both 2D and 3D bioreactors. It is designed to work seamlessly with our upstream exosome manufacturing products and integrates directly into our DSP workflows.

Q. Have you evaluated different collection timelines? Is there an optimal harvest duration?

Yes. We’ve evaluated differing collection timing in our standard workflows. Best-practice guidance can be found in our available process recommendations [13] or tailored to your specific cell type and bioreactor format. [14] Timing matters both for yield and potentially for the functional CQAs of the EVs collected.

Q. Can EVs be used directly from collection media, or is downstream processing required for clinical use?

Some level of processing is recommended for most applications. Raw conditioned media contains residual proteins, DNA, and media components that add complexity to the end product and its reliable characterization. RoosterCollect^TM-EV is not a media meant to be an excipient, and EVs must ultimately be formulated in a physiologically compatible buffer suitable for in vivo application. The specific formulation strategy is determined by safety and stability considerations and guidance.

Q. What are the best resuspension buffers for EVs — water or PBS?

In our experience, buffered solutions such as PBS for extracellular vesicle resuspension did not have negative impacts on EV quality under our tested conditions. In general, unbuffered water is not recommended for biological contents as pH fluctuations can affect their stability.

Q. Have you done any work on EV storage stability, including different storage buffers?

We have conducted preliminary evaluations of extracellular vesicle stability across different formulation buffers and storage temperatures. This is an area of ongoing study, and we look forward to sharing more comprehensive data as it becomes available.

Q. Do you optimize EV loading under GMP conditions?

Drug-loaded and/or targeted extracellular vesicles and GMP platform readiness are areas we are actively expanding. We welcome partnership discussions for groups working on engineered or loaded EV therapeutics and are building toward an integrated, end-to-end solution from cells to characterized EV product. Please contact us to learn about current options and custom development opportunities.

Theme 5: Evaluating a Platform Partner — How Flexible and Accessible Are Your Services?

Several questions probed whether assays work on non-MSC EV types, whether proprietary markers can be incorporated, and whether external samples are accepted. These weren’t purely academic questions. They were technical due diligence to evaluate whether RoosterBio could serve as a flexible partner that can grow with their program, not just serve as a vendor to offer a rigid menu of tests. Researchers at this stage are thinking about integration and long-term fit, not just a box-checking service transaction.

Q. What are the CD63, CD9, and CD81 percentage cutoffs per ISEV guidelines?

We are not aware of any requirement for a percentage cut off set by ISEV. The percentage is sample dependent.

Q. Are these assays for internal use only, or can external samples be submitted?

Yes. We routinely work with external collaborators and can perform characterization services on submitted EV samples. Whether you’re at the discovery stage characterizing your first harvests or preparing analytical comparability data for regulatory filings, we can support you. Reach out to our team at services@roosterbio.com to discuss your project and receive a custom quote.

Q. How much should I budget for characterizing 10 mL of conditioned medium?

We recommend contacting the RoosterBio team directly at services@roosterbio.com for a custom quote based on your sample and assay needs.

Q. Do your services cover EVs only, or can you also work with conditioned media?

Both. Depending on your needs, we can work with already-purified extracellular vesicle preparations or process EVs directly from conditioned media. Sample format requirements vary by assay panel, and we’re happy to tailor a package to your situation.

Q. Can other flow cytometry platforms be used to develop and run quantitative assays for EV surface markers?

The analytical framework used for CD73 and tetraspanin quantification can, in principle, be adapted to other platforms and markers. If you have a proprietary marker or a specific protein of interest, we can collaborate to develop that approach for your application.

Q. How many EVs are needed for your characterization assay services?

Particle requirements depend on the specific assay and the complexity of your sample. As a general guideline, nanoparticle tracking analysis (NTA) typically requires approximately 1×10⁸ particles/mL. More complex assays may require greater inputs. Factors like your sample’s processing and matrix also matter, so we recommend discussing your material with us directly to define the right input levels for your project.

FINAL QUESTION:

Q. If you could “wave a magic wand” and solve one challenge in MSC-EV characterization, what would it be?

Great question. We have two answers that could ultimately affect how EV therapeutics perform in vivo.

(1) Solve detection limit issues. In nearly all assays we are contending with the signal measurement approaching the assay detection limit and/or conventionally acceptable signal-to-noise ratios. This makes it difficult to confidently assess EV contents, especially when we expect that some functionally relevant contents remain unknown.

