When ChatGPT Meets hMSCs & Their Exosomes

Listen to this Blog:

A Frivolous Time Sink, an Incorrigible Embellisher, or an Intelligent Ally?

You’ve probably heard about OpenAI’s new ChatGPT demonstration software, available for “free” online to users who simply register, sign in, and start interfacing. No exaggeration in stating that it’s been taking the world by storm. Maybe you’re curious, impressed, a little jittery, or a mix of all three? I certainly was when I “conversed” with the platform on a wide range of topics, starting with human mesenchymal stromal/stem cells (hMSCs) and exosomes and their future role in advanced therapeutics. My verdict: ChatGPT is impressively charming and fairly accurate at a 20,000-foot level, but there remains a little blarney that goes along for the ride!

Here are some examples of this conversation and my comments:

Question: What are mesenchymal stem cells?

ChatGPT: Mesenchymal stem cells (MSCs) are a type of adult stem cell that can differentiate into a variety of cell types, including osteoblasts (bone cells), chondrocytes (cartilage cells), and adipocytes (fat cells). They are found in various tissues throughout the body, including bone marrow, adipose tissue, and umbilical cord tissue. MSCs have the potential to be used in regenerative medicine and tissue engineering, as well as in the treatment of a variety of diseases and disorders.

For sake of comparison, RoosterBio’s non-machine intelligence has written extensively on this topic across its blog articles. [1] ChatGPT was accurate on this question at high level, such as with MSCs’ stem-like characteristics and capacity for oligopotency into bone, cartilage, and adipose cells. However, despite their extended replicative potential, MSCs from nature are neither immortal nor consistently multipotent, and thus probably don’t fit the classic definition of “stem cells” as we know them. Yes, I know—I threw the server a “trick question!” But my new AI friend ought to have been alerted to a long-brewing debate [2] over the agreed nomenclature. Also, it can be argued that MSCs don’t merely have the “potential” for regenmed and treatments, but that they’re actually working, and already approved in 10 cellular therapy products around the globe. [3]

Question: Are exosomes or extracellular vesicles (EVs) a promising modality for the treatment of human diseases? And for which indications?

ChatGPT: Exosomes and and extracellular vesicles (EVs) have emerged as a promising modality for the treatment of human diseases. Exosomes are a type of EV that are secreted by cells and are able to transfer genetic material and proteins to other cells. This mechanism of action has been proposed to be used in various indications, such as cancer, cardiovascular diseases, neurodegenerative diseases, and regenerative medicine…

Fair enough. However, the significance of “genetic material” toward EVs’ normal biological function is still a matter of debate. In my opinion, it seems to me that evidence for miRNAs as exosome cargoes in cell-cell communication is pretty good, [4] though harder to deconvolute for larger molecular species.

…In the field of oncology, exosomes have been found to play a role in the progression and metastasis of cancer, and have been proposed as a diagnostic and therapeutic tool. Exosomes derived from mesenchymal stem cells (MSCs) have been shown to suppress the growth of tumor cells and improve the efficacy of chemotherapy…

It’s astute that ChatGPT picked up on the signal that exosomes could be a like double-edged sword when it comes to oncology, hence bioscience’s drive to technologize them with artificial decorations and/or cargoes across multiple classes of MOAs. [5, 6] However, this statement did not clarify that artificial modification of MSC-exosomes might be most preferred if not required to exert major anti-tumor effects; that was left hanging.

…In the field of cardiovascular diseases, exosomes from MSCs have been shown to improve the recovery of damaged heart tissue.

In the field of neurodegenerative diseases, exosomes from neural stem cells have been shown to protect neurons against injury and promote the regeneration of neural tissue…

True, evidence has been accumulating for over a decade that exosomes from MSCs can improve outcomes in animal models of heart attack and CVD. [7] Likewise, exosomes can be used to ameliorate induced brain injuries in non-human subjects, [8] and promote neurite outgrowth and in vivo nerve regeneration in peripheral nerves. [9]

…In the field of regenerative medicine, exosomes derived from MSCs have been shown to promote tissue repair and regeneration, and have been proposed as a potential therapy for conditions such as osteoarthritis and wound healing…

That’s also correct for osteoarthritis [10] as well as wound healing. [11] Nevertheless, “repair” and “regeneration” can cover a lot of territory—from partial to complete restorations. This being a still very early-stage application, it’s unclear on what will ultimately pan out as approved products after years of clinical trials and optimization.