(2) Standardize reporting across all labs to properly consider sample properties. Oftentimes, labs report “Our EVs show x…y…z… characteristics” without considering the fundamental sample properties. For example, is the sample purified? Is it conditioned medium? Is it processed through ultracentrifugation? We believe that a substantial amount of confusion in the literature stems from comparing results across samples with fundamentally different properties.

The EV field is asking more nuanced questions now. We’re here to help answer them.

For questions about RoosterBio’s products or services, contact us at services@roosterbio.com

 

References

1. RoosterBio. EV Content vs Function: Using Analytical Tools to Reveal the Connection. 2026; Available from: https://share.hsforms.com/1BcaHeOOARxmH1ApoPXddWA3564o.
2. RoosterBio. From Form to Function, What Does EV Content Really Tell Us? 2026; Available from: https://www.roosterbio.com/blog/from-form-to-function-what-does-ev-content-really-tell-us/.
3. RoosterBio. Analytics: A Referee for the Tug-of-War Between MSC-EVs’ Nature & Nurture. 2024; Available from: https://www.roosterbio.com/blog/analytics-a-referee-for-the-tug-of-war-between-msc-evs-nature-nurture/.
4. RoosterBio. When Is an Exosome Not an Exosome? 2025; Available from: https://www.roosterbio.com/blog/when-is-an-exosome-not-an-exosome/.
5. Cramer, Madeline. CD73: A Team Player Caught in the “AKT” of Wound Healing & Cell Survival via MSC Exosomes? 2024; Available from: https://www.roosterbio.com/blog/cd73-a-team-player-caught-in-the-akt-of-wound-healing-cell-survival-via-msc-exosomes/.
6. RoosterBio. Cytiva & RoosterBio’s Collaboration Yields Breakthroughs to Make Exosome Manufacture Easier from Benchtop to Bedside. 2024; Available from: https://www.roosterbio.com/blog/cytiva-roosterbios-collaboration-is-yielding-breakthroughs-to-make-exosome-manufacture-easier-from-benchtop-to-bedside/.
7. Jonathan Carson, Stephen Lenzini, Jae Jung. Countdown to Zero: Overcoming Downstream Processing’s Top Five Challenges for Viral Vectors & Exosomes. 2024; Available from: https://www.roosterbio.com/blog/countdown-to-zero-overcoming-downstream-processings-top-five-challenges-for-viral-vectors-exosomes/.
8. RoosterBio. To 10X & Beyond, AgentV™-DSP Arrives Ready to Rock. 2025; Available from: https://www.roosterbio.com/blog/to-10x-beyond-agentv-dsp-arrives-ready-to-rock/.
9. Elie Zakhem (RoosterBio), Jae Jung (RoosterBio), Cameron Garland (RoosterBio), Michael Boychyn (RoosterBio), Lauren Torres (Repligen), Mario Sinani (Repligen), Carl Breuning (Repligen), Jeremy Neidert (Repligen). Development of Manufacturing Therapeutic Platform for EVs Derived from MSC Using Tangential Flow Depth Filtration (TFDF®) & Tangential Flow Filtration (TFF). 2023; Available from: https://info.roosterbio.com/hubfs/Posters/Poster_2023_RoosterBio-Repligen_Development-of-Manufacturing-Therapeutic-Platform-for-EVs_v3.pdf.
10. Madeline Cramer, Elie Zakhem, Jon Rowley. Elevating MSC-EV Analysis: Development and Qualification of a CD73 Bioactivity. 2025; Available from: https://www.roosterbio.com/resource/elevating-msc-ev-analysis-development-qualification-of-a-cd73-bioactivity-assay/.
11. RoosterBio. When Data Blooms: Insights into MSC-EV Bioprocessing. 2025; Available from: https://www.roosterbio.com/blog/when-data-blooms-insights-into-msc-ev-bioprocessing/.
12. RoosterBio. RoosterCollect™-EV. 2026; Available from: https://www.roosterbio.com/products/roostercollect-ev-m2001/.
13. RoosterBio. Recommended EV Collection Protocol with RoosterCollect-EV 2025; Available from: https://www.roosterbio.com/wp-content/uploads/2019/12/M2001-RoosterCollect-Recommended-Protocol.pdf.
14. Stephen Lenzini, Jae Jung, Sanket Jadhav, Priyanka Gupta, Rukmini Ladi, Michael Boychyn, Jon Rowley, Elie Zakhem. Successful Development of a Scalable & Robust Process for MSC-EV Production. 2023; Available from: https://www.roosterbio.com/resource/isev-isct-2023-successful-development-of-a-scalable-robust-process-for-msc-ev-production/.