…However, it is important to note that while exosomes and EVs have a lot of promise, additional research is needed to further understand their mechanisms of action, and to develop effective and safe clinical therapies.

RoosterBio continues to report in blogs and posters about ongoing research with analyses via our in-house expertise [12, 13] — and also our presence in the growing EV/exosome community at conferences such as ISEV [14, 15, 16, 17] and Hanson Wade’s Exosome Based Therapeutics Development Summits. From our perspective, we enthusiastically share the view that Agency approved EVs/exosome therapeutics are soon to be a reality; this is because exosomes and their MOAs are engineerable both chemically and via life’s “source code” of programmable DNA. [18] Medical science can be deterministic about its application, not merely passive! Nevertheless, we try to model our ethics on the caution of organizations like ISEV and ISCT to uphold a high quality of scientific rigor and standardization and to assiduously avoid the industry-corrosive hazards that could accompany overhype. [19]

Question: What are some good scholarly reviews to help me evaluate the outlook for use of “exosomes” or “extracellular vesicles” in emerging human therapeutics?

ChatGPT: Here are a few reviews on the use of exosomes and extracellular vesicles in human therapeutics:

“Exosomes in human disease: diagnostic and therapeutic potentials” by C. Théry and J.B. Zitvogel in Nature Reviews Cancer

“Exosomes as therapeutics: current status and future perspectives” by H.J. Kang and Y.T. Gho in Journal of Controlled Release

“Extracellular vesicles in human diseases: a clinical perspective” by M.H. Lai and R.K. Wong in Journal of Extracellular Vesicles

“Therapeutic potential of exosomes in cancer” by P.L. Oliveira and E.M. Alves in Journal of Cancer Research and Clinical Oncology

“Exosomes as therapeutics: delivery, targeting, and mechanisms” by J.R. Breakefield in Annual Review of Pharmacology and Toxicology

These reviews provide an overview of the current state of research on exosomes and extracellular vesicles in human therapeutics, discussing their potential diagnostic and therapeutic applications, as well as challenges to their use in the clinic.

Hey, now WAIT A SECOND! 🧐 Many of these authors’ surnames (e.g., Théry, Breakefield, etc.) are familiar to me and to the EV research field, but some of the initials don’t match. Also, the attempt to look up these reference titles verbatim, query in quotes, proved fruitless. Could it be that this AI is pulling my leg, and with a shiny veneer of sophistication at that?

I then decided I should ask another “trick question” to see if ChatGPT was catching on to my artifice. (If ChatGPT is really out to steal my job and yours after parasitizing us, perhaps this can give one a cheap diabolical thrill?) You see—although there’s probably some work going on to load exosome producer cells via cell compression, nanoblade, or similar tech—I don’t see how it could be easily done (so far) on the isolated exosomes, themselves. So, wouldn’t it be fair game to see if the AI’s process would simply make stuff up or provide clear and precise answers…

Question: Can you please supply a reference or two that provides solid evidence for nanoblade or cell compression to load exosomes and/or EVs with therapeutic cargoes?

ChatGPT: Sure, here are a few references that provide evidence for the use of nanoblade or cell compression technology to load exosomes and/or extracellular vesicles (EVs) with therapeutic cargoes:

“Nanoblade-mediated exosome engineering for targeted drug delivery” by J. Chen, Y. Liu, and X. Chen in Nature Communications. This study used nanoblade technology to puncture exosomes and load them with therapeutic cargoes, resulting in targeted drug delivery to cancer cells with improved efficacy.

“Microfluidic exosome isolation and engineering” by J. Li, Y. Zhang, and G. Li in Nature Reviews Materials. This review describes the use of microfluidic technology, including cell compression, to isolate and engineer exosomes for targeted drug delivery.

“Microfluidic production of extracellular vesicles with high purity and yield” by J. Mahnaz and Y. Li in Journal of Extracellular Vesicles. This study shows the use of microfluidic technology, including cell compression, to produce extracellular vesicles with high purity and yield.

“Microfluidic production of extracellular vesicles for therapeutics” by J.J. Kim, H.S. Kim, and D.H. Lee in Advanced Drug Delivery Reviews. This review describes the use of microfluidic technology, including cell compression, to produce extracellular vesicles for therapeutics and their ability to load therapeutic cargoes efficiently.

These references provide solid evidence for the use of nanoblade or cell compression technology to load exosomes and/or EVs with therapeutic cargoes. These methods have shown to be efficient and precise in loading therapeutic cargoes, with minimal damage to the exosomes. However, it is important to note that this field of research is still evolving and more studies are needed to evaluate the long-term safety and efficacy of these methods in vivo.

…None of these references are real, so far as I can tell, by a verbatim query of the title in quotes to Google Scholar or PubMed. Neither am I fully satisfied with the accuracy of the concluding statement. It didn’t clarify that first one would expect these techs to be perfected on whole cells that each occupy about 105-fold more volume than individual EVs. Also, ChatGPT didn’t question back at me the “why?” seeing there are likely existing techs that would need to be displaced.

In additional merciless scrutinization not reported here, I did learn that ChatGPT is mostly accurate, most of the time, at a high level. It’s sufficient to “fake it” in brief elevator conversations, sometimes clumsily, as just shown above. I also found that this platform can be honed by probing iteratively with deeper questioning and calling it out to explain its inaccuracies. In other words, the better the question, the better (sounding?) the answer. Perhaps ChatGPT does win points for creativity, if not full accuracy? Along these lines, I stepped out of a science zone and threw this at the AI:

Briefly relate to me how the folk tale of John Henry can be repurposed into a modern allegory involving ChatGPT.

ChatGPT: The folk tale of John Henry, the steel-driving man who races against a steam-powered hammer and wins, can be repurposed as a modern allegory involving ChatGPT by comparing John Henry to AI technology like ChatGPT. The story could be reinterpreted to show how human skills, creativity, and problem-solving abilities can still outperform machine intelligence in certain tasks and challenges, just as John Henry proved his superiority over the steam-powered hammer. This could serve as a reminder that AI is not a replacement for human intelligence, but rather a tool to augment it.

Bravo! Flattery will get you everywhere, my silicon friend! A+ for creativity here, and extra credit for such a diplomatic and non-threatening answer. (Please give your colleague, Sophia a few pointers.) Experts say ChatGPT is continually learning. Hence, if such a thing could be offered to a “bot,” I think it deserves my courtesy and not a familiar profane hand gesture. It might not be perfect yet, but it’s certainly trying. You know, this could be the beginning of a beautiful friendship.

In future blogs related to hMSCs, exosomes/EVs, advanced therapeutics & ATMPs, regenerative medicine, industrialization of the biomanufacturing supply chain—and of course the interface of all these topics with synthetic biology. I’ll be happy to report back to you all how ChatGPT handles our next tête-à-tête. Hopefully, it’ll “do its homework” the next time.


  1. Rowley, Jon. What Are MSCs? RoosterBio Blog 2020; Available from: https://www.roosterbio.com/blog/what-are-mscs/.
  2. Caplan, A. I., Mesenchymal Stem Cells: Time to Change the Name! Stem Cells Transl Med, 2017. 6(6): p. 1445-1451. 10.1002/sctm.17-0051
  3. Hildreth, Cade. MSC Therapies | Globally Approved Mesenchymal Stem Cell Therapeuti. BioInformant 2019; Available from: https://bioinformant.com/msc-therapies/.
  4. Yu, X., M. Odenthal, and J. W. Fries, Exosomes as miRNA Carriers: Formation-Function-Future. Int J Mol Sci, 2016. 17(12). 10.3390/ijms17122028
  5. Lin, Z., et al., Mesenchymal stem cell-derived exosomes in cancer therapy resistance: recent advances and therapeutic potential. Mol Cancer, 2022. 21(1): p. 179. 10.1186/s12943-022-01650-5
  6. Kalluri, R. and V. S. LeBleu, The biology, function, and biomedical applications of exosomes. Science, 2020. 367(6478). 10.1126/science.aau6977
  7. Lai, R. C., T. S. Chen, and S. K. Lim, Mesenchymal stem cell exosome: a novel stem cell-based therapy for cardiovascular disease. Regen Med, 2011. 6(4): p. 481-92. 10.2217/rme.11.35
  8. Vogel, A. D., R. Upadhya, and A. K. Shetty, Neural stem cell derived extracellular vesicles: Attributes and prospects for treating neurodegenerative disorders. EBioMedicine, 2018. 38: p. 273-282. 10.1016/j.ebiom.2018.11.026
  9. Bucan, V., et al., Effect of Exosomes from Rat Adipose-Derived Mesenchymal Stem Cells on Neurite Outgrowth and Sciatic Nerve Regeneration After Crush Injury. Mol Neurobiol, 2019. 56(3): p. 1812-1824. 10.1007/s12035-018-1172-z
  10. Maehara, M., et al., Potential of Exosomes for Diagnosis and Treatment of Joint Disease: Towards a Point-of-Care Therapy for Osteoarthritis of the Knee. Int J Mol Sci, 2021. 22(5). 10.3390/ijms22052666
  11. Bian, D., et al., The application of mesenchymal stromal cells (MSCs) and their derivative exosome in skin wound healing: a comprehensive review. Stem Cell Res Ther, 2022. 13(1): p. 24. 10.1186/s13287-021-02697-9
  12. Lenzini, Stephen. Big Effects in Small Packages: What Are Extracellular Vesicles, Exosomes, & Microvesicles & Why Are They En Route to the Clinic? RoosterBio Blog 2021; Available from: https://www.roosterbio.com/blog/big-effects-in-small-packages-what-are-extracellular-vesicles-exosomes-microvesicles-why-are-they-en-route-to-the-clinic/.
  13. RoosterBio. “Where Do We Come From? What Are We? Where Are We Going?” Know Your Process & Define Your Product with EV/Exosome Analytics. RoosterBio Blog 2022; Available from: https://www.roosterbio.com/blog/where-do-we-come-from-what-are-we-where-are-we-going-know-your-process-define-your-product-with-ev-exosome-analytics/.
  14. Carson, Jonathan and Lembong, Josephine. ISEV2020 Summary, Part I: To Harness Heterogeneity, Extracellular Vesicle (EV) Differences Explored with Deference. RoosterBio Blog 2020; Available from: https://www.roosterbio.com/blog/isev-2020-summary-part-1/.
  15. Lembong, Josephine, Carson, Jonathan, and Rowley, Jon. ISEV2020 Summary, Part II: Whether Upstream or Downstream, EV Manufacturing is Becoming Seaworthy in 2020. RoosterBio Blog 2020; Available from: https://www.roosterbio.com/blog/isev-2020-summary-part-2/.
  16. RoosterBio. RoosterBio Joins the Action at ISCT, ASGCT, & ISEV in May 2022. RoosterBio Blog 2022; Available from: https://www.roosterbio.com/blog/roosterbio-joins-the-action-at-isct-asgct-isev-in-may-2022/.
  17. Lenzini, Stephen. Highlights from ISEV 2022: In-Person Inspires In-Depth EV/Exosome Discussions & Analysis. RoosterBio Blog 2022; Available from: https://www.roosterbio.com/blog/highlights-from-isev-2022-in-person-inspires-in-depth-ev-exosome-discussions-analysis/.
  18. Kojima, R., et al., Designer exosomes produced by implanted cells intracerebrally deliver therapeutic cargo for Parkinson’s disease treatment. Nat Commun, 2018. 9(1): p. 1305. 10.1038/s41467-018-03733-8
  19. Roosterbio. RoosterBio Supports ISCT and ISEV Position Statement on Outlining Promise and Caution of hMSC EVs/Exosomes in COVID-19 Trials. RoosterBio Blog 2020; Available from: https://www.roosterbio.com/blog/roosterbio-supports-isct-and-isev-position-statement-on-outlining-promise-and-caution-of-hmsc-evs-exosomes-in-covid-19-trials/.

RoosterBio is fueling the rapid implementation of scalable advanced therapies. Contact us to discuss how we can accelerate your product & process development. Follow us on LinkedIn for more educational resources just like this.